University of Southern California Dissertations and Theses

Essays in macroeconomics and macro-finance

(USC Thesis Other)

Essays in macroeconomics and macro-finance

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Essays in Macroeconomics and Macro-Finance
by
Dario Laudati
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(ECONOMICS)
May 2024
Copyright 2024 Dario Laudati

In memoriam of all my grandparents,
who left us during these years;
And to Ettore Majorana,
hoping that, by disappearing from this world,
he found himself in his Universe.
Was aus Liebe getan wird,
geschieht immer jenseits von Gut und Böse.
[ F. Nietzsche (1886), Beyond Good and Evil ]
ηθοζ ανθρωπωι δαιµων
[ Heraclitus ]
ii

Acknowledgments
My doctoral years have not been easy on a personal level. When beginning this section,
it was hard to resist the temptation of even mentioning any of the things happened over
this time span. Yet, just as I persisted and never gave up during the previous years, by the
same token I concluded that a new page can only be written by leaving the previous ones
behind. The only thing you can actually do in this life is to keep working on yourself, give
your best, and be patient. The odds sooner or later revert.
In the midst of this perfect storm, I am glad I have met great academic minds as the
ones who are part of my committee, the ones that I had the honor to co-author with, and all
the others I could exchange even just a few words with. I cannot start but by mentioning
my advisor, Vincenzo Quadrini: The kindest spirit and a great scholar. He was always
there, ready to discuss ideas, research progress, and more, whenever I demanded to. I am
sincere when I say that I could have not hoped for a better supervisor, and I am grateful
to have worked with him.
I certainly owe much to Caroline Betts and Romain Rancière for their support and
thoughtful comments on my research, and the effort to make USC Economics a better
place. In addition, they understood my problems one year ago and allowed me to stay
one extra year in order to conduct my job market and complete my dissertation in a more
meaningful way.
I am deeply thankful to Pablo Kurlat and Wenhao Li for their academic help, effort and
presence in their respective departments. I have always been impressed by their intelligence, and I hope that with time I will get closer to where they are.
I am also indebted to M. Hashem Pesaran for having spent time to “train me” both
from an empirical point of view, and as a researcher more in general. His rigor was beyond
mesmerizing since I first met him during his first Time Series class.
Last but not least, Irma Alfaro, Annie Le, Akiko Matsukiyo, and Young Miller should
iii

be thanked for their excellent administrative support in all phases of the program. Overall,
I am glad I landed at USC even though I hoped to bring home a better job market result
both for me and the department.
In the process, it was nice to share ideas and stories with a few new good friends such
as: Amanda Ang, Zhan Gao, Yi-Ju Hung, Thurston Nash, Brijesh Pinto, Fatou Thioune,
and Qitong Wang. I have also been blessed to have old friends like Chiara Chiurazzi and
Carlo Iengo, who patiently kept reaching out to me throughout these years even after
I disappeared for a long while. Antonio Di Matteo remains one of the most important
persons in my life since I met him on "his" track long time ago in a lazy summer afternoon.
That track has been unbelievably dismantled today thereby depriving many kids of that
positive environment; yet, our bond has remained strong over time, and way passed the
coach-athlete relationship.
My biggest thanks goes to my family for their support. The relationship with my father
and my sister has become stronger, more reliable, and more mature over time; but it is my
mother in particular that I need to thank for everything. She is the only one I could actually
fall upon in the worst moments, and she never stopped being there even when the sky was
falling down. God only knows what we had to go through as a family, but we came out of
it alive and well.
As I approach the end of my Ph.D., I am mostly left with a big willingness to unleash
everything that has been tapped over the past few years. Replonger en moi. May New York
City represent my actual landing in the New World to open a new, intense, and rewarding
chapter. From Caserta to Los Angeles that introvert kid on which no one may have bet did
end up walking many miles... many more are still waiting to be run.
Post Fata Resurgo. Semper.
iv

Table of Contents
Epigraph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Chapter 1: Identifying the Effects of Sanctions on the Iranian Economy using
Newspaper Coverage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Sanctions and the Iranian economy: an overview . . . . . . . . . . . . . . . . 8
1.3 Measures of sanctions intensity . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.4 Identification of sanctions effects: methodological issues . . . . . . . . . . . 19
1.4.1 Estimates of sanctions-induced output losses . . . . . . . . . . . . . . 25
1.5 Sanctions-augmented structural VAR model for Iran . . . . . . . . . . . . . . 27
1.5.1 Structural model estimation . . . . . . . . . . . . . . . . . . . . . . . . 30
1.5.2 Impulse response analysis . . . . . . . . . . . . . . . . . . . . . . . . . 36
1.5.3 Forecast error variance decompositions . . . . . . . . . . . . . . . . . 41
1.6 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Chapter 2: Inequality and the Rise of Finance . . . . . . . . . . . . . . . . . . . . . . 47
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.2 Stylized facts and background . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.3 A macro-finance model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.3.1 Sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.4 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
2.5 Quantitative results and policy experiments . . . . . . . . . . . . . . . . . . . 73
2.5.1 Baseline scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
2.5.2 Dividends taxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
2.5.3 Unconstrained public safe assets supply . . . . . . . . . . . . . . . . . 76
2.6 Empirical analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
2.6.1 The long-run co-variability of inequality and finance . . . . . . . . . . 78
2.6.2 Testing for the identifying mechanism across advanced economies . 81
v

2.7 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Chapter 3: The Political Economy of Banking and Shadow Banking Competition . 87
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.2 Stylized facts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
3.3 Empirical assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
3.4 Theoretical framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
3.4.1 Baseline set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
3.4.2 Extensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
3.5 Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Appendix A: Appendix to Chapter 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
A.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
A.2 Data appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
A.2.1 Sanctions intensity variable . . . . . . . . . . . . . . . . . . . . . . . . 135
A.2.2 Sanctions intensity variable construction . . . . . . . . . . . . . . . . . 138
A.2.3 U.S. Treasury sanctions variable construction . . . . . . . . . . . . . . 140
A.2.4 Sanctions dummy variables . . . . . . . . . . . . . . . . . . . . . . . . 142
A.2.5 Conversions from Iranian to Gregorian calendar . . . . . . . . . . . . 145
A.2.6 Economic and socio-demographic variables . . . . . . . . . . . . . . . 146
A.3 Computation of IRFs, FEVDs and their error bands by bootstrap . . . . . . . 151
A.3.1 Impulse response analysis for SVAR model of the Iranian economy . 151
A.3.2 Forecast error variance decompositions . . . . . . . . . . . . . . . . . 153
A.3.3 IRFs and FEVDs alternative computation . . . . . . . . . . . . . . . . 155
A.3.4 Bootstrapping procedure . . . . . . . . . . . . . . . . . . . . . . . . . . 157
A.4 Additional empirical results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
A.4.1 Reduced form output growth equation including current and lagged
sanction variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
A.4.2 Re-ordering the variables in the SVAR model . . . . . . . . . . . . . . 158
A.4.3 Using sanctions dummy variables . . . . . . . . . . . . . . . . . . . . 159
A.4.4 Additional IRFs and FEVDs . . . . . . . . . . . . . . . . . . . . . . . . 160
A.4.5 Additional results with robust standard errors . . . . . . . . . . . . . 161
A.4.6 Additional sanctions-augmented SVAR analyses . . . . . . . . . . . . 165
A.4.7 Additional IRFs and FEVDs results . . . . . . . . . . . . . . . . . . . 195
A.4.8 Additional analyses using heteroskedastic-consistent standard errors 200
A.5 Sanctions chronology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
Appendix B: Appendix to Chapter 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
B.1 Additional facts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
B.2 Model derivations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
B.3 Numerical solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
B.4 Empirical robustness checks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
B.5 Data sources and construction . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
vi

Appendix C: Appendix to Chapter 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
C.1 Anecdotal evidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
C.2 Additional theoretical proofs . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
C.3 Additional empirical evidence . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
C.4 Data sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
vii

List of Tables
1.1 Free market and official foreign exchange rate depreciation, inflation, real
output growth, and sanctions intensity over the period 1979q3–2021q1 . . . 11
1.2 Descriptive statistics of the sanctions intensity variable over relevant time
periods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.3 Estimates of the reduced form Iran’s output growth equation estimated over
the period 1989q1–2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.4 Quarterly estimates of the SVAR model of Iran with domestic variables ordered as: oil exports, exchange rate returns, money supply growth, inflation, and output growth, estimated over the period 1989q1-2019q4 . . . . . . 35
1.5 Forecast error variance decomposition for domestic variables in the SVAR
model with a single shock to sanctions . . . . . . . . . . . . . . . . . . . . . . 42
2.1 Calibrated parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
2.2 Quantitative results for the baseline model . . . . . . . . . . . . . . . . . . . 74
2.3 Policy experiments results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
2.4 Long-run co-variability results over the three relevant sub-periods . . . . . . 80
2.5a Regression results for the effects of inequality on total loans issuance for a
host of 18 economies over the period 1974-2019 . . . . . . . . . . . . . . . . . 84
2.5b Regression results for the effects of total loans on inequality for a host of 18
economies over the period 1974-2019 . . . . . . . . . . . . . . . . . . . . . . . 84
3.1 Quarterly estimates of the effect of funding competition on deregulation
over the period 1980q1-2008q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
3.2 Quarterly estimates of the effect of deregulation on funding competition
over the period 1980q1-2008q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
3.3 Effect of funding competition on off-balance items as a fraction of total assets. Sample period 1990q1-2007q4 . . . . . . . . . . . . . . . . . . . . . . . . 102
A.1 Sources of newspaper articles over the period 1979m1–2020m9 . . . . . . . . 138
A.2 Quarterly estimates of the log-likelihood of Equation (A.1) estimated over
the period 1989q1-2019q4 for values of w ∈ {0.1, 0.2, . . . 0.9} . . . . . . . . . 139
A.3 Sanctions dummy variable description over the period 1989q1-2019q4 . . . . 142
A.4 Discretized sanctions intensity variable description . . . . . . . . . . . . . . . 144
A.5 Sources of quarterly data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
viii

A.6 Quarterly estimates of the sanctions intensity variable AR(1) and AR(2)
models over the period 1989q1–2020q3 . . . . . . . . . . . . . . . . . . . . . . 162
A.7 Quarterly estimates of the world real output growth AR(1) and AR(2) models over the period 1989q1–2019q4 . . . . . . . . . . . . . . . . . . . . . . . . 162
A.8 Size of one standard error shock for the endogenous variables used in the
IRFs analyses in Figure 1.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
A.9 Reduced form Iran’s output growth equation including contemporaneous
sanctions variable and estimated over the period 1989q1–2019q4 . . . . . . . 164
A.10 Quarterly estimates of the equation for the oil export variable in the SVAR
model of Iran with domestic variables ordered as: oil exports, exchange
rate returns, money supply growth, inflation, and output growth, estimated
over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
A.11 Quarterly estimates of the equation for exchange rate returns in the SVAR
model of Iran with domestic variables ordered as: oil exports, exchange
rate returns, money supply growth, inflation, and output growth, estimated
over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
A.12 Quarterly estimates of the equation for money supply growth in the SVAR
model of Iran with domestic variables ordered as: oil exports, exchange
rate returns, money supply growth, inflation, and output growth, estimated
over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
A.13 Quarterly estimates of the equation for inflation in the SVAR model of Iran
with domestic variables ordered as: oil exports, exchange rate returns, money
supply growth, inflation, and output growth, estimated over the period
1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
A.14 Quarterly estimates of the equation for output growth in the SVAR model
of Iran with domestic variables ordered as: oil exports, exchange rate returns, money supply growth, inflation, and output growth, estimated over
the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
A.15 Quarterly estimates of the SVAR model of Iran with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output growth,
estimated over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . 170
A.16 Quarterly estimates of the equation for the oil exports variable in the SVAR
model of Iran with domestic variables ordered as: oil exports, exchange rate
returns, inflation, and output growth, estimated over the period 1989q1-
2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
A.17 Quarterly estimates of the equation for exchange rate returns in the SVAR
model of Iran with domestic variables ordered as: oil exports, exchange rate
returns, inflation, and output growth, estimated over the period 1989q1-
2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
A.18 Quarterly estimates of the equation for inflation in the SVAR model of Iran
with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output growth, estimated over the period 1989q1-2019q4 . . . . 173
A.19 Quarterly estimates of the equation for output growth in the SVAR model of
Iran with domestic variables ordered as: oil exports, exchange rate returns,
inflation, and output growth, estimated over the period 1989q1-2019q4 . . . 174
ix

A.20 Quarterly estimates of the SVAR model of Iran with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output growth,
estimated over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . 175
A.21 Quarterly estimates of the equation for the oil exports variable in the SVAR
model of Iran with domestic variables ordered as: oil exports, exchange rate
returns, inflation, and output growth, estimated over the period 1989q1-
2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
A.22 Quarterly estimates of the equation for exchange rate returns in the SVAR
model of Iran with domestic variables ordered as: oil exports, exchange rate
returns, inflation, and output growth, estimated over the period 1989q1-
2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
A.23 Quarterly estimates of the equation for inflation in the SVAR model of Iran
with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output growth, estimated over the period 1989q1-2019q4 . . . . 178
A.24 Quarterly estimates of the equation for output growth in the SVAR model of
Iran with domestic variables ordered as: oil exports, exchange rate returns,
inflation, and output growth, estimated over the period 1989q1-2019q4 . . . 179
A.25 Quarterly estimates of the SVAR model of Iran with domestic variables ordered as: exchange rate returns, oil exports, money supply growth, inflation, and output growth, estimated over the period 1989q1-2019q4 . . . . . . 180
A.26 Quarterly estimates of the equation for the exchange rate returns in the
SVAR model of Iran with domestic variables ordered as: exchange rate returns, oil exports, money supply growth, inflation, and output growth, estimated over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . 181
A.27 Quarterly estimates of the equation for the oil export variable in the SVAR
model of Iran with domestic variables ordered as: exchange rate returns,
oil exports, money supply growth, inflation, and output growth, estimated
over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
A.28 Quarterly estimates of the equation for money supply growth in the SVAR
model of Iran with domestic variables ordered as: exchange rate returns,
oil exports, money supply growth, inflation, and output growth, estimated
over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
A.29 Quarterly estimates of the equation for inflation in the SVAR model of Iran
with domestic variables ordered as: exchange rate returns, oil exports, money
supply growth, inflation, and output growth, estimated over the period
1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
A.30 Quarterly estimates of the equation for output growth in the SVAR model
of Iran with domestic variables ordered as: exchange rate returns, oil exports, money supply growth, inflation, and output growth, estimated over
the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
A.31 Quarterly estimates of the SVAR model of Iran with domestic variables ordered as: exchange rate returns, oil exports, inflation, and output growth,
estimated over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . 186
x

A.32 Quarterly estimates of the equation for exchange rate returns in the SVAR
model of Iran with domestic variables ordered as: exchange rate returns,
oil exports, inflation, and output growth, estimated over the period 1989q1-
2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
A.33 Quarterly estimates of the equation for the oil exports variable in the SVAR
model of Iran with domestic variables ordered as: exchange rate returns,
oil exports, inflation, and output growth, estimated over the period 1989q1-
2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
A.34 Quarterly estimates of the equation for inflation in the SVAR model of Iran
with domestic variables ordered as: exchange rate returns, oil exports, inflation, and output growth, estimated over the period 1989q1-2019q4 . . . . 189
A.35 Quarterly estimates of the equation for output growth in the SVAR model of
Iran with domestic variables ordered as: exchange rate returns, oil exports,
inflation, and output growth, estimated over the period 1989q1-2019q4 . . . 190
A.36 Estimates of the reduced form Iran’s output growth equation using a sanctions dummy variable estimated over the period 1989q1- 2019q4 . . . . . . . 191
A.37 Reduced form Iran’s output growth equation using a discretized sanctions
intensity variable estimated over the period 1989q1–2019q4 . . . . . . . . . . 192
A.38 Quarterly estimates of the SVAR model of Iran using a sanctions dummy
variable and with domestic variables ordered as: oil exports, exchange rate
returns, money supply growth, inflation, and output growth, estimated
over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
A.39 Quarterly estimates of the SVAR model of Iran using a discretized sanctions
intensity variable and with domestic variables ordered as: oil exports, exchange rate returns, money supply growth, inflation, and output growth,
estimated over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . 194
A.40 Forecast error variance decomposition in the SVAR model with domestic
variables ordered as exchange rate returns, oil exports, inflation, and output
growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
A.41 Estimates of the reduced form Iran’s output growth equation estimated over
the period 1989q1–2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
A.42 Estimates of the reduced form of the output growth equation for Iran including contemporaneous sanctions variable and estimated over the period
1989q1–2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
A.43 Quarterly estimates of the SVAR model of Iran with domestic variables ordered as: oil exports, exchange rate returns, money supply growth, inflation, and output growth, estimated over the period 1989q1-2019q4 . . . . . . 202
A.44 Quarterly estimates of the equation for the oil export variable in the SVAR
model of Iran with domestic variables ordered as: oil exports, exchange
rate returns, money supply growth, inflation, and output growth, estimated
over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
A.45 Quarterly estimates of the equation for exchange rate returns in the SVAR
model of Iran with domestic variables ordered as: oil exports, exchange
rate returns, money supply growth, inflation, and output growth, estimated
over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
xi

A.46 Quarterly estimates of the equation for money supply growth in the SVAR
model of Iran with domestic variables ordered as: oil exports, exchange
rate returns, money supply growth, inflation, and output growth, estimated
over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
A.47 Quarterly estimates of the equation for inflation in the SVAR model of Iran
with domestic variables ordered as: oil exports, exchange rate returns, money
supply growth, inflation, and output growth, estimated over the period
1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
A.48 Quarterly estimates of the equation for output growth in the SVAR model
of Iran with domestic variables ordered as: oil exports, exchange rate returns, money supply growth, inflation, and output growth, estimated over
the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
A.49 Quarterly estimates of the SVAR model of Iran with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output growth,
estimated over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . 208
A.50 Quarterly estimates of the equation for the oil exports variable in the SVAR
model of Iran with domestic variables ordered as: oil exports, exchange rate
returns, inflation, and output growth, estimated over the period 1989q1-
2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
A.51 Quarterly estimates of the equation for exchange rate returns in the SVAR
model of Iran with domestic variables ordered as: oil exports, exchange rate
returns, inflation, and output growth, estimated over the period 1989q1-
2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
A.52 Quarterly estimates of the equation for inflation in the SVAR model of Iran
with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output growth, estimated over the period 1989q1-2019q4 . . . . 211
A.53 Quarterly estimates of the equation for output growth in the SVAR model of
Iran with domestic variables ordered as: oil exports, exchange rate returns,
inflation, and output growth, estimated over the period 1989q1-2019q4 . . . 212
A.54 Quarterly estimates of the SVAR model of Iran with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output growth,
estimated over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . 213
A.55 Quarterly estimates of the equation for the oil exports variable in the SVAR
model of Iran with domestic variables ordered as: oil exports, exchange rate
returns, inflation, and output growth, estimated over the period 1989q1-
2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
A.56 Quarterly estimates of the equation for exchange rate returns in the SVAR
model of Iran with domestic variables ordered as: oil exports, exchange rate
returns, inflation, and output growth, estimated over the period 1989q1-
2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
A.57 Quarterly estimates of the equation for inflation in the SVAR model of Iran
with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output growth, estimated over the period 1989q1-2019q4 . . . . 216
xii

A.58 Quarterly estimates of the equation for output growth in the SVAR model of
Iran with domestic variables ordered as: oil exports, exchange rate returns,
inflation, and output growth, estimated over the period 1989q1-2019q4 . . . 217
A.59 Quarterly estimates of the SVAR model of Iran with domestic variables ordered as: exchange rate returns, oil exports, money supply growth, inflation, and output growth, estimated over the period 1989q1-2019q4 . . . . . . 218
A.60 Quarterly estimates of the equation for the exchange rate returns in the
SVAR model of Iran with domestic variables ordered as: exchange rate returns, oil exports, money supply growth, inflation, and output growth, estimated over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . 219
A.61 Quarterly estimates of the equation for the oil export variable in the SVAR
model of Iran with domestic variables ordered as: exchange rate returns,
oil exports, money supply growth, inflation, and output growth, estimated
over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
A.62 Quarterly estimates of the equation for money supply growth in the SVAR
model of Iran with domestic variables ordered as: exchange rate returns,
oil exports, money supply growth, inflation, and output growth, estimated
over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
A.63 Quarterly estimates of the equation for inflation in the SVAR model of Iran
with domestic variables ordered as: exchange rate returns, oil exports, money
supply growth, inflation, and output growth, estimated over the period
1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
A.64 Quarterly estimates of the equation for output growth in the SVAR model
of Iran with domestic variables ordered as: exchange rate returns, oil exports, money supply growth, inflation, and output growth, estimated over
the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
A.65 Quarterly estimates of the SVAR model of Iran with domestic variables ordered as: Exchange rate returns, oil exports, inflation, and output growth,
estimated over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . 224
A.66 Quarterly estimates of the equation for exchange rate returns in the SVAR
model of Iran with domestic variables ordered as: exchange rate returns,
oil exports, inflation, and output growth, estimated over the period 1989q1-
2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
A.67 Quarterly estimates of the equation for the oil exports variable in the SVAR
model of Iran with domestic variables ordered as: exchange rate returns,
oil exports, inflation, and output growth, estimated over the period 1989q1-
2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
A.68 Quarterly estimates of the equation for inflation in the SVAR model of Iran
with domestic variables ordered as: exchange rate returns, oil exports, inflation, and output growth, estimated over the period 1989q1-2019q4 . . . . 227
A.69 Quarterly estimates of the equation for output growth in the SVAR model of
Iran with domestic variables ordered as: exchange rate returns, oil exports,
inflation, and output growth, estimated over the period 1989q1-2019q4 . . . 228
A.70 Estimates of the reduced form Iran’s output growth equation using a sanctions dummy variable estimated over the period 1989q1- 2019q4 . . . . . . . 229
xiii

A.71 Quarterly estimates of the reduced form Iran’s output growth equation using a discretized sanctions intensity variable estimated over the period 1989q1-
2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
A.72 Quarterly estimates of the SVAR model of Iran using a sanctions dummy
variable and with domestic variables ordered as: oil exports, exchange rate
returns, money supply growth, inflation, and output growth, estimated
over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
A.73 Quarterly estimates of the SVAR model of Iran using a discretized sanctions
intensity variable and with domestic variables ordered as: oil exports, exchange rate returns, money supply growth, inflation, and output growth,
estimated over the period 1989q1-2019q4 . . . . . . . . . . . . . . . . . . . . . 232
A.74 Chronology of major sanctions events against Iran over the period from
November 1979 to January 2021 . . . . . . . . . . . . . . . . . . . . . . . . . . 234
B.1 Regression results for the short- and long-run effects of inequality on total
loans for a host of 18 economies over the period 1970-2019 . . . . . . . . . . 251
B.2 Regression results for the short- and long-run effects of total loans on inequality for a host of 18 economies over the period 1970-2019 . . . . . . . . . 251
B.3 Regression results for the short- and long-run effects of inequality on total
loans using the Arellano-Bond estimator for a host of 18 economies over the
period 1970-2019 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
B.4 Regression results for the short- and long-run effects of total loans on inequality using the Arellano-Bond estimator for a host of 18 economies over
the period 1970-2019 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
B.5 Variables description and sources . . . . . . . . . . . . . . . . . . . . . . . . . 254
B.6 Construction of macro-financial variables with series references . . . . . . . 257
C.1 Gross real and nominal growth rates for each sector of the economy . . . . . 268
C.2 Effect of funding competition on deregulation in the United States over the
period 1980q1-2008q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
C.3 Data sources utilized in both the descriptive statistics and empirical analysis 272
xiv

List of Figures
1.1 Sanctions intensity variable over the period 1989q1–2020q3 . . . . . . . . . . 17
1.2 Impulse responses of the effects of sanctions and domestic shocks on oil
exports, foreign exchange, inflation, and output growth . . . . . . . . . . . . 38
1.3 Forecast error variance decomposition for domestic variables in the SVAR
model with a cumulative shock to sanctions, and domestic variables ordered
as oil exports, exchange rate returns, inflation, and output growth . . . . . . 43
2.1 Gross and net measures of financial assets intermediated as a share of GDP
and top 1 per cent income share in the United States over the long run . . . . 50
2.2 Traditional and shadow banking assets as a share of GDP and top 1 per cent
income share (left); and domestically-held safe assets as a share of GDP
and top 1 per cent income share (right), in the United States over the period
1960-2019 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.3 Decomposition of the Unites States financial assets held domestically vis-àvis by the rest of the world (left); capital income share and top 5 per cent
wealth share in the United States over the period 1960-2019 (right) . . . . . 57
2.4 Visual representation of the model . . . . . . . . . . . . . . . . . . . . . . . . 67
2.5 Sensitivity analysis under different volatility scenarios . . . . . . . . . . . . . 69
2.6 Sensitivity analysis under different capital share scenarios . . . . . . . . . . . 70
2.7 Long Run Projections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
2.8 Total loans to GDP ratio, and top 5 per cent income share for a panel of 18
advanced economies over the period 1912-2019 . . . . . . . . . . . . . . . . . 82
3.1 Traditional bank assets relative to the total amount of assets issued by other
non-bank financial institutions in the United States over the period 1960q1
– 2019q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
3.2 Banks off-balance sheet assets vis-à-vis non-bank financial institution assets
(left), and funding competition vis-à-vis deregulation index (right) . . . . . 94
3.3 Local projections over the period 1980q1–2008q4 . . . . . . . . . . . . . . . . 101
3.4 Model baseline description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
3.5 Model description: Deregulation extension. . . . . . . . . . . . . . . . . . . . 117
3.6 Model description: Off-balance sheet extension. . . . . . . . . . . . . . . . . . 120
A.1 Sanctions intensity variable and standardized sanctions intensity variable
over the period 1989q1–2020q3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
xv

A.2 Sanctions intensity variable and the U.S. Treasury sanctions variable over
the period 2006q1–2020q3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
A.3 Sanctions intensity variable, and discretized sanctions intensity variable,
over the period 1989q1–2020q3 . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
A.4 Relevant Iran’s and World macroeconomic and financial time series over the
period 1979–2020 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
A.5 Impulse responses of the effects of a world output shock on oil exports, foreign exchange, inflation, and output growth . . . . . . . . . . . . . . . . . . . 195
A.6 Impulse responses of the effects of sanctions and domestic shocks on foreign
exchange, oil exports, inflation, and output growth . . . . . . . . . . . . . . . 196
A.7 Impulse responses of the effects of sanctions and domestic shocks on foreign
exchange, oil exports, inflation, and output growth . . . . . . . . . . . . . . . 197
A.8 Forecast error variance decomposition for domestic variables in the SVAR
model with a cumulative shock to sanctions, and domestic variables ordered
as FX returns, oil exports, inflation, and output growth . . . . . . . . . . . . 199
B.1 Share of U.S. Treasuries held domestically vis-à-vis abroad over the period
1970–2019 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
B.2 Domestically-held private safe assets as a share of domestically-held public
safe assets, and shadow banking sector as a share of traditional banking
sector in the United States over the period 1950-2019 . . . . . . . . . . . . . . 241
B.3 Top 1 percent share of the income distribution and households’ portfolio
share invested in institutional investors in the United States over the period
1971-2019 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
B.4 Quantile plot showing the correlation between top 5 per cent share of the
income distribution and credit to GDP (in logs) for 18 economies over the
period 1970-2019 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
B.5 Income volatility by income level in the United States across quantiles over
the period 1998-2019 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
C.1 Anecdotal evidence of regulators and traditional banks assessment of the
changing banking landscape . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
C.2 Proportion of bank loans relative to the total amount of loans issued by
financial institutions (in blue), and shadow banks (in red) in the United
States over the period 1960q1 – 2019q4 . . . . . . . . . . . . . . . . . . . . . . 268
xvi

Abstract
This dissertation on macroeconomics and macro-finance is composed of three chapters.
The order has been established in terms of each chapter’s readiness for publication on
academic reviews. The different nature of the projects with regard to their methodology
and the specific themes treated therein tracks my intellectual evolution during the doctoral
years.
The first chapter is "Identifying the Effects of Sanctions on the Iranian Economy using
Newspaper Coverage". It is joint work with M. Hashem Pesaran, and it has been published
on the Journal of Applied Econometrics as lead article (Laudati and Pesaran, 2023). The
following broad and daunting questions motivate the paper: What is the long-run economic effect of sanctions on the Iranian economy over the past 30 years (if any); and which
channels are key to understand the observed dynamics. We innovate both the time series
and the sanctions literature by proposing a new identification and measurement strategy
to answer the previous research questions. Identification-wise, we show that a global factor model with a small (but isolated) open economy is able to overcome the theoretical
shortcomings present in the "difference-in-differences" and in the "synthetic control methods" literature. In terms of quantitative estimation, we build a new sanctions index based
on the newspapers coverage of sanctions. By doing so, we innovate on the previous studies, which treated sanctions as a dichotomous variable with arbitrary starting and ending
periods. We find that over the 1989-2019 period Iran’s average annual growth could have
been around 4-5 per cent, as compared to the 3 per cent realized. Estimates of the proposed
sanctions-augmented structural vector autoregressive model show that sanctions significantly decrease oil export revenues, result in substantial depreciation of the Iranian rial,
followed by subsequent increases in inflation and falls in output growth. Keeping other
shocks fixed, two years of sanctions can explain up to 60 per cent of output growth forecast error variance, although a single quarter sanction shock proves to have quantitatively
xvii

small effects.
The second chapter – "Inequality and the Rise of Finance" – tackles an equally big question from a theoretical point of view. It studies the causes behind the rise of the financial
sector observed in the United States from the 1980s. The theory proposed claims that the
growth of the financial sector (interpreted as the amount of financial assets intermediated
as a share of GDP) can be seen as an endogenous rise of non-bank financial institutions
(shadow banking sector) reacting to secular forces in the macro-economy. The paper connects for the first time disjoint parts of the macroeconomics and finance literature in a
single internally-consistent framework according to the following lines. An increase in
wealth inequality induces a higher amount of savings to invest in the hands of the wealthier households – the investors. Investors need to allocate their holdings between risky and
safe assets following portfolio optimization decisions. Given a constrained supply of public safe assets, real interest rates decline to accommodate the larger demand. A compression of the real interest rates reduces the costs of issuing debt for the poorer households,
and represents the incentive for the shadow banking system to step in. The shadow banks
complete a market by transforming the debt that the poorer households use to finance part
of their consumption into the private safe assets that the investors demand to hedge their
risks. As such, the shadow banking sector rises (at least in part) as a result of a domestic
safe asset shortage. Also, one of the model’s additional contributions is to allow for an
endogenous and non-mechanical feedback loop between inequality and finance. Furthermore, the primitive increase in wealth inequality is obtained through non-trivial dynamics
generated by an exogenous decline of the labor share, which allows to connect for the first
time to my knowledge the production technology of an economy with its banking structure. The paper is quantitative in spirit with a few empirical exercises corroborating the
model predictions.
The final chapter is "The Political Economy of Banking and Shadow Banking Competition". This project is virtually nested into the previous. It starts by taking the larger inflow
xviii

of funds towards the financial sector as being the result of a competitive advantage driven
by regulatory arbitrage. With that in hand, it attempts to provide a macro-banking theory for the endogenous emergence of financial deregulation from the 1980s. If traditional
banks become increasingly squeezed by the better shadow banking competitors, then this
provides a rationale to lobby harder in order to obtain financial deregulation and level the
playing field. Interestingly, the theory predicts that – if traditional banks can gain an alternative competitive advantage by investing in financial innovation – a stiffer regulatory
approach may induce a faster increase in banking activities. This is the result of banks
pushing even more strongly on activities outside the scope of regulation to keep up with
other financial institutions. As such, the paper is not only the first to provide a root cause
explanation for the rise and timing of financial deregulation in the U.S.. It also sheds light
on the importance of carefully crafting new regulation, and potentially include financial
innovation itself into the scope of macro-prudential policies. The preliminary time series
analyses are in line with the previous points, although a more granular battery of tests
will be needed in the future to further corroborate the theoretical elements just described.
To maintain a relatively parsimonious structure of the dissertation, the published paper on geographic inequality in Italy – "Accounting for the Duality of the Italian Economy"
– has been omitted (Fernández-Villaverde, Laudati, Ohanian, and Quadrini, 2023). It remains part of the work carried out over these past few years, and it may be the starting
point for more projects on this subject in the years to come.
xix

Chapter 1
Identifying the Effects of Sanctions on
the Iranian Economy using Newspaper
Coverage1
1.1 Introduction
Over the past forty years Iran has been subject to varying degrees of economic and financial sanctions and asset freezes, which began in November 1979 when the U.S. placed
an embargo on Iranian oil trade and froze $12 billion of Iranian assets held outside Iran
with the aim of securing the release of U.S. hostages. Although this particular sanction
episode was successfully negotiated in January 1981, U.S. policy towards Iran became increasingly entrenched, aimed at curtailing the economic and political influence of Iran in
the Middle East region and beyond; a process which escalated over Iran’s nuclear program. As a result, the Iranian economy has been operating for a prolonged period under
severe and often quite harsh international restrictions, perhaps unique for a sizable economy with deep historical roots in the global economy. Given the uncertainty and durability of sanction regimes, it is also important to bear in mind that, besides the direct effects
of sanctions (arising from loss of oil export revenues, loss of access to currency reserves
1This chapter is joint work with M. Hashem Pesaran (Trinity College – University of Cambridge, and University of
Southern California). This article, published on the Journal of Applied Econometrics, is available under the Creative
Commons CC-BY-NC-ND license and permits non-commercial use of the work as published, without adaptation or
alteration provided the work is fully attributed.
1

and other trade-related losses), sanctions also result in important and lasting indirect effects, such as rent-seeking, resource allocation distortions, and general costs associated
with efforts involved in mitigating and circumventing the sanctions regimes. These indirect effects are likely to be more serious the longer the sanctions are in place, particularly
when the prospect of a sanctions free outcome seems very remote.
The focus of this paper is on the identification and quantitative evaluation of the direct
and indirect effects of sanctions on the Iranian economy over the period 1989–2019, which
intentionally excludes the period 1979–1988 due to the special circumstances of the 1979
Revolution, the hostage crises and the ensuing eight year war with Iraq, which ended
in August 1988, as well as the post 2019 period to avoid the confounding effects of the
Covid-19 pandemic. We are primarily concerned with economic rather than international
political dimensions of the sanctions, and will not be addressing the issue of the efficacy
of sanctions in achieving their political aims.2
Sanctions against Iran span a period of more than four decades over which the degree of sanctions intensity has varied considerably. There are no clear "sanctions on" and
sanctions off" periods, required for application of comparative approaches used in the literature for policy evaluations, such as the synthetic control method (SCM) proposed by
Abadie and Gardeazabal (2003), and the panel data approach proposed by Hsiao, Ching,
and Wan (2012). These techniques require pre-policy intervention outcomes to estimate
weighted averages of post policy outcomes for a "pre-selected" control group to be used
as the basis of comparisons. It is also unclear which countries should be included in the
control group given the continued importance of the Iranian economy in the Middle East
region.
2The effectiveness of sanctions in achieving foreign policy goals has been studied extensively in the literature. Hufbauer,
Schott, and Elliott (1990) examine 116 case studies covering the period from the economic blockade of Germany during
World War I through the U.N.-U.S. embargo of Iraq in 1990. Further overviews are provided in Morgan, Bapat, and
Kobayashi (2014).and Doxey (1996). Critical assessments of sanctions as a policy tool are provided by Weiss, Cortright,
Lopez, and Minear (1997), Pape (1997, 1998), Andreas (2005), and Peksen and Drury (2010). These studies highlight
possible counterproductive effects of economic sanctions. Naghavi and Pignataro (2015) provide a game-theoretic
analysis of sanctions and its application to Iran.
2

In this paper we propose a time series approach that takes account of variations in sanctions intensity over the past forty years, without requiring an arbitrary classification of the
time periods into sanctions on and sanctions off periods. To this end, we construct a time
series index of sanctions intensity based on daily newspaper coverage of the sanctions,
their imposition, the intensity of their use, as well as their occasional removal. Given the
absence of clear "sanctions off" periods, it follows also that simple (0,1) dummy variables
may not be sufficient to capture the rich variations in sanctions intensity that are observed
over the past forty years.3
The idea of a newspaper coverage index was developed by Baker, Bloom, and Davis
(2016) for measurement of economic uncertainty, but to our knowledge it has not been
utilized in the analysis of sanctions. As we shall see, the evolution of our proposed sanctions intensity index closely tracks the main sanctions time points.4 See Figure 1.1. The
sanctions intensity measure also correlates closely with the U.S. Treasury "Specially Designated Nationals and Blocked Persons List" (SDN) for Iran which has been publicly available since 1994, but with usable data on Iran only since 2006.
We augment a structural vector autoregressive (SVAR) model of the Iranian macroeconomy with our sanctions intensity variable to identify short run and long run effects
of sanction shocks on oil export revenues, Iran’s rial/USD exchange rate, money supply
growth, inflation, and output growth, whilst controlling for foreign output growth, and
other global factors such as global equity market volatility. We also consider the effects of
sanction shocks when sustained over a prolonged period. This is particularly relevant to
the case of Iran where government policy responses often have led to large fuel and food
3To investigate the value added of our proposed measure of sanctions intensity, as suggested by a referee, we also considered dummy variables constructed based on historical narratives, as well as by a discretization of our own newspapers
index. In all cases we found our sanction intensity variable performs much better than dummy variable measures in
explaining the variations in key macroeconomic variables of the Iranian economy.
4The most notable are: the U.S. Iran and Libya Sanctions Act of 1996, the U.S. export ban in 1997, the U.S. investment bans
and asset freezes in 2006 and 2007 ("Iran Freedom and Support Act", and Executive Order 13438), the United Nations
nuclear Resolutions (1737, 1747) during 2006 and 2007, the U.S. Comprehensive Iran Sanctions, Accountability, and
Divestment Act of 2010, the U.S. National Defence Authorization Act of 2012, the partial lifting of U.N. sanctions under
the Joint Comprehensive Plan of Action (JCPOA) in 2015 and its subsequent implementation in January 2016, and
finally President Trump’s unilateral withdrawal from the JCPOA agreement in 2018.
3

subsidies, multiple exchange rates, and lax budgetary and credit policies, which in turn
have resulted in economic mismanagement and rent-seeking, and corruption on a large
scale.5 Seen from this more general perspective, in addition to direct (and often immediate) effects of sanctions on oil exports and exchange rates, there are also indirect effects
that result from government policy responses, some of which are inevitable as the government tries to come to terms with the adverse effects of sanctions, particularly on the
economic conditions of the low income groups on fixed wages and salaries. Whilst we
acknowledge such indirect effects of sanctions, it is beyond the scope of the present paper to disentangle the direct and indirect effects of sanctions. This drawback particularly
applies to the long run effects of the sanctions that move beyond the immediate effects
on oil exports and exchange rates that are much easier to identify. At the same time, it is
true that the Iranian economy would have been subject to distortions and economic mismanagement even in the absence of any sanctions, and it seems unlikely that one could
separate sanctions-induced distortions from all other distortions. Therefore, the estimates
we present can be viewed as measuring the combined effects of sanctions and sanctionsinduced distortions, broadly defined.
We find that the sanctions intensity variable has highly statistically significant effects
on oil exports revenues, exchange rates, inflation and output growth, but not on money
supply growth. These estimates proved to be robust to alternative specifications and after
allowing for a host of control variables. Our results also show that large reductions in oil
exports, currency depreciations (with substantial overshooting), and high inflation are
important channels through which sanctions affect the real economy. But we do not find
monetary expansion to have an independent impact on the real economy, once we control
for inflation and exchange rates.
Using impulse response analysis and forecast error variance decomposition, we also
5
Subsidies on essential food items and energy (fuel as well as electricity) have created inefficiencies, smuggling, and
damaging unintended consequences. Subsidies on electricity, for example, have led to excessive ground water withdrawals from electricity-powered irrigation wells, and more recently for mining crypto-currencies, one of the sources
of Iran’s worsening water shortages, and frequent electricity blackouts.
4

find a significant drop in oil exports, followed by an over-reaction of the rial to a positive sanctions shock, with a subsequent rise in inflation and a fall in output shortly after.
The economy adapts reasonably quickly to sanction shocks, a property that has also been
documented by Esfahani, Mohaddes, and Pesaran (2013), who consider the effects of oil
revenue shocks on output growth and inflation, but do not allow for changes in sanctions
environment. Forecast error variance decompositions also show that, despite the inclusion
of the sanctions intensity variable in the SVAR, around 80 per cent of variations in foreign
exchange and 82 per cent of variations in output growth remain unexplained, and most
likely relate to many other latent factors that drive the Iranian economy. These estimates
relate to the effects of a single-period sanctions shock as it is standard in the empirical
macro literature. The effects of sanctions on the economy become much more pronounced
once we consider the effects of shocks that last for a consecutive number of periods. We
find that such shock scenarios could explain up to 80 per cent of output growth variations
after five years. We also estimate that in the absence of sanctions Iran’s output growth on
average could have been around 4–5 per cent per annum, as compared to the 3 per cent
realized.
Sanctions have also had a number of positive unintended consequences. Interestingly,
the Iranian economy at the onset of sanctions was as heavily dependent on oil exports as
countries such as Saudi Arabia. Restricting oil exports over a relatively long time period
has led to important structural transformations of the Iranian economy, with significant
increases in non-oil exports, most notably petrochemicals, light manufacturing products
and agricultural goods. In addition, it is likely that U.S. sanctions have been partly responsible for the rapid rise of high-tech and knowledge-based companies in Iran over the past
decade.
Overall, there seems little doubt that sanctions have harmed the Iranian economy. But
removal of sanctions on their own is unlikely to ensure a period of sustained growth
and low and stable inflation, and many policy reforms are needed to address sanctions5

induced price distortions as well as other distortions due to general economic mismanagement, poor governance, and the ambiguities that surround the relative roles of semigovernment agencies and the private sector in the economy.
Related literature. Studies that are more closely related to our paper either consider a
specific sub-period or use shocks to oil export revenues as representing a sanction shock.
Gharehgozli (2017) considers the effects of sanctions just before the 2015 nuclear agreement, Joint Comprehensive Plan of Action (JCPOA), which we discuss in further detail in
Section 1.4 below. Dizaji and van Bergeijk (2013) study the impact of economic sanctions
via changes in oil revenues using an unrestricted VAR model. They show that sanctions
have adverse output effects in the short-run but their effects fade with time. Similar results
are reported by Esfahani, Mohaddes, and Pesaran (2013), who find that shocks to foreign
output and oil exports are rather short-lived for Iran. This is an important feature of the
Iranian economy which is also confirmed by our analysis using the new sanctions intensity variable. Farzanegan, Khabbazan, and Sadeghi (2016) develop a Computable General
Equilibrium model for Iran and conduct a number of different comparative static exercises,
finding large effects of oil sanctions on the macro-economy and households welfare under
their perfectly competitive set up.
Haidar (2017) uses micro-data over the period 2006–2011 to find that two-thirds of
Iran’s sanctioned non-oil exports were redirected to other non-sanctioning countries. It
is also found that large exporters appear to be less affected by export sanctions. Popova
and Rasoulinezhad (2016) find a similar geographical redirection of Iran’s non-oil exports
over the period 2006–2013 of trading partners away from Western economies to countries
in the region (notably Iraq), China and other Asian economies. Farzanegan (2014) studies
the role of military spending to explain output losses due to oil shocks. Farzanegan and
Hayo (2019) analyze the effect of sanctions to expand the shadow economy. Although not
strictly quantitative in nature, a number of studies maintained that the burden of economic
6

sanctions for Iranian growth was high but not decisive to bring about political change in
Iran. (Carswell, 1981; Amuzegar, 1997a; Amuzegar, 1997b; Dadkhah and Zangeneh, 1998;
Downs and Maloney, 2011; and Borszik, 2016).
Sanctions have also played an important role in shaping Iran’s monetary and financial
system. Mazarei (2019) analyzes the current state of the Iranian financial system and its
fragility. Farzanegan and Markwardt (2009) focus on the extent to which Iran suffers from
a form of "Dutch disease", thus advocating for a sovereign oil fund to mitigate inflationary
pressures and risks of currency crises. Mazarei (2020) highlights the danger of inflation
for Iran in the wake of sanctions and the pandemic. There are also several studies on the
determinants of inflation in Iran (not related to sanctions), which could be of interest. See,
for example, the studies by Liu and Adedeji (2000), Celasun and Goswami (2002), and
Bonato (2008).
Sanctions have often led to the establishment of multiple exchange rate markets with
important rent-seeking opportunities and related economic distortions. Currently, there
are three different exchange rates for the rial.6 Bahmani-Oskooee (1996) provides an earlier account of the gains obtainable in Iran from the black market premium, and the need to
consider the free market rate rather than the official one when the Iranian money demand
is to be assessed; we follow this approach when conducting our analyses and disregard the
official rate. The economic implications of multiple exchange rates in Iran are discussed
in Pesaran (1992), Farzanegan (2013) and Majidpour (2013).
Our contributions are both methodological, by expanding the program evaluation literature with a novel econometric approach, and empirical in terms of the measurement of
sanction intensity using textual analysis and its incorporation in a quarterly macroeconometric model of Iran, which has not been done before.7
6The three exchange rates are: (i) The official exchange rate to import essential items such as medicine, grain and sugar;
(ii) The Nima rate, officially set at 2 per cent above the official rate by Bank Markazi daily, but in practice it is subject to
much higher premiums and is reserved for non-oil exporters; (iii) The free market rate used for all other transactions.
7We are not aware of any study that is able to use data at quarterly frequency for over thirty years to evaluate the long-run
effect of sanctions. This is relevant for the dynamics of the SVAR model and increases the precision of our estimates.
7

The rest of the paper is organized as follows. Section 1.2 offers an overview of the
Iranian economy under sanctions. Section 1.3 explains how we construct the sanctions intensity index from six leading newspapers, and its co-movements with historical events.
Section 1.4 discusses alternative approaches to the analysis of policy interventions, and
develops a framework with latent factors used to identify the effects of sanctions on the
Iranian economy, as well as providing an estimate of sanctions-induced output losses. Section 1.5 reports estimates of sanctions-augmented SVAR models, and discusses the channels through which sanctions affect the Iranian economy. Impulse responses and error
variance decompositions are presented and their robustness to a different ordering of the
variables in the SVAR model are discussed. Section 1.6 ends with some concluding remarks. The chapter’s appendix provides details on the construction of our sanctions intensity variable, the data sources, with further methodological notes and empirical results.
The same appendix also contains a comprehensive list of economic and financial sanctions
imposed against Iran over the past forty years.
1.2 Sanctions and the Iranian economy: an overview
The evolution of the Iranian economy over the past forty years has been largely shaped
by the Revolution and the eight-years war with Iraq (1979-1988), prolonged episodes of
economic and financial sanctions, and often very different policy responses to sanctions
and economic management under the four presidents that have held office over the period
1989-2019. Initially, U.S. sanctions were much more clearly targeted. The goal of the 1980–
81 sanctions was to negotiate the release of U.S. hostages, and the 1987 sanctions to end
hostilities in the Persian Gulf and bring about an end to the war with Iraq. These sanctions
aimed at limiting Iran’s access to foreign exchange earnings through asset freezes and,
more importantly, by reducing Iran’s capacity and ability to produce and export oil.8
Iran’s oil exports had been already cut by half from the pre-Revolution peak of 6 mil8
For an overview of U.S. sanctions against Iran see also Chapter 9 of Maloney (2015).
8

lions barrels per day (mb/d).9 The first U.S. sanctions drove Iran’s oil exports to the low
of 700,000 b/d before recovering somewhat after the sanctions were lifted in January 1981.
However, since the lifting of the sanctions coincided with the intensification of the war
with Iraq, oil exports did not recover fully till after the war ended in 1988. From 1989 to
2005, with improvements in the diplomatic relationships between Iran and the U.S. and
other Western countries, oil exports started to rise and stabilized to around 2.5 mb/d under the presidencies of Rafsanjani (1989q3–1997q2) and Khatami (1997q3–2005q2). Oil
exports began to decline again from 2007 after the imposition of U.S. and U.N. sanctions in
December 2006 aimed at halting Iran’s uranium enrichment program which had gathered
pace under the newly elected President Ahmadinejad (2005q3-2013q2). Initially, sanctions targeted investments in oil, gas and petrochemicals, and exports of refined products,
but were later expanded to include banking, insurance and shipping. Additional financial
sanctions were imposed on Iran from July 2013. The coverage of U.N. and U.S. sanctions increased well beyond the oil and gas sectors and affected all aspects of Iranian foreign trade
and international finance, and even the international payment system of Bank Markazi
(Iran’s Central Bank). The extensive coverage of the sanctions, their multilateral nature,
coupled with the start of President Rouhani’s moderate administration (2013q3–2021q2),
paved the way for the 2015 nuclear agreement (JCPOA), implemented in January 2016
which led to the easing of some of the U.S. sanctions and the lifting of U.N. and European
Union sanctions against Iran. But the benefits of the JCPOA to Iran were limited, as many
non-U.S. global companies and banks were hesitant to deal with Iran because of the remaining U.S. sanctions, as well as concerns over money laundering, opacity of ownership,
and the fragility of the Iranian banking system. As it turned out, JCPOA was also short
lived, with oil exports sharply declining after May 2018, when U.S. President Trump unilaterally withdrew from JCPOA, and adopted the policy of "maximum pressure" against
Iran. With the election of President Biden in November 2020, there are negotiations for
9
See Panel A of Figure A.4 in the Appendix.
9

the U.S. to return to the 2015 nuclear agreement, although our analysis will be pre-dating
these negotiations.
The U.S. sanctions against Iran were mainly of extra-territorial nature. Iran-U.S. trade
had already been cut drastically after the Revolution and did not recover after the resolution of the hostage crisis. In response to sanctions, the Iranian government made concerted
efforts to re-direct Iran’s foreign trade from the West to the East and to the neighboring
countries. The sources of foreign exchange were also diversified from oil to non-oil exports of goods and services. The share of oil and gas exports declined steadily from 96
per cent of total exports in 1979 to around 60 per cent in 2018, before the full impact of
the U.S. withdrawal from Iran’s oil exports.10 Over the same period non-oil exports have
increased from 753 million dollars to 37 billion dollars.
In contrast, Iran was not able to adapt to financial sanctions sufficiently quickly, resulting in large depreciations of the free market rate of the rial against the U.S. dollar,
with the official rate lagging behind for a number of years, thus creating opportunities for
rent-seeking and often corrupt business practices.11 Given the relevance of imports in the
Iranian economy, and the role of the U.S. dollar as the store of value and as a hedge against
inflation for many Iranians, the fall in value of the rial quite rapidly translates into higher
consumer prices, with the rise in prices somewhat moderated due to government imports
of food and medicine at official rates. But, as the gap between the official and market rates
closes over time, consumer prices end up reflecting the full extent of depreciation of the rial
on the free market.12 As can be seen from Table 1.1, over the period 1989q1–2021q1 the free
market rate has depreciated around 17.4 per cent per annum as compared to the average
annual rate of inflation of around 18.3 per cent over the same period, representing a gap
of around 1 per cent between the two series. But according to the Purchasing Power Parity
10See Panel B of Figure A.4 in the Appendix.
11See Panel C of Figure A.4 in the Appendix. The development of the free market exchange rate, also known as the ‘black’
market rate during the 1979-1988 period, is discussed in Pesaran (1992).
12The time profiles of free market rate and consumer prices (in log scales) are depicted in Panel D of Figure A.4 in the
Appendix. As can be seen there is a very close association between the two series.
10

Table 1.1: Free market and official foreign exchange rate depreciation, inflation, real output
growth, and sanctions intensity over the period 1979q3–2021q1
Per cent per annum
Periods Free FX depreciation
Official FX
depreciation
Inflation Output
growth
Sanctions
intensity (0,1)
Mean Median
Revolution and
Iran-Iraq War1
19.94 0.28 18.29 -1.60 0.20 0.11
(1979q3–1989q2)
Rafsanjani presidency 16.55 39.83 21.17 5.16 0.11 0.10
(1989q3–1997q2)
Khatami presidency 7.90 20.34 14.53 4.72 0.15 0.13
(1997q3–2005q2)
Ahmadinejad
presidency
17.08 5.16 18.15 1.68 0.38 0.39
(2005q3–2013q2)
Rouhani presidency2 25.34 14.66 19.61 0.61 0.34 0.27
(2013q3–2021q2)
Post-revolution full
sample2
17.39 15.30 18.34 1.98 0.23 0.15
(1979q3–2021q1)
Post-War full sample2 17.38 19.88 18.30 3.08 0.24 0.16
(1989q1–2021q1)
Notes: 1. Data on free market foreign exchange rate start in 1980q2. 2. Data on foreign exchange rates (free market
and official rate), and inflation end in 2021q1, data on output growth end in 2020q1, data on sanctions intensity end
in 2020q3. See Section 1.3 of the paper for the sanctions intensity variable definition over the range (0,1). See Sections
A.2.1, A.2.2, A.2.5, and A.2.6 in the data appendix for details on the construction of the sanctions intensity variable,
calendar conversions, and sources of the data used.
(PPP), the difference between inflation and exchange rate depreciation should match the
average annual U.S. inflation, which is estimated to be around 2.5 per cent over the same
period.13
It is also important to note that not all foreign exchange crises can be traced to the intensification of sanctions. Iran has witnessed major currency crises during all the four
presidencies since 1989, whilst only the last two currency crises can be directly attributed
to intensification of the sanctions. The currency crises during Rafsanjani and Khatami
13The PPP is a long-run relationship that relates the exchange rate between two currencies to their relative price of goods:
Pt = EtP
∗
t
, with Et being the exchange rate representing the number of domestic currency units that can be bought
with one unit of foreign currency, Pt and P
∗
t denote the domestic and foreign price levels.
11

presidencies have domestic roots resulting from the rapid expansion of imports and low
oil prices, coupled with accommodating fiscal and monetary policies.14 As shown in Table 1.1, the average rate of inflation has been systematically high across all the four presidencies, and does not seem to correlate with changes in sanctions intensity. Even under
Khatami’s Presidency, the average annual inflation still amounted to 14.5 per cent, despite
his conciliatory foreign policy and a much lower rate of currency depreciation (7.9 per
cent as compared to 17.4 per cent over the full sample).
Comparing Iran’s output growth with that of world output growth over the 1989–2019
period15 also suggests an output growth shortfall of around 1 per cent per annum, which
could be contributed to the sanctions, although such a comparison does not take account
of Iran’s potential as an emerging economy. Even if we exclude the war periods, we also
observe a much larger output growth volatility in Iran as compared to the volatility of
world output growth volatility or a number of emerging economies of similar size to Iran,
such as Turkey or Saudi Arabia. Iran’s output growth volatility (as measured by standard
deviations of output growth) was almost five times as large as the global output growth
volatility (12.61 versus 2.69 per cent).16 Comparing Iran and Turkey over the same period,
we also find that Turkey grew at an average annual rate of 4 per cent with volatility of 10.8
per cent, a country also known for high inflation and repeated currency crises.17 Finally,
sanctions have most likely also contributed to the de-coupling of the Iranian economy
from the rest of the world. Again comparing Iran and Turkey, we note that the correlation
of Iran’s output growth with the world output growth is around 0.12, barely statistically
14During the reconstruction period under President Rafsanjani imports of goods and services doubled over the period
1989–1991 rising from 13.5 to 25 billion dollars, and Iran’s foreign debt rose to 23.2 billion dollars by the end of 1993.
For further details of the developments that led to the currency crisis under President Rafsanjani, see Section 3 of
Pesaran (2000).
15World output is computed as a weighted average of some of the largest 33 economies with details provided in the
Appendix.
16Mohaddes and Pesaran (2013) document the high volatility of Iran’s oil export revenues as one of the factors behind
Iran’s low growth and high volatility. A large part of the volatility of Iran’s oil export revenues is traced to high
volatility of barrels of oil exported, largely due to vagaries of sanctions. By comparison the volatility of oil prices is
of secondary importance. This contrasts to the volatility of Saudi Arabia oil revenues which is largely governed by
changes in international oil prices.
17The average annual output growth of Saudi Arabia over the 2005-2019 period was also similar to Turkey and amounted
to 4.3 per cent.
12

significant, as compared with a correlation of 0.33 for Turkey.
There seems to be little doubt that sanctions have adversely affected the Iranian economy, contributing to low growth, high inflation and increased volatility. What is less clear
is how to carry out a quantitative evaluation and identification of channels through which
sanctions have affected the Iranian economy over time, in particular the dynamics of responses and the time horizon over which the effects of sanctions filter out throughout the
economy. To this end, a formal model is required where conditions under which the effects of sanctions can be identified are made explicit, and their dynamic implications are
estimated and evaluated. It is to this task that we now turn in the rest of this paper.
1.3 Measures of sanctions intensity
Sanctions against Iran have been imposed with different degrees of intensity over the
past forty years. To account for both the prolonged nature of sanctions and their timevarying intensity, we construct "sanctions on" and "sanctions off" measures based on newspaper coverage of the imposition and the occasional lifting of sanctions. Newspaper coverage has been used in the literature and was initially formalized by Baker, Bloom, and
Davis (2016) for measuring the effects of economic uncertainty on macroeconomic aggregates. But, to our knowledge, the idea of using newspaper coverage to quantify sanctions
intensity is new.
We consider six of the world’s major daily newspapers, namely the New York Times,
the Washington Post, the Los Angeles Times and the Wall Street Journal in the U.S., and
the Guardian and the Financial Times in the U.K.. We then count the number of articles
published in these newspapers that deal with sanctions and Iran.18 We are careful not to
confound our measures with articles that cover international sanctions against Iraq but
also mention Iran. Sources and details of how the searches were carried out are provided
18The selected newspapers represent a sample of the most-read and well-informed articles over the past forty years, and
provide a good blend of both generalist press and those that focus on economic-finance issues. Also, by including two
different geographic regions, we hope to cover a more diversified sample.
13

in Section A.2.1 of Appendix A.
We also considered including Iranian newspaper sources in our textual analysis, but
decided not to do so for three main reasons. First, newspaper articles written about sanctions in Iran have a political dimension (e.g. strengthen the theocracy by levering on the
idea of the "resistance economy"), which does not necessarily relate to changes in sanctions
intensity. Second, sanctions are announced, decided, and implemented by the U.S. and
other major U.N. countries. Therefore, Western media offer a more accurate and timely
changes in new and ongoing sanctions against Iran. Third, there are not many Iranian
newspapers that reliably cover the whole forty years time period that we are considering,
and including available data from Iranian newspapers could bias our sanctions indicator.
One can think of our approach as measuring a proxy for an underlying latent sanctions
intensity process. The true process generates a signal, part of which is captured in daily
articles published in the six newspapers under consideration. To be specific, let njdt be
the number of articles published about sanctions on Iran in newspaper j during day d of
month t, and denote the true (latent) sanctions intensity variable during month t by s
∗
dt.
The relationship between njdt and s
∗
dt is specified as:
njdt = θjs
∗
dt + ζjdt, (1.1)
where θj > 0 is the loading of newspaper j on the true signal, s
∗
dt, and ζjdt is an idiosyncratic serially uncorrelated error term assumed to be distributed independently of the true
signal, s
∗
dt, with zero mean and finite variance. Equation (1.1) could be viewed as a single
factor model where θj
is the newspaper-specific factor loading. The number of articles
published in newspaper j correlates with the true signal depending on the size of θj and
the variance of the idiosyncratic term. Clearly, not all published articles capture the true
signals, but by averaging across newspapers and different days in a given month it is possible to reduce the effects of the noise, ζjdt, and obtain a consistent estimator of s
∗
dt, up to a
14

scalar constant. Both simple and weighted averages can be used. Taking a simple average
across the J newspapers and the number of days, Dt
, in month t, we have nt = θJ s
∗
t + ζt
,
where nt = J
−1D
−1
t
PJ
j=1
PDt
d=1 njdt, s
∗
t = D
−1
t
PDt
d=1 s
∗
dt, and θJ = J
−1 PJ
j=1 θj
. We considered 6 newspapers (J = 6) over a number of publishing days per month Dt
, typically
26 days, resulting in about 156 data points over which to average. This in turn ensures
that the idiosyncratic errors get diversified, and as a result the average error, ζt
, becomes
reasonably small. Specifically
ζt = J
−1D
−1
t
X
J
j=1
X
Dt
d=1
ζjdt = Op

J
−1D
−1
t

,
and we have s
∗
t = ¯θ
−1
J nt + op(1).These monthly measures can then be time aggregated
further to obtain quarterly or annual series which are then used to identify the effects of
s
∗
t
(up to the scaling factor ¯θ
−1
J
) in our macro-econometric model. We could also consider
a weighted average version of n¯t along the lines suggested in the literature, where the
number of newspaper articles (the raw count) is weighted by the inverse of their respective standard deviations, σˆjT , computed over the full data set, using σˆjT =
q
(T − 1)−1×
qPT
t=1 (njt − nj )
2
, njt = D
−1
t
PDt
d=1 njdt, and nj = T
−1 PT
t=1 njt. See Baker, Bloom, and
Davis (2016) and Plante (2019). But, as reported in Figure A.1 of the Appendix, the simple and weighted averages, after being suitably scaled, are very close in the case of our
application.
Although most sanctions news has been about imposing new or tightening old sanctions, there are some isolated periods where sanctions have been lifted, as in 1981 after the
release of the U.S. hostages, and over the period 2016q1–2018q2 after the implementation
of JCPOA. Accordingly, we construct two sanctions measures: an ‘on’ measure, denoted
by st,on, and an ‘off’ measure, denoted by st,of f , and we normalize them such that they
both lie between 0 and 1, with 1 representing the maximum sanctions intensity over the
full sample. We then define a net sanctions measure as st = st,on − w × st,of f , where w > 0
1

represents the weight attached to the sanctions off indicator compared to the sanctions
on indicator. The weight, w, is estimated to be wb = 0.4 using a grid search method over
values of w ∈ (0, 1).
19
Figure 1.1 displays the quarterly estimates of st over the period from 1989q1 to 2020q3,
which takes its maximum value at the end of 2011 when Iran was sanctioned simultaneously by the U.N., the U.S. and the E.U.. Important historical events are annotated in the
lower part of the figure, while specifics of particular sanctions are shown on the upper
part of the figure.
The fact that intensity of sanctions against Iran has been quite varied can be clearly seen
from Figure 1.1. Most notably there are three major spikes in sanctions intensity. The first
is in 2006 after Ahmadinejad was elected and Iran began its uranium enrichment program,
when the U.S. passed the "Iran Freedom and Support Act", which extended the coercive
measures against Iran – most notably secondary sanctions on non-U.S. corporations and
institutions doing business with Iran and very strict sanctions related to investments in
the energy sector. An even larger spike occurs between 2011 and 2012, when the Obama
administration joined efforts with the United Nations and the European Union to tighten
the sanctions even further with the aim of bringing Iran to negotiations over the nuclear
program. The U.S. passed stiff measures at the end of December 2011 under the "National
Defense and Authorization Act for Fiscal Year 2012", with Iran threatening to block oil
shipments through the Strait of Hormuz as a response. At the same time, the E.U. initiated
a total disconnect of Iranian financial institutions from the international payments system
(SWIFT) in March 2012,20 while the U.N. proceeded to extend the mandates of their previous resolutions between June 2011 and June 2012. The third, and most recent, spike is registered in 2018q2 after Trump decided to unilaterally withdraw the U.S. from the JCPOA
19The grid search was performed by running the regressions: ∆yt = β0+β1∆yt−1+β2st−1(w)+εt, w ∈ {0.1, 0.2, ..., 0.9}
over the period 1989q1–2019q4, and w selected by the maximum likelihood method. Further details are provided in
Section A.2 of the Appendix.
20SWIFT stands for the "Society for Worldwide Interbank Financial Telecommunications", and it is a vast and secure
network used by banks and other financial institutions to operate financial transactions across the globe.
16

Figure 1.1: Sanctions intensity variable over the period 1989q1–2020q3
Notes: Major events related to the Middle East are indicated by arrows below the x-axis. Major sanctions episodes related to Iran are indicated by arrows above the
x-axis. See Sections A.2.1 and A.2.2 in the data appendix for details on the construction of the sanctions intensity variable.
17

accord and begin a strategy of "maximum pressure". There are also minor spikes in 1996
when the Clinton administration signed the "Iran and Libya Sanctions Act", and in 1997
when the U.S. introduced an export ban to reduce the threat of potential weapons of mass
destruction being built, and in 2010 when the CISADA ("Comprehensive Iran Sanctions
Accountability and Divestment Act") was signed into law and the U.N. Security Council
passed the fourth round of sanctions against Iran with its 1929 resolution.
Lows of the sanctions intensity variable are recorded during the reconstruction period
under President Rafsanjani and the pragmatic rule under Khatami’s administration, and
more recently over the period between the JCPOA agreement in August 2015 and January
2018, when the U.S. unilaterally withdrew from the agreement. Table 1.2 provides summary statistics (minimum, median, mean, maximum and standard deviations) of st over a
number of sub-periods. A number of interesting observations follow from this table. First,
the summary statistics for st over the low sanctions periods under Rafsanjani and Khatami
are very close to those recorded for the period 2015q1–2018q1 when sanctions were partially lifted after JCPOA. Second, the peak of sanctions occurred during the internationally
coordinated efforts of 2011/2012 rather than after 2018, when the U.S. began their "maximum pressure" strategy under Trump and Bolton.21 In the period after 2018q2, the degree
of intensity of our indicator is 82 per cent of its peak in 2011. However, after 2018 the intensity of sanctions against Iran seems to have been much more persistent: the mean and
median are higher during the 2018q2–2020q3 period than during 2012q1–2014q4. Finally,
we notice that after the Iran-Iraq War, the median of the sanctions intensity has been only
two thirds of the mean: 0.16 versus 0.24. This feature stems from the several tail events
that characterize the series of sanctions against Iran, and as an overall measure the median
is to be preferred to the mean.
For the analysis of the effects of sanctions on Iran, it is also important to note that st
21John Bolton served as the 26th United States National Security Advisor from April 2018 to September 2019 under the
Presidency of Donald Trump. He has been a long-standing "policy hawk" advocating for regime change in several
strategic countries not aligned with the U.S. such as Iran and North Korea, among others.
18

shows a considerable degree of persistence over time. When sanctions are intensified they
tend to remain high for some time before subsiding. Table A.6 in the Appendix provides
estimates of first- and second-order autoregressive processes (AR) fitted to st
, and shows
that an AR(1) model captures well the sanctions intensity process, with a relatively large
and highly significant AR coefficient, namely 0.743 (0.059).
Table 1.2: Descriptive statistics of the sanctions intensity variable over relevant time periods
Time period Min Median Mean Max St. Dev.
Rafsanjani & Khatami presidencies 1989q3–2005q2 0.02 0.12 0.13 0.36 0.07
Ahmadinejad presidency 2005q3–2013q2 0.11 0.39 0.38 1.0 0.17
U.N./U.S. max sanctions 2012q1–2014q4 0.27 0.45 0.48 1.0 0.18
JCPOA agreement 2015q1–2018q1 0.06 0.11 0.14 0.33 0.07
U.S. "maximum pressure" 2018q2–2020q3 0.21 0.63 0.56 0.82 0.21
Full sample (post Iran-Iraq War) 1989q1–2020q3 0.02 0.16 0.24 1.0 0.19
Notes: See Sections A.2.1 and A.2.2 in the data appendix for details on the construction of the sanctions intensity variable.
Finally, as a robustness check we also attempted to create an alternative measure of
sanctions intensity based on the number of Iranian entities being sanctioned by the U.S..
We used the U.S. Treasury data set on entries and exits of sanctioned companies, individuals and vessels. We were able to build an indicator from 2006 to present. Although the
two measures (newspaper coverage and U.S. Treasury data) capture the sanctions phenomenon from different perspectives, they correlate rather well at 38 per cent. For further
details see Section A.2.3 of the Appendix.
1.4 Identification of sanctions effects: methodological issues
Identifying the effects of sanctions on the Iranian economy is challenging even if a reliable measure of sanctions intensity is available. As with all macro policy interventions,
when identifying the effects of sanctions we also need to take account of confounding fac19

tors that are correlated with changes in sanctions intensity, and which at the same time
have a causal influence on target variable(s) of interest such as output growth and inflation. In situations where a policy intervention has differential effects over time and across
many different units such as households or firms, difference-in-difference techniques are
used whereby changes in outcomes during policy on and policy off periods for those affected by the intervention are compared to corresponding changes in outcomes for a control group that is not directly affected by the intervention. This method is clearly not applicable to the analysis of policy interventions that target a particular entity such as a region
or country, and a different approach is needed. Currently, there are two such approaches:
the Synthetic Control Method (SCM) advanced by Abadie and Gardeazabal (2003) and
the Panel Data Approach (PDA) proposed by Hsiao, Ching, and Wan (2012).22 Both approaches compare outcomes for the country (region) subject to the intervention with a
weighted average of outcomes from a control group. The former was originally applied
to quantify the economic costs of political instability in the Basque Country in Spain, and
the latter to evaluate the economic effects of the hand-over of Hong Kong to China in 1997.
Both studies consider discrete policy interventions and do not allow for the policy intensity to vary over time. Perhaps most importantly, they both use pre-policy outcomes to
estimate the weights applied to the countries included in the control group. The main
difference between the two approaches lies in way the weights are estimated.23
The application of these approaches to the case of Iran is complicated by the fact that
the imposition of sanctions coincided with the onset of the Revolution, which renders the
pre-sanctions period of limited relevance. Also, as noted earlier, the scope and intensity
of sanctions against Iran have undergone considerable changes over the past forty years
and there are no clear cut periods that one could identify as "sanctions on" periods to be
compared to "sanctions off" periods, in which all sanctions were levied. There is also the
22Further details and extensions of SCM are discussed in Abadie, Diamond, and Hainmueller (2010) and Doudchenko
and Imbens (2016).
23Gardeazabal and Vega-Bayo (2017) provide a comparative simulation analysis of SCM and PDA, with a follow up
critique by Wan, Xie, and Hsiao (2018).
20

additional challenge of identifying countries for inclusion in the control group.
To our knowledge, the only study that applies SCM to Iran is by Gharehgozli (2017),
who considers the effects of the intensification of sanctions just before the JCPOA agreement in July 2015 on Iran’s real GDP, treating the years 2011–2014 as the "sanctions on"
period as compared to the preceding years 1995–2010 as the "sanctions off" period. She
then selects 13 countries worldwide to mimic a "synthetic" Iran as a weighted average of
GDP of these economies with their respective weights determined using the SCM based
on seven different macroeconomic indicators. She concludes that the 2011–2014 sanctions
resulted in Iran’s real GDP to fall by as much as 17 per cent, as compared to the synthetic
sanctions free Iran, with all the output short fall attributed to sanctions.
We depart from the mainstream literature reviewed above and consider the following
reduced-form model for Iran’s quarterly output growth
∆yt = α + λ∆yt−1 + ψ0st + ψ1st−1 + β
0xt + γ
0
ft + ut
, (1.2)
where ∆yt
is the output growth, st measures the intensity of sanctions against Iran, xt
and ft are respectively observed and unobserved control variables, and ut
is an idiosyncratic error term, distributed independently of (st
, xt
,ft). It is assumed that part of the
change in the intensity of sanctions affects Iran’s output growth with a lag, thus distinguishing between short term, ψ0, and long term, θ = (ψ0 +ψ1)/(1−λ), effects of sanctions.
As discussed above, sanctions affect output growth through a number of channels, most
importantly oil exports, exchange rate, liquidity, and inflation to be addressed in Section
1.5. However, here we are concerned with both direct and indirect effects of sanctions on
output growth, and to avoid confounding these effects we will not be including contemporaneous values domestic variables in the output growth equation. For example, including
changes in volume of oil exports in (1.2) will most likely result in under-estimating the
effects of the sanctions, since one important aim of the sanctions is to reduce Iran’s oil
exports. The same also applies to other domestic variables, such as exchange rate or infla21

tion, that are affected by sanctions and their inclusion bias the estimates of ψ0 and ψ1. But
it is important that observed and unobserved external factors that are not affected by sanctions, but potentially can impact Iran’s output growth are included in (1.2). One important
example is changes in international oil prices, which affect Iran’s output growth through
changes in government foreign exchange revenues, but do not seem to have been affected
by sanctions, particularly due the accommodating oil production and export policies followed by Saudi Arabia.24 Accordingly, we include changes in international oil prices
as an element of xt
. We could not identify other observed external factors with obvious
effects on the Iranian economy, and focused on identification of unobserved common factors, ft
. In this regard, our approach is closely related to the PDA (Hsiao, Ching, and Wan
(2012)). To this end, we consider the following equations for output growth for the rest
of the world25
∆yit = αiy + β
0
iyxit + γ
0
iyft + uy,it, i = 1, 2, ..., n, (1.3)
where ∆yit denotes output growth in country i (excluding Iran), xit is a k × 1 vector
of control variables specific to country i, and ft
is the m × 1 vector of unobserved common
factors, and uy,it are idiosyncratic shocks to output growth that are serially uncorrelated
but could be weakly cross correlated.26 By allowing the factor loadings, γi
, to differ across
countries, we do not assume that all economies are equally affected by the same factors,
an assumption that underlies the DiD approach. We also depart from SCM and PDA
and, unlike these approaches, we do not require a "donor pool" of countries to be selected
for comparative analysis. Instead, we assume that xit also follows similar multi-factor
structures, and impose a rank condition which allows us to identify ft as weighted averages
24See Section 5.2 in Mohaddes and Pesaran (2016), where it is shown that an adverse shock to Iran’s oil supply induces
a rise in Saudi oil supplies. Another reason why sanctions against Iran have not led to important oil price rises is the
prolonged nature of these sanctions, allowing the international oil market to adjust to reduced oil exports from Iran.
25It is assumed that sanctions against Iran have had only negligible impacts on the rest of the world economies.
26A set of random variables, {uit, i = 1, 2, ..., n} is said to be weakly cross correlated if supj
Pn
i=1 |Cov(uit, ujt)| <
C < ∞. It then follows that Pn
i=1 wiuit = Op(n
−1/2
), for any granular weights, wi, such that wi = O(n
−1
P
) and
n
i=1 w
2
i = O(n
−1
). An obvious example is the simple weights wi = 1/n. For further details see Chudik, Pesaran,
and Tosetti (2011).
22

of ∆yit and xit overi (excluding Iran). Any granular weights can be used to construct these
averages, such as simple averages. But in cases where n is not sufficiently large and there
are dominant economies such as the U.S., it is advisable to use output shares as weights.
Accordingly, suppose that
xit = αix + Γ
0
ixft + ux,it, i = 1, 2, ..., n, (1.4)
where Γix is a k × m matrix of factor loadings, and ux,it is a k × 1 vector that follows
general stationary processes that are weakly cross-sectionally correlated. Combining (1.3)
and (1.4) we have:


1 −β
0
iy
0 Ik


zit =


αiy
αix

 +


γ
0
iy
Γ
0
ix


ft +


uy,it
ux,it

 ,
which yields zit = ci + Aift + Biuit, where
ci =


αiy + β
0
iyαix
αix

 , Ai =


γ
0
iy + β
0
iyΓ
0
ix
Γ
0
ix


ft
, and Bi =


1 β
0
iy
0 Ik

 .
Averaging zit overi using the weightswi we now have zwt = cw+Awft+
Pn
i=1 wiBiuit,where
zwt =
Pn
i=1 wizit, cw =
Pn
i=1 wici
, and Aw =
Pn
i=1 wiAi
. Suppose now that the (k + 1) × m
matrix Aw is full column rank (that requires m ≤ k + 1), and A
0
wAw →p> 0, as n → ∞.
Then ft can be solved as27
ft = awf + Hwzwt − Hw
Xn
i=1
wiBiuit!
,
where
27See Pesaran (2006) for further details in a related context.
23

awf = −

A
0
wAw
−1
A
0
wcw and Hw =

A
0
wAw
−1
A
0
w.
Under the rank condition, the terms awf and Hw tend to finite limits, whilst under the
assumptions that uit are weakly cross correlated, the final term of ft tends to zero for any
choice of weights wi that are granular, ft can be identified up to linear transformations in
terms of zwt = (∆ywt, x
0
wt)
0 = (Pn
i=1 wi∆yit,
Pn
i=1 wix
0
it)
0
. More specifically, Pn
i=1 wiBiuit =
Op(n
−1/2
), and we have ft = awf + Hwzwt + Op(n
−1/2
), which can be used to eliminate the
unobserved factors, ft
, from Iran’s output growth equation. Specifically, we obtain
∆yt = αyw + λ∆yt−1 + ψ0st + ψ1st−1 + β
0xt + θyw∆ywt + θ
0
xwxwt + ut + Op(n
−1/2
). (1.5)
Hence, for n sufficiently large, and considering that the Iranian economy is quite small
relative to the rest of the world, the sanctions coefficients ψ0, and ψ1 can be identified
by augmenting the output growth equations with the rest of the world average output
growth, ∆ywt, and the weighted averages of the observed drivers of the rest of the world
output growth, xwt.
It is interesting to note that our approach does not require selecting a pool of countries
that are close to Iran, but recommends including all countries, weighted for their relative
importance in the world economy. Selecting specific countries could bias the results by
restricting the number included in the construction of cross section averages. The rank
condition, rank
A
0
wAw

= m , for a given n, and as n → ∞, ensures that ft has a reasonably pervasive effect on most economies which in turn allows us to use ∆ywt, and xwt as
reliable proxies for ft
.
The analysis of sanctions effects can also be extended to other macro variables such as
inflation and unemployment, and even to some key socioeconomic indicators such as life
expectancy, death rate or educational achievement. See Section 7 in Laudati and Pesaran
(2021).
24

1.4.1 Estimates of sanctions-induced output losses
Initially, we report regression results for the reduced form output growth regressions
set out in Equation (1.5), and focus on specifications with st−1 as the intervention variable.
We favor this specification over the one that includes both st and st−1 since "sanctions
news" do not contain anticipatory effects, and one would not expect contemporaneous
changes in st to affect output growth, as time is required for the real economy to adjust to
sanctions news.28 The estimates of the reduced form output growth equations computed
over the period 1989q1–2019q4 are summarized in Table 1.3, where we report both the
short- and long-run effects of sanctions on output growth, whilst allowing for a host of
lagged values of domestic variables as well as contemporaneous foreign control variables
and international oil price returns.29 The parameter of interest is the long run effect of sanctions on output growth reported at the bottom panel of Table 1.3. It is estimated to be about
−0.027 (0.013), which is statistically significant and remarkably robust across the seven different specifications reported. The estimates suggest output growth losses of around 2 per
cent per annum if we use the median value of st over the sample under consideration, or 3
per cent if we use the mean value of st
.
30 Due to the large outliers in the sanctions intensity
variable, we favor the lower estimate of 2 per cent based on the median value of st
, which
in turn suggests that in the absence of sanctions and sanctions-induced mismanagement
Iran’s average annual growth over 1989q1–2019q4 could have been around 4-5 per cent, as
compared to the 3 per cent realized, a counterfactual outcome which is close to the growth
of emerging economies such as Indonesia, South Korea, Thailand, and Turkey whose average annual growth rate over the same sample period amounted to 4.8, 4.5, 4.2 and 4.0
28We are grateful to Nick Bloom for drawing our attention to this point.
29Amongst the domestic variables, only lagged inflation has a statistically significant impact on output growth. The
negative effect of inflation on output growth could be due to price distortions and allocation inefficiencies that are
often associated with high and persistent levels of inflation, as has been the case in Iran. We find that global factors
such as global volatility or output growth do not affect Iran’s output growth, largely due to Iran’s relative economic
isolation. Amongst the global factors, the only factor with statistically significant impact on Iran’s output growth
turned out to be the global exchange rate variable. However, the negative effect of the global exchange rate variable
on output growth is more difficult to rationalize.
30The median and mean values of st, are 0.16 and 0.24, respectively, as summarized in Table 1.2.
25

Table 1.3: Estimates of the reduced form Iran’s output growth equation estimated over the period
1989q1–2019q4
∆yt
(1) (2) (3) (4) (5) (6) (7)
st−1(βst−1
) −0.033∗∗ −0.032∗∗ −0.032∗∗ −0.034∗∗ −0.034∗∗ −0.034∗∗ −0.035∗∗
(0.016) (0.016) (0.016) (0.016) (0.016) (0.016) (0.016)
∆yt−1(λ∆yt−1
) −0.204∗∗ −0.202∗∗ −0.203∗∗ −0.200∗∗ −0.214∗∗ −0.214∗∗ −0.218∗∗
(0.091) (0.092) (0.092) (0.092) (0.091) (0.092) (0.092)
∆x
0
t−1
0.016 0.016 0.016 0.017 0.014 0.014 0.015
(0.014) (0.014) (0.014) (0.014) (0.014) (0.014) (0.014)
∆ef,t−1 −0.004 −0.004 −0.004 0.0002 0.004 0.004 0.002
(0.033) (0.033) (0.034) (0.034) (0.033) (0.034) (0.034)
∆mt−1 −0.028 −0.037 −0.041 −0.032 −0.053 −0.056 −0.063
(0.100) (0.102) (0.104) (0.104) (0.103) (0.104) (0.106)
∆pt−1 −0.239∗ −0.234∗ −0.232∗ −0.246∗∗ −0.268∗∗ −0.273∗∗ −0.274∗∗
(0.122) (0.123) (0.123) (0.124) (0.123) (0.125) (0.125)
∆ywt 0.228 0.160 0.215 −0.129 −0.162 −0.117
(0.553) (0.602) (0.604) (0.625) (0.635) (0.643)
∆reqwt 0.013 0.021 0.013 0.002 −0.0001
(0.045) (0.046) (0.045) (0.057) (0.057)
∆rwt −4.518 −4.311 −4.474 −3.490
(4.141) (4.097) (4.143) (4.611)
∆ewt −0.278∗ −0.272∗ −0.309∗
(0.148) (0.150) (0.168)
grvt −0.038 −0.044
(0.114) (0.115)
∆p
0
t −0.012
(0.024)
βst−1
/(1 − λ∆yt−1
) −0.027∗∗ −0.027∗∗ −0.027∗∗ −0.028∗∗ −0.028∗∗ −0.028∗∗ −0.028∗∗
(0.013) (0.013) (0.013) (0.013) (0.013) (0.013) (0.013)
Adjusted R2 0.083 0.077 0.069 0.071 0.091 0.084 0.077
Notes: ∆yt = ln(Yt/Yt−1), Yt is the quarterly real output of Iran. st is the quarterly sanctions intensity variable. βst−1
and λ∆yt−1
are the coefficients of st−1 and ∆yt−1, respectively; βst−1 /(1 − λ∆yt−1
) represents the long run effect
of sanctions on output growth. See Chapter 6 of Pesaran (2015). ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the oil exports
revenues in U.S. dollars; ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly rial/U.S. dollar free market exchange rate; ∆mt =
(M2t−M2,t−1)/M2,t−1, M2t is the monetary aggregate M2 obtained by summing the aggregates M1 and "quasi-money";
∆pt = ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran. ∆ywt is the quarterly world output growth,
computed as ywt =
Pn
i=1 wiyit, with {yit}
n
i=1 being the natural log of real output for 33 major economies, and {wi}
n
i=1
are GDP-PPP weights. ∆reqwt is the quarterly rate of change of the global real equity price index: reqwt =
Pn
i=1 wireqit,
reqit is the natural log of the real equity price of country i in quarter t. ∆rwt is the quarterly per cent change of the global
nominal long term interest rate: rwt =
Pn
i=1 wirit, rit is the long term nominal interest rate of country i in quarter t.
∆ewt is the quarterly rate of change of the global real exchange rate vis-à-vis the U.S. dollar: ewt =
Pn
i=1 wieit, eit
is the natural log of the real exchange rate of country i in quarter t. grvt is the quarterly global realized volatility.
∆p
0
t = ln(P
0
t /P0
t−1), P0
t
is the quarterly oil price (Brent crude). Numbers in parentheses are least squares standard
errors. ***p < 0.01, **p < 0.05, *p < 0.1.
See Sections A.2.1, A.2.2, A.2.5, and A.2.6 in the data appendix for details on the construction of the sanctions intensity
variable, calendar conversions, and sources of the data used.
26

per cent, respectively. Similar estimates are obtained if both st and st−1 are included in
the regressions. See Table A.9 of the Appendix. Furthermore, Tables A.41 and A.42 of the
Appendix show that similar results are obtained when we use heteroskedastic-consistent
standard errors following the approach proposed by White (1980).
1.5 Sanctions-augmented structural VAR model for Iran
We now consider the main channels through which sanctions affect the Iranian economy, and provide estimates of the time profiles of their effects. Initially, U.S. sanctions
targeted the Iranian oil industry with the aim of reducing oil exports and limiting Iran’s capacity to produce oil. More recently, financial sanctions have been used more extensively.
As a result new sanctions, or even their announcement, have invariably led to reduced oil
exports, with a significant depreciation of the Iranian rial, followed by a sharp rise in price
inflation and output losses within 3–6 months after the imposition of the new sanctions.
We model the dynamic inter-relationships of oil exports, exchange rate, money supply, inflation and output growth using a structural vector autoregressive (SVAR for short) model
augmented with the sanctions intensity variable and the global control variables, denoted
by zwt above.
We denote by qt = (∆x
0
t
, ∆ef t, ∆mt
, ∆pt
, ∆yt)
0
an m×1 (with m = 5) vector of endogenous domestic variables, where ∆x
0
t
is the oil export revenues, ∆ef t represents the rate of
change of free market foreign exchange rate,31 ∆mt
is the growth rate of money supply,
∆pt
is the rate of inflation, and ∆yt
is real output growth.
To distinguish between different types of shocks and their implications for the Iranian
economy, in our SVAR we assume the direction of causality goes from ∆x
0
t
to exchange
rate depreciation, to money supply growth, to inflation, and then to output growth, as
represented by the ordering of the five endogenous variables in qt
. Under this causal
31We also tried a weighted average of the free market and official exchange rates, but found that the free market rate
provides a more accurate and timely measure of the exchange rate movements for Iran given its higher responsiveness
to sanctions. The exchange rate variable is expressed as the number of Iranian rials per one U.S. dollar.
27

ordering, we are able to distinguish changes in qt that are due to variations in the intensity of sanctions from those that are the result of domestic policy shocks.32 The assumed
causal ordering can be justified in terms of relative speed with which the Iranian economy
responds to crises. Oil exports are usually targeted by sanctions and their revenues fall immediately by design, then it is the value of the rial in free market that weakens, followed
by a potential expansion of liquidity, a rise in the price of imported commodities, before
the real economy starts to adjust to higher prices and interest rates. Due to the relatively
underdeveloped nature of money and capital markets in Iran, monetary policy tends to
be accommodating, typically allowing liquidity to rise in line with inflation.
We consider the following augmented SVAR model in qt
A0qt = aq + A1qt−1 + A2qt−2 + γ0s
st + γ1s
st−1 + Dw zwt + εt
, (1.6)
where, as before, st
is our measure of sanctions intensity, and zwt =
(∆ywt, ∆reqwt, ∆rwt, grvt
, ∆ewt)
0
is a k × 1 (k = 5) vector of control variables that
includes: global output growth, ∆ywt, global equity returns, ∆reqwt, global long term
interest rates, ∆rwt, global realized volatility, grvt
, and the rate of change of the global
real exchange rate, ∆ewt.
33 Given the assumed causal ordering, matrix A0 is restricted to
have the following lower triangular form:
32It is also possible to use non-recursive identification schemes such as sign restrictions, or the more recently developed
Bayesian approach by Baumeister and Hamilton (2015) to point identify and estimate contemporaneous effects in the
SVAR model and associated impulse responses using priors. This could be the subject of future research. However,
we do not expect that the main results of our paper that relate to the effects of sanctions to be much affected by such
alternative identification schemes.
33Details on data sources and the computation of the global variables are given in Section A.2 of the Appendix.
28

A0 =


1 0 . . . 0
−a
0
∆e,∆x0
.
.
.
.
.
.
.
.
.
−a
0
∆m,∆x0 −a
0
∆m,∆e
−a
0
∆p,∆x0 −a
0
∆p,∆e −a
0
∆p,∆m 1 0
−a
0
∆y,∆x0 −a
0
∆y,∆e −a
0
∆y,∆m −a
0
∆y,∆p
1


, (1.7)
where we expect a
0
∆p,∆e ≥ 0, with inflation responding positively to a contemporaneous rise in ef t (rial depreciation), and a
0
∆y,∆x0 ≥ 0, with output rising as a
result of higher oil revenues. The signs of the contemporaneous impacts of ∆ef t,
∆mt and ∆pt on output growth are less clear cut. The structural shocks, εt =
(ε∆x0,t, ε∆e,t, ε∆m,t, ε∆p,t, ε∆y,t)
0
, are assumed to be serially uncorrelated with zero means,
E(εt) = 0, and mutually uncorrelated with the diagonal covariance matrix E(εtε
0
t
) = Σ =
Diag (σ∆x0,∆x0
, σ∆e,∆e, σ∆m,∆m, σ∆p,∆p, σ∆y,∆y). Since we condition on sanctions intensity
and global indicators, the structural shocks can be viewed as "domestic" shocks attributed
to policy changes that are unrelated to sanctions. Specifically, it is assumed that εt are
uncorrelated with st and zwt. Under these assumptions it is now possible to distinguish
between the effects of a unit change in the sanctions variable, from domestic policy changes
initiated by a unit standard error change to the domestic shocks, εt
. Specifically, for contemporaneous effects we have ∂qt/∂st = A−1
0 γ0s
, and ∂qt/∂εjt =
√σjjA−1
0 ej where A0 is
given by (1.7), ej (j = ∆x
0
, ∆ef , ∆m, ∆p, ∆y) are the vectors of zeros except for their j-th
component, which is one.
For the purpose of computing impulse responses and forecast error variance decompositions, we model st and zwt as autoregressive processes:
st = as + ρsst−1 + ηt
, (1.8)
zwt = azw + Azwzw,t−1 + vwt, (1.9)
29

where the sanctions and global shocks, ηt and vwt, are serially uncorrelated with zero
means, and variances ω
2
s and Ωw. Combining equations (1.6), (1.8), and (1.9), we obtain
the following SVAR model in zt = (q
0
t
, st
, z
0
wt)
0
,
Ψ0zt = a+Ψ1zt−1 + Ψ2zt−2 + ut
, (1.10)
where a =

a
0
q
, as, a
0
zw0
and ut = (ε
0
t
, ηt
, v
0
wt)
0
, are (m + k + 1) × 1 vectors, and
Ψ0 =


A0 −γ0s −Dw
0 1 0
0 0 Ik


, Ψ1 =


A1 γ1s 0
0 ρs 0
0 0 Azw


, Ψ2 =


A2 0 0
0 0 0
0 0 0


,
are (m+k+1)×(m+k+1) matrices. Standard techniques can now be applied to the SVAR
model in (1.10) to obtain impulse response functions and error variance decompositions
assuming the global shocks, vwt, are uncorrelated with domestic and sanctions shocks
(namely εt
, and ηt).34 This is a standard small open economy assumption which applies
to the Iranian economy in particular since its relative size in the world economy is small
and has been declining over the past forty years.
1.5.1 Structural model estimation
We estimated the five equations of the augmented SVAR model in (1.6), experimenting with different sub-sets of the control variables: world output growth, global realized
volatility, world real equity returns, changes in long term interest rates, and global real exchange rate changes against the U.S. dollar.35 The full set of results are provided in Tables
A.10 to A.14 of the Appendix.36 As can be seen, with the exception of the world output
34Further details are provided in Section A.3 of the Appendix.
35To take account of possible seasonal variations all regressions are also augmented with seasonal dummies which turn
out to be highly significant in the money supply growth equation.
36The figures in parentheses in these tables report the least squares standard errors. But to check the robustness of our
inference we also provide White (1980)’s heteroskedastic robust standard errors in Section A.4.8 of the Appendix.
As to be expected the use of robust standard errors results in reduced level of statistical significance for most of the
3

growth, none of the other control variables play a significant role in the regressions for
inflation and output growth. Accordingly, we consider a simplified specification and in
Table 1.4 we provide estimates of the SVAR model including only the world output growth
(∆ywt) as the control variable. As can be seen from this Table, the sanction variable is statistically significant for four out of the five domestic variables, with changes in oil exports
and output growth being affected after one quarter. In contrast, exchange rate changes and
inflation are affected significantly by the sanctions contemporaneously as well as with one
quarter lag. The only variable which seems to be unaffected by the sanctions is the money
supply growth. It is also worth noting that none of regressions in the SVAR model display
residual serial correlation, which is an important consideration for impulse response and
variance decomposition analyses that follow.
To assess the quantitative importance of the sanctions, we compute the effects of sanctions by multiplying the estimated coefficients of st and st−1 by the median value of the
sanctions variable, which is around 0.16. See Table 1.2. The median presents a more robust measure of a central value for sanctions intensity as compared to the average which
is likely to be sensitive to the outlier values of st over time. Using the median we are able
to provide an estimate of the effects of moving from a no sanction case (with st = 0) to a
situation where st
is set to its median value. We refer to these estimates as (counterfactual)
median estimates of the sanctions.
We now consider the results of the individual equations in the SVAR model. In the case
of oil exports, we note that in addition to sanctions, changes in oil exports are also affected
significantly by world output growth with some feedback effects from the exchange rate
variable. The positive impact of world output growth on Iran’s oil exports makes sense and
suggests that sanctions have not been completely effective in making Iran’s oil exports nonresponsive to world economic conditions. The median estimate of the effects of sanctions
parameters, but as can be seen the differences are largely inconsequential. Further, we shall be using bootstrap standard
error bands in our impulse response analyses and the bootstrap procedure will automatically account for possible
heteroskedasticity and non-Gaussian errors.
31

on oil exports is around 4.6 per cent per quarter, or about 18.4 per year. See Table A.10.
Turning to the estimates of the exchange rate equation (given in column 2 of Table 1.4),
we first note that exchange rate changes have been moderately persistent with a coefficient
of 0.350 (0.094), which is statistically highly significant. In most developed markets, we do
not expect exchange rate changes to be persistent, and the result for rial points to possible
inefficiencies in Iran’s foreign exchange market.37 Second, we observe that the rial depreciates strongly in the same quarter in which sanctions are raised. The median fall in its
value is around 4.9 per cent per quarter. However, there is a significant degree of overshooting, with the sanctions variable having the opposite effect on exchange rate after one
quarter. The rial appreciates by about 3.7 per cent in the following quarter, resulting in a
less pronounced overall impact of sanctions on the rial depreciation of around 1.2 per cent
per quarter, or 4.8 per cent per annum, which is still quite substantial.38 As can be seen
from Table A.11 of the Appendix, these estimates are remarkably stable and statistically
significant at the 1 per cent level across all specifications regardless of the number of global
control variables included in the regression equation. In fact, none of the domestic variables (oil exports, inflation, money supply growth, and output growth) have a statistically
significant effect on the exchange rate, and only global realized volatility and foreign output growth prove to be statistically significant at 10 per cent level but not robust across all
specifications. The adjusted R2 of the exchange rate equations with world output growth
included is around 21 per cent. This is high by the standard of exchange rate equations,
and is partly explained by the presence of the contemporaneous sanctions variable in the
regression. Its use for prediction requires predicting the sanctions variable which adds
another layer of uncertainty.
The estimates for the money supply growth (∆mt) equation are summarized in column 3 of Table 1.4. As can be seen, only lagged money supply growth is statistically sig37Note that the exchange rate is expressed as the number of Iranian rials per one U.S. dollar, and therefore a rise in the
exchange rate variable corresponds to a depreciation of the rial.
38Such overshooting is well-known in the international finance literature. See, for example, Dornbusch (1976).
32

nificant, and moderately persistent with a coefficient of 0.218 (0.096). Notably, we do not
find any feedback effects from inflation to money supply growth, even when we include
a second lag of inflation to the money supply growth equation.
The estimates for inflation (∆pt) are summarized in column 4 of Table 1.4. As discussed
in Section 1.2, inflation in Iran has been persistently high over the past forty years, and
to capture its persistence it proved necessary to include ∆pt−2, as well as ∆pt−1 in the
regression equation. It does not seem necessary to include second order lags of other
variables in the inflation equation.39 Perhaps not surprisingly, the estimates also show
that exchange rate depreciation is an important determinant of inflation in Iran, a factor
which is statistically significant and quantitatively important. The immediate effect of one
per cent depreciation of the free market exchange rate is to raise prices by around 0.15 to
0.17 per cent, as many imported goods items tend to rise with the fall in exchange rate.
Sanctions affect inflation indirectly through the exchange rate as well as directly, but the
direct effects of sanctions do not last and the net direct effects of sanctions on inflation seem
to be negligible. It is also interesting and quite surprising that money supply growth,
oil exports, or lagged output growth do not seem to have any significant direct effects
on inflation. But we do find some evidence of global output growth positively affecting
inflation, a kind of international Phillips curve effect that leads to higher international
prices that are in turn reflected in Iran’s import prices and hence domestic inflation.
Finally, column 5 of Table 1.4 provides the results for real output growth. Output
growth in Iran is negatively autocorrelated, with a coefficient estimated to be around
−0.22 which is statistically significant. This contrasts the positively autocorrelated output growth observed for many other countries. The sanctions intensity variable affects
output growth with a lag, as it takes a few months for different sectors of the economy
to adjust to sanctions. After only one quarter, the effect of sanctions on output growth is
39See also Table A.13 of the Appendix where different sub-sets of control variables are also included in the regressions
for the inflation equation.
33

statistically highly significant.40 Within two quarters the regression predicts Iran’s output
growth to slow down by about 0.9 per cent per quarter (3.6 per cent per annum). In addition to this direct effect, sanctions also influence output growth through exchange rate
depreciation, which is also highly statistically significant. This indirect effect amounts to
around 0.14 per cent per quarter drop in output growth when the rial depreciates by one
per cent. Output growth is also negatively affected by lower oil exports, and by lagged
inflation, which highlights the adverse effects of high and persistent inflation on output
growth, without any short term Phillips curve type of trade off between inflation and output growth. Interestingly enough, none of the global factors seem to have any significant
effects on Iran’s output growth, partly due to Iran’s relative economic and financial isolation from the rest of the global economy. See Table A.14 of the Appendix for further
details.
Since – in the SVAR model – money supply growth plays a minimal role in the determination of inflation and exchange rate variations, and exchange rate remains the primary
driver of inflation and output growth, we decided to simplify the model by dropping the
money supply growth from the SVAR model. The estimation results for this simplified
model are summarized in Tables A.15 to A.19, and A.20 to A.24 of the Appendix.41 As
can be seen, the estimates for the four equations in the current SVAR model are very close
to those in the model with money supply growth, confirming further that money supply
growth is not essential for the analysis of the interrelationships of exchange rate, inflation
and output growth in Iran, which is the primary concern of our analysis. It is also worth
noting that our main findings are not much affected by re-ordering of the domestic variables. In Section A.4 of the Appendix we provide results of estimating the SVAR model in
(1.6), with the following ordering of the domestic variables: {∆ef t, ∆x
0
t
, ∆mt
, ∆pt
, ∆yt}.
For this ordering, the foreign exchange is placed first and oil export revenues second to
40Table A.14 in the Appendix shows that the results are reasonably robust to different choices of control variables.
41Dropping the money supply growth from the SVAR model, also renders the seasonal dummies statistically insignificant. Thus seasonal dummies are not included in the SVAR model that excludes the money supply growth variable.
34

Table 1.4: Quarterly estimates of the SVAR model of Iran with domestic variables ordered as: oil
exports, exchange rate returns, money supply growth, inflation, and output growth, estimated
over the period 1989q1-2019q4
∆x
0
t ∆ef,t ∆mt ∆pt ∆yt
(1) (2) (3) (4) (5)
st 0.107 0.305∗∗∗ −0.002 −0.033∗∗ 0.029
(0.150) (0.064) (0.017) (0.013) (0.026)
st−1 −0.288∗ −0.233∗∗∗ 0.015 0.037∗∗∗ −0.056∗∗
(0.155) (0.067) (0.017) (0.013) (0.026)
∆x
0
t 0.029 0.006 −0.003 0.025∗
(0.040) (0.010) (0.007) (0.014)
∆ef,t −0.007 0.163∗∗∗ −0.141∗∗∗
(0.023) (0.017) (0.045)
∆mt −0.073 0.063
(0.073) (0.142)
∆pt 0.387∗∗
(0.181)
∆ywt 8.406∗∗ −2.639∗ 0.233 0.865∗∗∗ −0.520
(3.649) (1.590) (0.389) (0.298) (0.592)
∆x
0
t−1 −0.051 0.044 −0.005 −0.003 0.023∗
(0.090) (0.038) (0.009) (0.007) (0.014)
∆ef,t−1 −0.441∗∗ 0.350∗∗∗ −0.025 −0.009 0.041
(0.217) (0.094) (0.027) (0.020) (0.036)
∆mt−1 −0.715 0.149 0.218∗∗ −0.025 0.046
(0.930) (0.397) (0.096) (0.075) (0.144)
∆pt−1 0.052 −0.341 0.167 0.488∗∗∗ −0.505∗∗∗
(0.794) (0.338) (0.115) (0.089) (0.167)
∆yt−1 0.122 −0.145 0.025 0.042 −0.221∗∗
(0.592) (0.252) (0.063) (0.048) (0.090)
∆pt−2 −0.070 0.183∗∗
(0.104) (0.079)
Residual serial 2.406 6.212 7.640 8.061 7.240
correlation test [0.662] [0.184] [0.106] [0.089] [0.124]
Adjusted R2 0.122 0.209 0.466 0.659 0.124
Notes: The variables are ordered as: ∆x
0
t
, ∆eft, ∆mt, ∆pt, and ∆yt, where: ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the
oil exports revenues in U.S. dollars; ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly rial/U.S. dollar free market exchange
rate; ∆mt = (M2t − M2,t−1)/M2,t−1, M2t is the monetary aggregate M2 obtained by summing the aggregates M1
and "quasi-money"; ∆pt = ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran; ∆yt = ln(Yt/Yt−1), Yt is
the quarterly real output of Iran. st is the quarterly sanctions intensity variable. ∆ywt is the quarterly world output
growth, computed as ywt =
Pn
i=1 wiyit, with {yit}
n
i=1 being the natural log of real output for 33 major economies, and
{wi}
n
i=1 are GDP-PPP weights. Seasonal dummies are included to allow for possible seasonality of the variables in the
regressions of the SVAR model in Equation (1.6) with qt =

∆x
0
t
, ∆eft, ∆mt, ∆pt, ∆yt
0
and zwt = (∆ywt)
0
. Numbers
in parentheses are least squares standard errors, and those in square brackets are p-values. ***p < 0.01, **p < 0.05,
*p < 0.1. "Residual serial correlation test" is the Breusch–Godfrey LM test of serially uncorrelated errors with lag order
of the test set to 4. See Sections A.2.1, A.2.2, A.2.5, and A.2.6 in the data appendix for details on the construction of the
sanctions intensity variable, calendar conversions, and sources of the data used. Regressions results that include other
global control variables (e.g. global realized volatility) are provided in Tables A.10 –A.14 in the Appendix.
3

capture the idea that the rial may react even faster than ∆x
0
t
to announcements of new
sanctions. The results are summarized in Tables A.25 to A.30, and – as aforementioned –
they are largely not affected by this change. In Tables A.31 to A.35, we consider the effects of re-ordering of the variables in the case of the simplified model without the money
supply growth or seasonal dummies, and it is once again confirmed that the results are
reasonably robust to the re-ordering of the variables under consideration.
Overall sanctions have affected Iran in a number of ways and through different direct
and indirect channels, the most important of which are falls in oil export revenues and the
exchange rate depreciation. The exchange rate depreciation itself could have its roots in
persistently high levels of inflation, coupled with a reduction in oil revenues and anticipated decline in private sector activity. The currency depreciation in turn leads to higher
import prices and lower economic growth. We also find that the direct effect of sanctions
on inflation is rather small, compared to an average annual inflation norm of around 18
per cent in Iran (See Table 1.1).
Money supply growth seems to follow patterns which are neither related to sanctions
nor to any of the domestic variables, notably inflation, which could be due to the underdevelopment of capital and money markets in Iran, as highlighted recently by Mazarei
(2019). These results seem quite robust to other measures of liquidity such as M1 or private sector credit.42
1.5.2 Impulse response analysis
The estimates of the individual equations provided in Table 1.4 provide a snap-shot of
how sanctions interact with some of the key macroeconomic variables. However, given
the dynamic and simultaneous nature of the model, to fully understand and evaluate
the nature and consequences of these interactions, we compute impulse response functions (IRFs) and forecast error variance decompositions (FEVDs) for the augmented SVAR
42Estimates based on these alternative measures of liquidity are available upon request.
36

model given by (1.6).
43 We have seen that money supply growth does not play much of a
role in the determination of inflation and output growth, and is hardly affected by sanctions. Also, amongst the control variables, only foreign output growth seems to exert statistically significant effects on inflation and output growth. For these reasons, to compute
IRFs and FEVDs we will be focusing on the SVAR model with qt = (∆x
0
t
, ∆ef t, ∆pt
, ∆yt)
0
,
augmented with the sanction variables and ∆ywt as the foreign control variable. We also
use AR(1) models for st and ∆ywt to capture the dynamics of these exogenous processes.44
The IRFs for positive one standard error (s.e.) shocks to st and qt are displayed in
Figure 1.2.
45 Panel A of this figure shows the results for the sanction shock. One standard
deviation for st
is equal to 0.120, which represents half of the average sanctions intensity
over the period considered (s1989q1−2019q4 = 0.24).46 A single quarter shock to sanctions
intensity causes oil exports to decrease by almost 5 per cent after one quarter, with some
reversal thereafter. But the negative effects of sanctions on oil export revenues continue
to be important even after four quarters with losses that are still about 1 per cent. The
positive shock to sanctions also causes the foreign exchange rate to depreciate by about 3
per cent in the same quarter, but its effects are rather short lived and become statistically
insignificant two quarters after the shock. For inflation and output growth, the effects of
the sanction shock last much longer. Its effects on inflation are particularly persistent and
last at least for four years after the shock, although its magnitude is relatively small: 0.3 per
cent increase per quarter in the first year. The effects of sanction shock on output growth,
on the other hand, are much larger in size. A single period one standard error shock to
sanctions causes output growth to fall by more than 0.4 per cent per quarter (1.6 per cent
43Detail of the derivations of IRFs and FEVDs are given in Sections A.3.1 and A.3.2 of the Appendix, respectively.
44Time series evidence in support of our choice of AR(1) specifications for st and ∆ywt are provided in Tables A.6 and
A.7 of the Appendix. It is also worth noting that the assumed AR(1) processes for st and ∆ywt only affect the IRFs
and FEVDs, and do not affect the estimates of the SVAR model.
45The error bands for the point estimates shown in these figures are computed using the bootstrap procedure described
in Sub-section A.3.4 of the Appendix.
46See Table 1.2 for the descriptive statistics of the sanctions intensity indicator, and note that one s.e. sanction shock is
computed using the AR(1) specification assumed for st – it is smaller than the one standard deviation of st. Information on the size of one standard error shock in the case of the endogenous variables in the SVAR model are provided
in Table A.8 of the Appendix.
37

Figure 1.2: Impulse responses of the effects of sanctions and domestic shocks on oil exports,
foreign exchange, inflation, and output growth
Panel A: One positive standard error shock to the sanctions intensity variable
Panel B: One positive standard error shock to the oil exports
Panel C: One positive standard error shock to the exchange rate
per annum). The loss in output growth is still close to 0.2 per cent per quarter two years
after the shock.
Panel B of Figure 1.2 displays the results for a single quarter shock to oil revenues.
The effect on oil export revenues themselves is very large and positive, although rather
short-lived, reflecting the rather volatile nature of oil export revenues. The effects of oil
revenue shock on foreign exchange rate is not that large, around 1.2 after one quarter, and
then falling to zero after four quarters. Its effects on inflation is positive but quite small,
38

around 0.2 per cent after two quarters. The positive shock to oil revenues induces a rise in
output of around 0.5 per cent on impact which is statistically significant, but this effect is
short lived and tends to zero quite rapidly.
The results for the foreign exchange rate shock are given in Panel C of Figure 1.2. The
effect of this shock on oil export revenues is negative and amounts to − 4 per cent one
quarter after the shock before reverting to zero thereafter. More interestingly, one quarter
exchange rate shock induces a sizeable and precisely estimated effect (of around 8 per
cent per quarter) on exchange rate, but similar to the effects of the sanction shock, it does
not last long and its effects dissipate very quickly after two quarters. The exchange rate
shock raises inflation on impact by around 1.2 per cent per quarter, and then starts to fall
and vanishes completely after about two years. The same is not true of real output growth.
The direct effects of foreign exchange shock on output growth are negative and statistically
significant but small in magnitude, around −0.50 per cent on impact, which then moves
towards zero very quickly.
Panel D of Figure 1.2 gives the results for an inflation shock (for example, due to a
domestic expansionary policy). Again, because of the highly persistent nature of inflation
in Iran, the most pronounced effects of the inflation shock is on inflation itself, raising
inflation by 1.5 per cent per quarter on impact and then falling gradually to zero after
two years. Interestingly, the effect of inflation shock on exchange rate is not statistically
significant, suggesting that the causal link between them is from exchange rate to inflation
and not vice versa. Compare the IRFs for exchange rate and inflation in Panels C and D of
Figure 1.2. The effects of inflation shock on output growth are positive on impact but
small in magnitude, and reverse quickly after one quarter, suggesting that it might not be
possible to increase output by expansionary policies. The effects on oil export revenues
do not appear to be statistically significant.
Finally, the IRFs of the effects of a positive shock to output growth are displayed in
Panel E of Figure 1.2. A positive output shock could be due to technological advance or
39

Figure 1.2: Impulse responses of the effects of sanctions and domestic shocks on oil exports,
foreign exchange, inflation, and output growth
Panel D: One positive standard error shock to inflation
Panel E: One positive standard error shock to Iran output growth
fundamental reforms that reduce economic distortions and enhance production opportunities. The output shock seems to have little impact (in short or medium term) on both
oil exports and exchange rate, which seem to be primarily driven by sanctions and their
own dynamics. The positive output shock also has a minimal effect on inflation, increasing inflation by less than 0.1 per cent per quarter after two quarters. The primary effects
of the output shock are on output itself, raising output by 2.8 per cent per quarter on impact before losing momentum in less than a year. The initial very large increase in output
is somewhat of an over-reaction which is then corrected slightly, yet providing a net 2
per cent rise in output within the year of the shock. Once again this result highlights the
importance of supply side policies for improving Iran’s output growth in the long run.47
The impulse response analysis confirms some of the preliminary conclusions set out
47In the Appendix, we provide impulse responses for a positive shock to the world output growth in Figure A.5.
40

in Section 1.5.1. Sanctions have their most impact on oil exports, free market exchange
rate, and to a lesser extent on output growth. Inflation has its own dynamics and is hardly
affected by sanctions. The roots of high and persistent inflation must be found in domestic
economic mismanagement. Also, sanctions do adversely affect output growth after one
quarter but such effects are short lived.
1.5.3 Forecast error variance decompositions
We now turn to a quantification of the relative importance of sanctions as compared to
the four domestic shocks and the foreign output shock. Table 1.5 presents the results.48 In
Panel A we report estimates of the FEVDs of a unit shock to oil export revenues. As can be
seen, around 96 per cent of the forecast error variance of oil export revenues is explained
by the shock to oil revenues itself. Other factors come into play in subsequent quarters,
but they explain only a small proportion of the total forecast error variance, with sanctions
explaining 6 per cent, foreign exchange 2 per cent, and world output growth around 4 per
cent. It is clear that a single isolated sanction shock is not enough to make a significant
impact on oil export revenues, and a prolonged period of sanctions is required for sanction
effects to cumulate and lead to a sizable effect.
Panel B of Table 1.5 gives the results for the foreign exchange variable. Not surprisingly,
foreign exchange shocks are the most important, and account for 82 per cent of forecast
error variance on impact and decline only slightly, falling to 80 per cent after one quarter.
Sanctions shock accounts for 17 per cent of the variance, with the other shocks contributing
very little. Therefore, isolated sanctions do not drive Iran’s exchange rate, and only become
a dominant force if we consider prolonged periods over which sanction shocks are in place
with the same intensity.
The FEVDs of inflation, reported in Panel C of Table 1.5, show that foreign exchange and
inflation shocks account for the bulk of the variance, with sanction shocks accounting for
48FEVDs are computed using Equations (A.9), (A.10), and (A.11).
41

Table 1.5: Forecast error variance decomposition for domestic variables in the SVAR model with
a single shock to sanctions
Panel A: FEVD for oil exports Panel B: FEVD for exchange rate
Quarter Proportion explained by a shock to: Quarter Proportion explained by a shock to:
ahead st ∆x
0
t ∆ef t ∆pt ∆yt ∆ywt ahead st ∆x
0
t ∆ef t ∆pt ∆yt ∆ywt
0 0.00 0.96 0.00 0.00 0.00 0.03 0 0.17 0.00 0.82 0.00 0.00 0.01
1 0.04 0.90 0.02 0.00 0.00 0.04 1 0.17 0.01 0.80 0.00 0.00 0.02
2 0.05 0.89 0.02 0.00 0.00 0.04 2 0.17 0.01 0.80 0.00 0.00 0.02
3 0.06 0.88 0.02 0.00 0.00 0.04 3 0.17 0.01 0.80 0.01 0.00 0.02
4 0.06 0.88 0.02 0.00 0.00 0.04 4 0.17 0.01 0.80 0.01 0.00 0.02
5 0.06 0.88 0.02 0.00 0.00 0.04 5 0.17 0.01 0.80 0.01 0.00 0.02
6 0.06 0.88 0.02 0.00 0.00 0.04 6 0.17 0.01 0.80 0.01 0.00 0.02
7 0.06 0.88 0.02 0.00 0.00 0.04 7 0.17 0.01 0.80 0.01 0.00 0.02
8 0.06 0.88 0.02 0.00 0.00 0.04 8 0.17 0.01 0.80 0.01 0.00 0.02
Panel C: FEVD for inflation Panel D: FEVD for output growth
Quarter Proportion explained by a shock to: Quarter Proportion explained by a shock to:
ahead st ∆x
0
t ∆ef t ∆pt ∆yt ∆ywt ahead st ∆x
0
t ∆ef t ∆pt ∆yt ∆ywt
0 0.01 0.00 0.43 0.55 0.00 0.01 0 0.00 0.03 0.05 0.03 0.90 0.00
1 0.04 0.00 0.48 0.48 0.00 0.01 1 0.02 0.03 0.05 0.06 0.85 0.00
2 0.05 0.00 0.49 0.45 0.00 0.01 2 0.03 0.03 0.05 0.06 0.83 0.00
3 0.06 0.00 0.50 0.43 0.00 0.01 3 0.04 0.03 0.05 0.06 0.83 0.00
4 0.06 0.00 0.50 0.43 0.00 0.01 4 0.04 0.03 0.05 0.06 0.82 0.00
5 0.06 0.00 0.50 0.43 0.00 0.01 5 0.04 0.03 0.05 0.06 0.82 0.00
6 0.07 0.00 0.50 0.42 0.00 0.01 6 0.05 0.03 0.05 0.06 0.82 0.00
7 0.07 0.00 0.50 0.42 0.00 0.01 7 0.05 0.03 0.05 0.06 0.82 0.00
8 0.07 0.00 0.50 0.42 0.00 0.01 8 0.05 0.03 0.05 0.06 0.82 0.00
Notes: st is the quarterly sanctions intensity variable. ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the oil exports revenues in U.S.
dollars. ∆eft = ln(Eft/Ef,t−1), Eft is the Iran rial/U.S. dollar quarterly free market exchange rate. ∆pt = ln(Pt/Pt−1),
Pt is the quarterly consumer price index of Iran. ∆yt = ln(Yt/Yt−1), Yt is the quarterly real output of Iran. ∆ywt is
the quarterly world output growth: ywt =
Pn
i=1 wiyit, with {yit}
n
i=1 being the natural log of real output for 33 major
economies, and wi the GDP-PPP weights.
See Sections A.2.1, A.2.2, A.2.5, and A.2.6 in the data appendix for details on the construction of the sanctions intensity
variable, calendar conversions, and sources of the data used.
the remainder. Oil exports, domestic and foreign output shocks make little contribution.
On impact, inflation shock accounts for 55 per cent of the variance, flattening out at 42 per
cent after six quarters. In contrast, the contribution of the foreign exchange shock rises
from 43 per cent on impact to 50 per cent after three quarters. The contribution of the
sanction shock is not particularly large, and starts at 1 per cent, but rises to 7 per cent after
six quarters. Once again, we see that inflation and exchange rates in Iran are mainly driven
42

Figure 1.3: Forecast error variance decomposition for domestic variables in the SVAR model
with a cumulative shock to sanctions, and domestic variables ordered as oil exports, exchange
rate returns, inflation, and output growth
by domestic factors. But sanctions effects could accumulate very quickly if we consider
sanctions being in place over a prolonged period of time.
Finally, the FEVDs of output growth are reported in Panel D of Table 1.5. As can be
seen, the output shock is by far the most important shock and accounts for 90 per cent
of forecast error variance of output growth on impact and falls only slightly to 82 per
cent after four quarters. In line with our estimates, sanctions shocks do not affect output
growth on impact, and end up explaining only 5 per cent of the variance after six quarters.
Foreign output shocks do not have any explanatory power for Iran’s output growth. The
other three domestic shocks (oil exports, inflation and exchange rate) together account
for 14 per cent of forecast error variance of output growth after one quarter, and do not
increase any further after that.
The outcome of FEVDs is very different if we consider the effects of a prolonged period
of sanctions, namely if sanctions are imposed for over 2 or more years. The results are
summarized in Figure 1.3. When sanctions are imposed with the same intensity for about
two years, sanctions explain more than 70 per cent of the forecast error variance of inflation
and around 60 per cent of the forecast error variance of output growth, keeping all other
shocks fixed.
43

1.6 Concluding remarks
In this paper, using a novel measure of the intensity of sanctions based on newspaper
coverage, we have quantified the effects of sanctions on oil exports, exchange rate, inflation, and output growth in Iran. In order to estimate the prolonged effect of sanctions on
the Iranian economy, we faced several measurement and econometric challenges. Iran’s
recent history formed by the Islamic Revolution, hostage taking and the eight year war
with Iraq, makes it hard to have a reliable "donor pool" of countries to construct a synthetic Iran. Furthermore, Dif-in-Dif methods cannot be applied because a relevant presanctions episode is not available. Finally, the degree of intensity of sanctions imposed on
Iran has varied considerably over time while never being completely lifted. For these reasons, a novel identification strategy was provided to overcome the difficulties that could
not be addressed by using approaches such as the Synthetic Control Method and the Panel
Data Approach (Hsiao, Ching, and Wan (2012)). In addition, we have proposed the first
newspaper-based indicator to track sanctions intensity. In doing so, it was possible to
solve the issue of not having a "sanction off" period, something impossible to capture with
a dummy variable estimator. With a novel econometric strategy and a sanctions index at
hand, we proceeded to analyze both the reduced-form long term effects of sanctions on
Iranian output, and the channels through which such losses manifested.
When evaluating the direct and indirect costs of sanctions, we have followed the literature and attempted to control for possible confounders, namely external and domestic
factors that affect the economy but are unrelated to sanctions, such as advances in technology, world output growth, and international prices. Using a reduced form regression
of output growth on our sanctions intensity variable, we estimate Iran’s output loss to be
around 2 per cent per annum, which is considerable when cumulated over time. There
is, of course, a high degree of uncertainty associated with such estimates which should
be borne in mind. But – even if we compare Iran’s growth performance over the 1989q1-
44

2021q1 period with that of Turkey and other similar size emerging economies – we find
that Iran’s realized output growth of 3 percent still lies below the average growth of 4.4 per
cent experienced by Indonesia, Turkey, South Korea and Thailand over the same period.49
A SVAR analysis augmented with the proposed sanctions variable as well as global
factors, allows us to identify the channels of transmission of sanctions to the broader economy. Oil exports revenues drop first as a direct consequence of new sanctions, accompanied by an instantaneous depreciation of the Iranian rial vis-à-vis the U.S. dollar, which is
subsequently translated into higher consumer prices, and slower economic growth. Monetary policy appeared to be passive, and accommodating the behavior of other macrofinancial variables once we control for a number of factors. Overall, the economy appeared
rather isolated from global factors.
There is no doubt that sanctions have harmed the Iranian economy, but one should
not underestimate the damage done by years of economic mismanagement. Iran’s low
output growth relative to its potential, high inflation and excess output growth volatility
cannot all be traced to sanctions and have domestic roots stemming from prolonged periods of economic mismanagement, distorted relative prices, rent seeking, a weak banking
system and under-developed financial institutions. Sanctions have accentuated some of
these trends and delayed the implementation of highly needed reforms.
A more comprehensive analysis of sanctions also requires detailed investigation into
how sanctions and their variability over the past forty years have affected policy decisions
at all levels, from monetary and fiscal policies to industrial, regional and social policies. It
is generally agreed that, at times of increased sanctions intensity, governments fearful of
political consequences are reluctant to curtail distortionary policies, such as large subsidies
on food and energy, and they might even accentuate them, or resort to multiple exchange
49If we take the 1990 value of GDP-PPP (constant international dollars) for Iran and cumulate the potential losses over
the period until 2019, we reach a similar conclusion. In the conservative scenario in which Iran grows at 4.5 per cent
per annum rather than 3.08, its output would be 18th in the world between Saudi Arabia and Thailand. By using a
less conservative yet still plausible estimate – if Iran were to grow at 5.5 per cent, its output would be double the level
experienced in 2019. It would be the 15th largest economy between South Korea and Spain – two developed countries
by now. We thank an anonymous referee for this idea.
45

rates to reduce the inflationary effects of sanctions.
Sanctions have also led to some positive unintended effects. Non-oil exports have risen
from $600 million before the Revolution to around $40 billion, resulting in greater foreign exchange diversification. The high-tech sector has seen exponential growth over the
past 10 years and is now one of the regions’ fastest growing sectors. Iran’s major webbased companies have been protected by potential competition from their U.S. counterparts shown in brackets including: Digikala (Amazon), Aparat (YouTube), Cafe Bazaar
(Google Play), Snapp (Uber), Divar (Craigslist). It is estimated that over 65 per cent of
Iranian households are now connected to the internet. This rapid expansion was facilitated by the government and security apparatus making affordable high-speed internet a
reality in Iran. The Mobile Telecommunication Company of Iran, largely controlled by the
Islamic Revolutionary Guard Corps now has over 43 million subscribers. Sanctions have
also resulted in significant advances in the areas of missiles and other military-related
technologies. It is estimated that IRGC control between 10-30 per cent of the economy,
with large stakes in the oil and gas sectors, construction, telecom, banking, and tourism.
One could argue that IRGC has been a major beneficiary of U.S. sanctions.
Our sample does not cover the period from January 2020 when Covid-19 effects started
to be felt in Iran. However, it is clear Covid-19 could have important medium term consequences, particularly for the traditional service sector. The full economic impact of Covid19 on the Iranian economy is unknown and requires further investigation.
46

Chapter 2
Inequality and the Rise of Finance
2.1 Introduction
The largest industry of the U.S. economy by value added is the one composed by Finance, Insurance, and Real Estate (FIRE). This was not always the case. After a period of
mild growth following WWII, the size of the financial sector strongly increased from the
early 1980s before reaching a plateau in the aftermath of the Great Financial Crisis of 2008.
A similar trend over time can be noticed both in the amount of financial assets intermediated relative to GDP, and in the wage premium earned by workers in the financial sector
vis-à-vis the ones employed elsewhere. Despite the economic significance of the rise of
finance, no consensus in the macroeconomics and finance literature has been reached on
the explanations of such growth (Philippon and Reshef, 2013; Philippon, 2015).
This paper highlights the importance of analyzing the rise of the financial sector in conjunction with the endogenous rise of other non-banking financial intermediaries. The rise
of finance is expressed in terms of assets intermediated by the financial sector as a share of
nominal output. However, such growth has not been characterized by a mere rise of the
same institutions and instruments that existed before the 1980s. Rather, it has been led
mostly by the explosive growth of other non-bank financial institutions, which often go
47

under the name of "shadow banking system". The shadow banking system is composed
of a network of institutions, operations, and instruments that replicate similar credit functions banks perform, but without relying on the traditional structure of depository chartered banks. In the theory I propose, the emergence of the shadow banking system occurs
endogenously when not enough risk-free financial assets are available for investors – what
goes under the name of "safe assets shortage" (Caballero, Farhi, and Gourinchas, 2017). In
this respect, the approach is consistent with the hypothesis advanced by part of the finance
literature (Gorton, 2017). However, I do not take the financial phenomenon in isolation,
and I treat the safe asset shortage as a proximate cause.
I seek a root cause explanation by looking at the structural transformations happening in the broader macro-economy. The decline of the labor share is taken as a primitive
change to connect factor income inequality with personal income inequality. A higher capital share (interpreted as non-labor share) generates higher income and wealth inequality
in the economy due to non-trivial dynamics. The wealthier quantiles of the population
need to solve a portfolio composition problem on how to allocate their savings between
risky and safe assets. When inequality increases non-homothetically, the amount of assets under management held by the richer quantiles increases, i.e., keeping the savings
rates constant, as a result of incomplete markets. If the public supply of safe assets is constrained, the higher demand for safe assets puts a downward pressure on the real interest
rates to clear the markets.
The lower the interest rates fall, the lower the costs of poorer households to issue debt
to finance part of their consumption. As a result of cheaper debt, households’ leverage increases. Such environment creates the conditions for other non-bank financial institutions
– the shadow banking system – to step in and complete a market by manufacturing private
safe assets. These are obtained by transforming the debt that the poorer households wish
to issue to finance part of their consumption into quasi-safe assets that investors wish to
hold to hedge their risks. The financial sector rises in size and changes in composition as
48

a result of the endogenous rise of shadow banks. Importantly, a higher amount of assets
under management leads also to higher price valuations, which endogenously create a
feedback mechanism that exacerbates wealth inequality.
The paper contributes to the literature along several dimensions. First, it pins down
the endogenous rise of the shadow banking system as emerging from structural macroeconomic forces. Second, it is able to link (for the first time to my knowledge) a change in
the production technology of the economy with a change of its banking structure. Third,
it allows to study in an internally-consistent framework both the direct channel of larger
inequality inducing a larger financial system, and the feedback loop of higher asset prices
valuation exacerbating the level of inequality. In short, the current parsimonious set-up
is able to connect disjoint parts of the macroeconomic and finance literature and predict:
The endogenous rise of income and wealth inequality, the compression of real interest
rates, the rise of households indebtedness, the higher leverage in the economy, the change
in size and composition of the financial sector, and the feedback loop between inequality and market-based finance in a parsimonious set-up.1 Finally, on the methodological
side – it is able to deliver the previous results by relying only on incomplete markets and
precautionary motives. Allowing for elements such as preferences for liquidity and/or
wealth and non-homothetic savings rates may further strengthen the overall quantitative
findings; however, these are not strictly necessary.
The paper is mostly quantitative in spirit, allowing for a model, its quantification, and
a series of policy experiments. The empirical section further tests the theoretical mechanisms and implications. Quantitatively, taking the periods 1970-1979 and 2010-2019 as
the initial and final steady states, the model is able to explain 73 per cent of the growth of
shadow banking, when wealth inequality increases by 20 per cent as a result of the corresponding capital share increase. In terms of real interest rates, the model can explain
1
See Mian and Sufi (2018) for a discussion on the importance of understanding whether the following patterns where
connected at all: (i) The rise of households indebtedness; (ii) The compression of the risk-free interest rate; and (iii)
The rise of finance itself (seen as amount of assets intermediated).
49

Figure 2.1: Gross and net measures of financial assets intermediated as a share of GDP and top
1 per cent income share in the United States over the long run
(a) Inequality and net measure of finance, 1913-2019 (b) Inequality and gross measure of finance, 1960-2019
Notes: In Panel (a) the net size of the financial sector up to 2010 refers to the domestically held claims adjusted to account
for informational quality produced by Philippon (2015). The series is spliced up to 2019 by accounting for the size of
the domestically held liabilities of the financial sector as a share of GDP. The top 1 percent income share includes capital
gains as developed by Piketty and Zucman (2014). In Panel (b), the fiscal income measure of inequality includes capital
gains and the adjustments made by by Auten and Splinter (2024). Sources: See Tables B.5 and B.6 in Section B.5 of the
Appendix for details on the variables sources and construction.
up to 40 per cent of such compression over time when the aforementioned increase in
inequality is obtained. The portfolio shares between risky and safe assets are not in line
with the values observed in the data even though they are not targeted and they are not
the main objective of study. The model also features a qualitative increase in the equity
risk premium in line with the data.
Historically speaking, when looking at the longue durée — the connection between inequality and finance seems to be systematically at play when financial markets are left
largely free of shackles. Figure 2.1a helps visualizing the relationship between inequality, measured by the fiscal income share of the top 1 per cent, and the domestic financial
system in the United States over the period 1913-2019.2 The two series strongly co-move
for almost a century, except for the thirty years from 1944 to 1973 following the Bretton
Woods Accords of July 1944. During the latter period, Western economies attempted to
immobilize financial markets in the hope that financial stability could be gained as a re2
Focusing on the fiscal income measure of inequality allows to account for capital gains – which are an essential element
for this paper. The top quantile is chosen because it is the only measure made available by Auten and Splinter (2024).
Data from the World Income Database suggest that a similar pattern has been followed also by the top 5 and 10 percent
of the distribution.
50

sult. In this respect, more than the limitations to speculative international capital per se
(coming in the form of "hot money"), the post-WWII world was characterized by outright
financial repression at the domestic level. In the rest of the paper, I will focus precisely on
the post-1970s world.3 Panel (b) shows that the relationship between gross total financial
sector assets and inequality from the 1960s is almost perfect. The two series correlate at 97
per cent and share the same hockey-stick pattern. As shown later, the tight relationship
between inequality and finance seems to be true more in general for a host of advanced
economies.
In the empirical section, I test the extent to which we can find statistically significant
evidence of the lockstep movements between inequality and finance over the past century.
In this respect, I build on the work by Müller and Watson (2018) on co-variability. I do
find evidence that in the pre- and post-Bretton Woods world the two series systematically
co-moved in growth rates. Such result can be thought to be in the same spirit as the "great
ratios" hypotheses related to the Kaldor facts.
Furthermore, I run more empirical analyses to test the importance of the "mechanism",
i.e., the role of market-based vs. bank-based financial structure, by analyzing a panel of
advanced economies. I follow the approach by Rancière, Tornell, and Westermann (2008),
and I find more evidence about the importance of considering a market-based financial
sector as a key aspect to think at the two-way relationship between finance and inequality.
Related literature. This work relates and tries to bridge different literature strands in
macroeconomics and finance. First, the rise of finance has been acknowledged to be an
important area of study after the Great Financial Crisis (GFC) through detailed data work
by Philippon and Reshef (2012, 2013), and Philippon (2015). Such effort naturally engendered a quest for theories to explain the trends. Gennaioli, Shleifer, and Vishny (2014)
provided an initial attempt by means of a neo-classical growth model augmented with
asset managers. However, this line of research overall did not gain sufficient traction to
3Chapter 3 explains how financial deregulation can be thought of as a natural element stemming from the present paper.
51

provide additional answers from a theoretical standpoint. Rather, in the post-GFC world
researchers have preferred to investigate the extent to which the increase in finance has
been "excessive" (Arcand, Berkes, and Panizza, 2015, among others).4 This work places
the stream of literature on the explanations for the secular rise of finance back to the fore,
and contributes to it by providing a hypothesis on its root causes. In this respect, it takes a
positive angle rather than a normative one by not addressing the issue about whether the
size of the current financial sector is excessive or not.
Second, the paper relates to the macroeconomic consequences of higher inequality.5
Kumhof, Rancière, and Winant (2015) established a link between increased inequality and
higher financial crises probabilities. Here, I do not look at the financial fragility component per se (even though the model does deliver higher leverage and lower capital buffers),
but rather at structural explanations for the secular rise of finance. More recently, inequality has been studied in conjunction with the rise of households indebtedness and richer
households savings in a series of papers by Mian, Straub, and Sufi (2021a,b,c). However,
this paper differs from the previous along several dimensions. First, it proposes a novel
connection between the production technological shifts and the rise of finance. Second, it
emphasizes that higher inequality leads to a larger financial sector because of its interactions with risks and a lack of public safe assets. Taking seriously the portfolio allocation
choices of the investors, higher inequality generates a domestically-driven safe asset shortage, which ultimately creates the conditions for the shadow banking system to emerge.
Third, as aforementioned, it looks at the change in finance size and the rise of other financial intermediaries as inextricably related. The growth of the sector becomes intertwined
4
See Cochrane (2013) for a rebuttal of this exercise. See Brunnermeier, Palia, Sastry, and Sims (2021) for a novel way of
empirically identifying the effects of financial deepening on output in the U.S..
5The debate on the causes leading to higher inequality is still open. Stansbury and Summers (2018) document an increasing gap between productivity and wages since at least 1973. Many other works have highlighted: the importance
of changes in the taxation regimes adopted (e.g. Piketty and Zucman, 2014), “China shocks” and off-shoring of jobs to
lower income countries (Autor, Dorn, and Hanson, 2016, for a review), the rise of automation and capital-enhancing
technologies (Acemoglu and Restrepo, 2022), the rise of college premium, and de-unionization. Such list is not meant
to be exhaustive and potentially a combination of all the previous items and others is important to explain the rise of
inequality. See Hubmer, Krusell, and Smith (2021) for a quantitative assessment of the different drivers explaining the
rise in wealth inequality.
52

with the rise of new financial intermediaries in response to the changes in the macroeconomic environment. Methodologically, I contribute to this latter literature by not relying
on non-homotheticities but on precautionary motives. I leave the savings rates constant
over time and across households to focus on the increase in the absolute amount of money
under management by investors, i.e. the change in the levels of savings.6
A host of other studies focus on the connection between inequality and other specific
macro-financial trends covered in this work. Favilukis (2013) and Auclert and Rognlie
(2017, 2020) focus on the macroeconomic implications of increased inequality connecting it with aggregate demand and decrease in real interest rates. A broader set of papers
looks at the effects of inequality for asset pricing. Lansing (2015), and Markiewicz and
Raciborski (2022) look at the implications of rising inequality (generated by higher capital income share) on lower interest rates and equity risk premium. Recently, Fagereng,
Gomez, Gouin-Bonenfant, Holm, Moll, and Natvik (2023) and Gomez (2023) look at the
aspects of inequality linked to asset valuations. Panageas (2020) carries out a literature
review and attempts to reconcile the most seminal studies on the matter. Also, the specific
effects of higher income inequality on households debts dynamics have been studied by Iacoviello (2008) and Coibion, Gorodnichenko, Kudlyak, and Mondragon (2020), amongst
others. Azzimonti, de Francisco, and Quadrini (2014) investigate the higher demand for
public debt stemming from higher inequality in an international setting.7
In this respect,
this paper has the benefit of encompassing different and relevant aspects in a single overarching framework.
This paper relates also to the broader safe assets shortage and "savings glut" literature.
The early literature on these topics is rather rich but almost exclusively focused on the
international dimension of the issue. More recently, Caballero and Farhi (2018) and Barro,
Fernández-Villaverde, Levintal, and Mollerus (2022) focus on the U.S., and look at the
macroeconomic implications that stem from acute safe assets shortages reaching some
6
See also Fagereng, Holm, Moll, and Natvik (2021) for an important study on wealth-related non-homotheticities.
7
See Benhabib and Bisin (2018) for review the literature on theories and empirics surrounding wealth inequality.
53

diverging conclusions on the importance of public safe assets to mitigate such issue. See
also Caballero, Farhi, and Gourinchas (2017) for a review. More similar to this paper,
Ordoñez and Piguillem (2021) also address the macroeconomic importance of the savings
glut from a domestic point of view by taking a demographic angle.
The recent macro-finance literature has also produced several works to place the safe
assets demand – and the insurance function of banking liabilities – at the center of stage
by looking at them as an extension of money instruments (Krishnamurthy and VissingJorgensen, 2012; Brunnermeier and Sannikov, 2016; Quadrini, 2017; Kiyotaki and Moore,
2019; Krishnamurthy and Li, 2023; among others).
Other aspects of the finance literature are also important to mention. On the one hand,
the paper speaks to the causes behind the rise of the asset management industry (Greenwood and Scharfstein, 2013) by linking it to inequality; on the other hand, the emergence
of the shadow banking system has been conjectured to be linked to the higher demand for
safe assets (Gorton, 2017), which is in line with what shown here. In a recent paper, Sarto
and Wang (2022) assess the connection between the rise of shadow banking and lower
interest rates, as theorized here.
The paper is structured as follows. Section 2.2 describes key stylized facts to place the
study in perspective. Section 2.3 describes the macro-finance model proposed. Section 2.4
explains the calibration exercise, and Section 2.5 assesses the quantitative performance of
the baseline scenario and a few policy experiments. Section 2.6 provides empirical evidence of the relations and channels between inequality and finance in the U.S. and across
countries. Section 2.7 concludes. Additional stylized facts, proofs, results, and data description are available in the Appendix.
54

2.2 Stylized facts and background
A number of stylized facts consistent with the proposed narrative can help to provide
descriptive evidence of the mechanism investigated.
The rise of finance from the 1980s has been characterized by the burgeoning rise of
"other non-bank financial institutions" and instruments. The literature has often called
this broad universe with the name of "shadow banking" or "parallel banking" system. Albeit being more lightly regulated than the traditional banking system, and not enjoying
an explicit safety net from the government, shadow banking should not be interpreted
as a set of operations catering to "shady" and illegal activities. Instead, it relates to the
creation of credit instruments by non-depository institutions (say, mortgages issued by
finance companies) ultimately funded with short-run money-like instruments such as repurchase agreements (RePos) and money markets mutual funds (MMMFs) shares which
differ from the publicly-guaranteed bank deposits. To this extent, I will sometimes prefer
to refer to the shadow banking system as the "market-based banking" system to reflect the
intrinsic banking nature of this network of transactions generated through market mechanisms.
In Panel (a) of Figure 2.2, I plot the size of traditional and shadow banking sector visà-vis the inequality measure over the period 1960-2019. Over a thirty years time span
from the 1980s to 2010s, the banking sector almost tripled in size as a share of GDP before
reaching a new steady state in the aftermath of the 2008 crisis. However, what is striking
is that most of the growth in the banking sector has been driven by market-based banking institutions. The sector skyrocketed from the 1980s becoming twice as large as the
amount of national output before stabilizing after the GFC. The traditional banking system, on the other hand, has been virtually flat from the the 1970s. As argued before, most
of the growth in banking is linked to the rise of other non-financial institutions rather than
55

Figure 2.2: Traditional and shadow banking assets as a share of GDP and top 1 per cent income
share (left); and domestically-held safe assets as a share of GDP and top 1 per cent income share
(right), in the United States over the period 1960-2019
(a) (b)
Notes: In Panel (a), the traditional banking sector is defined by the total financial assets of private depository institutions,
which are composed of: U.S.-chartered depository institutions, foreign banking offices in the U.S., banks in U.S.-affiliated
areas, and credit unions. The shadow banking sector is given by the sum of the financial assets of: agency- and GSEbacked mortgage pools, ETFs, finance companies, GSEs, issuers of asset-backed securities, money market mutual funds,
private pensions funds, real estate investment trusts, security brokers and dealers. In Panel (b), the domestically-held
safe assets are obtained by removing the rest of the world component from the following total financial assets: treasuries,
municipal bonds, checking deposits and currency, saving and time deposits, money market mutual funds, security
repurchase agreements (RePos), commercial paper, and GSEs and bonds accounted for 85 percent of their volume to be
consistent with the measurement proposed by Gorton, Lewellen, and Metrick (2012). Sources: See Tables B.5 and B.6 in
Section B.5 of the Appendix for details on the variables sources and construction.
banking activities traditionally intended.8 The share of income in the hands of the top 1
per cent of the income distribution (accounting for capital gains) plotted in blue, on the
other hand, rose from a steady level of about 0.08 in the 1960s to about 0.16 fifty years later.
One of the leading theories in the finance literature behind the rise of the shadow banking sector entertains the hypothesis of a shortage of safe assets. In Figure 2.2b, I show to
what extent the production of safe assets is connected to the rise of inequality after removing the claims held abroad.9 As inequality moved, the production of domestically-held
safe assets closely mirrored the same path.10 The overall correlation for a 60 years time
span (1960-2020) is 97 per cent. This should be the case if higher inequality generates
8
Similarly, it is possible to show that the size of market-based banking out of traditional banking tracks the rise of
inequality.
9
In the appendix, I show that the share of private safe assets to public safe assets is also tightly associated with the rise
of the shadow banking sector over time.
10The picture is the net measure after removing the foreign claims on U.S. safe assets.
56

Figure 2.3: Decomposition of the Unites States financial assets held domestically vis-à-vis by the
rest of the world (left); capital income share and top 5 per cent wealth share in the United States
over the period 1960-2019 (right)
(a) (b)
Sources: See Tables B.5 and B.6 in Section B.5 of the Appendix for details on the variables sources and construction.
a higher demand for safe assets by investors in order to hedge potential risks from their
investments.
After WWII, the world has become more globalized not only in terms of international
trade but also financially. Therefore, it is interesting to investigate whether such massive
rise of finance has been driven by international forces. Figure 2.3a looks at this by decomposing each U.S. financial instrument according to the location of the owners of the
claims — domestic vs. rest of the world —, as provided by the Financial Accounts of the
Federal Reserve. It is true that the share of finance has progressively moved more in the
hands of foreigners over time, however, foreigners have been holding at most 10 per cent
of the total financial claims over the past decades. Also, notice that until the mid-1990s the
share was not larger than it was back in the 1970s. The acceleration began after 1994 and
became more pronounced in the 2000s – in line with the arguments on foreign reserves
accumulation happening after the Mexican and Asian Financial Crisis of 1994 and 1997,
and the entering of China in the World Trade Organization in 2002.11
11Figure B.1 in the Appendix shows that the same cannot be said for U.S. Treasuries. In this case, the fraction held by
foreigners over the past two decades has fluctuated between 30 and over 40 per cent.
57

Given that the proposed theory takes the rise of the capital share as a primitive measure
to generate an increase in wealth inequality, Panel (b) displays such joint occurrence over
time. Up to a stronger cyclical behavior of the capital share, the two series are very closely
connected. They display a downward trend during the redistributive forces of the 1960s
and 1970s before rising again from the 1980s and stabilizing over the past decade.
With that in hand, it is of interest to generate a theoretical framework that is able to
rationalize such joint movements in a more formal fashion.
2.3 A macro-finance model
The current section builds a theoretical framework to analyze the mechanisms described in the introduction. The model is set in discrete time, and agents live over the
time horizon t ∈ {0, 1, . . . }. The model features a representative firm, two sets of heterogeneous households — investors and workers —, a financial technology (representing the
market-based banking system), and a Government budget constraint.12
Firms. Firms are price-takers, and maximize profits statically every period. They employ
two factors of production, capital and labor, and they produce a final good, yt
, whose price
I take as numéraire. Capital is represented by a non-reproducible stock, kt
, which is owned
by the investors. Labor, Nt
, is inelastically supplied by the workers.13 Firms combine the
two factors according to a Cobb-Douglas production function facing constant returns to
scale. The time-varying capital share is represented by αt
, and the labor share is 1 − αt
.
The rented capital is remunerated at rate, dt
, which represents the dividends paid off by
the capital stock. Labor is paid according to the wage schedule, wt
. The amount of capital
is normalized to one. Similarly, a constant population (and labor force) is normalized to
12I abstract from the traditional banking sector, as it was relatively constant over time. Therefore, one could think at this
as a time-invariant technology whose output is normalized to zero.
13Workers do not have access to financial markets, therefore markets are segmented. On the other hand, investors do
not supply labor to maintain a parsimonious structure.
58

one. The problem of the firms is provided in (PF ):
max
kt,Nt
yt − dtkt − wtNt sub yt = k
αt
t N
1−αt
t
(PF )
As a result of labor being inelastically supplied for a constant population, the production of final output is yt = 1 ∀t. It follows that the amount of dividends earned by capitalowners and the wages earned by workers correspond to the factor shares of capital and
labor, respectively: dt = αt
, wt = (1 − αt).
Households. Households (both investors and workers) discount future levels of utility
by the same discount factor β˜ ∈ (0, 1). The dynastic structure features a constant survival
probability, δ ∈ (0, 1); therefore, a constant fraction of the population (1 − δ) exits every
period. This fraction is re-born with average income at each time t to ensure that the
total population and assets in the economy are constant. The exit probability ensures the
stationarity of the model.14 I define the "effective" discount rate as: β , βδ˜ . Both agents
have homothetic preferences and maximize utility derived from personal consumption.
— Investors/Capital-owners. Investors face two joint problems: A consumptionsavings decision, and a portfolio allocation problem. In other words, in each period
investors need to decide the amount to allocate to consumption vis-à-vis savings, and
contemporaneously determine how to invest their savings across different financial asset classes. They maximize their utility according to log-preferences.15 The assets space
is composed of three instruments: risky, public safe assets, and privately-produced quasisafe assets. The recursive problem faced by investors is represented by the Bellman equa14This is a standard approach in the macroeconomic literature to prevent the accumulation of all the income by a single
individual, and it will become clearer from the law of motion of assets.
15Log preferences are a conservative choice for this model structure. Other CRRA utility functions would be perfectly
acceptable, and strengthen my results by allowing for more concavity of the utility function to exacerbate the precautionary motives.
59

tion in (PI ):
V
(I)
it (bit, mit, kit) = max
c
(I)
it ,bi,t+1,
mi,t+1,ki,t+1



log
c
(I)
it
+ βδ˜
|{z}
,β
Et
h
V
(I)
it (bi,t+1, mi,t+1, ki,t+1)
i



(PI )
sub c
(I)
it + pKtki,t+1 + qBtbi,t+1 + qM tmi,t+1 =
= (pKt(1 + it) + dt)kit + bit + (1 + ζ
M
it )mit
| {z }
,Ait
The problem for the investors features four control variables and three state variables.
The control variables are consumption in the current period, c
(I)
it , and investment decisions
over next period in: risky capital shares, ki,t+1, public safe assets, bi,t+1, and private quasisafe assets, mi,t+1. The total value of wealth for each household i is denoted by Ait, and
it is obtained by summing the latter three elements evaluated at market prices. Notice
that the assets distribution is not a state variable. This result stems from the fact that the
model features aggregation, as pertaining to the class of models with idiosyncratic capital
shocks described by Angeletos (2007). The returns on capital and quasi-safe assets are
subject to idiosyncratic shocks. The shocks on capital are drawn from a distribution such
that: F ∼ (0, σ), and the ones on quasi-safe assets from Fζ ∼ (0, σζ ). Both distributions
are assumed to be uniform for simplicity.
Investors own the capital stock of the economy (kt =
R
i
kit di), and each investor i
owns a share of the capital of the economy, kit. Capital is risky. Each investor is hit by
idiosyncratic shocks on their capital holdings, and risks cannot be insured away with adhoc contingent claims. It follows that perfect diversification is not attainable, and that
investors become ex-post heterogeneous after the shocks are realized even if they start
as ex-ante identical.16 Therefore, the model allows for a full-fledged distribution of wealth
types with some investors becoming zero-wealth holders and others extremely wealthy.
16Guvenen, Pistaferri, and Violante (2022) show that the volatility of income earnings for top quantiles is larger than for
the lower quantile. See Figure B.5 in the Appendix for the results for the whole population and divided by age cohort
and gender.
60

I do not allow for aggregate uncertainty, thus, the total supply of capital is known with
certainty. The capital stock is priced at value pKt, and it pays off non-storable dividends,
dt
, every period. Given that the shocks hit the shares of capital, this can be seen as leading
to a stochastic variation in the value of individual capital holdings.17
Safe assets, bit, are risk-free instruments issued by the Government. They generate
perfect insurance and they are provided in positive but limited net supply, ¯b.
Privately-issued quasi-safe assets, mit, are created by the financial sector by transforming the debt of the workers in exchange for a fee. Hence, investors fund the debt of the
workers. A classical no-arbitrage condition holds in the aggregate such that E[RM t] =
E[RBt]. In other words, the model continues to feature no aggregate uncertainty, and some
funds are able to provide small returns while others marginally "break the buck". Given
the complete absence of risk on the Government-guaranteed safe assets, this ensures that
a safety discount on public assets is achieved endogenously.18 However, the degree of uncertainty is small enough not to lead to substantial changes in the equilibrium outcome as
a result of this modeling choice. All prices and returns are endogenously determined by
trading on their respective markets.
Log preferences induce linear policy functions.19 Consumption and assets holdings
policy functions can be written as a linear function of the total assets owned by investors:
c
(I)
it =(1 − β)Ait (2.1)
qBtb
(I)
i,t+1 =βφ1tAit (2.2)
qM tm
(I)
i,t+1 =βφ2tAit (2.3)
pKtk
(I)
i,t+1 =β(1 − φ1t − φ2t)Ait (2.4)
17This is also isomorphic to idiosyncratic depreciation (or appreciation) rates of the individual capital shares.
18The existence of safe and quasi-safe assets will make easier to conduct comparable quantitative counterfactual exercises
later on.
19Log preference with total capital depreciation and aggregation lead to a Brock and Mirman (1972) world with analytically derivable policy functions.
61

where φ1t
, φ2t
, and (1 − φ1t − φ2t) are the portfolio shares of public safe assets, private
safe assets, and risky assets, respectively. It is important to stress that such portfolio shares
are endogenously determined as a result of risk-reward decisions that investors make in
equilibrium. The expressions for φ1t
, φ2t
, and (1−φ1t−φ2t) are provided by the no arbitrage
conditions in Equations (2.6)-(2.5), and solved for numerically.
Lemma 1. Investors allocate their savings endogenously to risky, safe, and quasi-safe assets so that
the following conditions are satisfied:
1 = Et

Ri,t+1
φ1tRB,t+1 + φ2tRM,t+1 + (1 − φ1t − φ2t)Ri,t+1
(2.5)
1 = Et

RB,t+1
φ1tRB,t+1 + φ2tRM,t+1 + (1 − φ1t − φ2t)Ri,t+1
(2.6)
1 = Et

RM,i,t+1
φ1tRB,t+1 + φ2tRM,t+1 + (1 − φ1t − φ2t)Ri,t+1
(2.7)
with Ri,t+1 = (pK,t+1(1 + i,t+1) + dt+1)/pKt, RBt = 1/qBt, and RM,i,t = (1 + ζ
M
it )/qM t.
Proof. See Section B.2 in the Appendix.
From the perspective of the single investor, the assets law of motion for the modeled
economy follows a CAPM-like setting. Lemma 2 makes this statement formally.
Lemma 2. The model features a two factors structure for investors, where Equation (2.8) represents
the law of motion of assets for each agent:
Ai,t+1 = βAit [φ1tRB,t+1 + φ2tRM,t+1 + (1 − φ1t − φ2t)Ri,t+1] (2.8)
with Ri,t+1 = (pK,t+1(1 + i,t+1) + dt+1)/pKt.
Proof. See Section B.2 in the Appendix.
62

Moving on to the model mechanics, it is important to stress the importance of precautionary demand for safe assets by investors. The idiosyncratic and non-insurable shocks
on capital create a demand to hedge against risk exposure. None of the investors has
informational advantage before shocks get realized; therefore, their ex-ante preferences
on portfolio composition are identical. Shocks on larger capital stocks induce a greater
hedging demand as a result of potential larger capital losses to face.20 Once the idiosyncratic shocks are realized, an ex-post distribution of investors is formed according to the
individual-specific capital gains and losses. In equilibrium, the model features a baseline
degree of inequality where a portion of agents that faced a series of negative shocks is left
with an arbitrarily small amount of assets while a small fraction of agents who faced a
series of positive shocks will hold a large fraction of the economy’s assets.
A change in the technological structure of the economy (αt higher) induces a larger
return on capital, i.e., higher dividends dt
, paid to the investors. When capital promises
higher dividend payoffs, investors’ wealth rises. However, as they get richer and they
own a larger amount of assets, they hold also a larger share of the risk in the economy.
Therefore, two opposite effects emerge as capital promises higher gains. On the one hand,
higher expected returns push investors to tilt their portfolios more towards risky assets,
as this could ensure to harvest a greater amount of valuable fruits in the future. It follows
that the trading activity of shares surges, and pKt increases to absorb the excess demand.
Effectively, there is a stock market boom. On the other hand, higher returns come with
higher risks. Risk aversion under incomplete markets gives rise to precautionary motives.
As such, investors buy jointly both shares of risky, safe, and quasi-safe assets in order to
minimize the potential losses from negative idiosyncratic shocks. This type of market
incompleteness leads to increased trading activity also for the safe and quasi-safe assets.
Consequently, safe assets prices also rise (higher qBt, qM t), and returns get compressed.
In general equilibrium, the share of risky assets holdings increases as a result of higher
20A similar argument is proposed by Di Tella (2019) to justify regulation of financial intermediaries.
63

expected dividends although the portfolio tilting towards the risky share gets partially
dampened by the lower returns.
It is also important to notice that the model features a feedback effect. As higher capital
returns generate higher demand, the price of risky assets pKt increases. Hence, capital
valuations increase, and induce investors to become effectively wealthier, and thus with
higher savings to invest back in the system.
— Workers. The fringe of labor-owning households maximizes its intertemporal utility
derived from consumption, c
(W)
t
, and chooses an optimal amount of loans to borrow, lt+1.
In order to pay for the interest on loans and for consumption, the workers inelastically
supply labor, Nt
, to the final good firms at the competitive wage, wt
. Labor earnings are
assumed to be deterministic, thus workers effectively do not feature either idiosyncratic
or aggregate risk. The constrained optimization presented in (PW ) describes the problem
faced by workers.
V
(W)
t
(lt) = max
c
(W)
t
,lt+1
n
u(c
(W)
t
) + βEt
h
V
(W)
t
(lt+1)
io (PW )
sub c
(W)
t + lt +
λ
2
qLt (lt+1 − L/λ)
2 + T
(W)
t = qLtlt+1 + wtNt
The amount of debt that workers can take on is limited by the quadratic costs on the left
hand side of the budget constraint. This represents a "soft" borrowing constraint where
each additional unit of debt gets more expensive to issue, and can be thought of as the
costs that financial intermediaries would impose on workers for greater monitoring. The
parameter λ governs the steepness of such borrowing constraint — the larger the value,
the steeper the costs, and therefore the more stringent the conditions for workers.21
The parameter L contributes to the formation of the wedge that allows to obtain an
21This type of borrowing constraint modeling is isomorphic to a penalty function in the utility function, and sometimes
modeled as such in the macro literature.
64

implicit representation of the financial sector.22 It contributes to the return spread per unit
of debt charged by financial intermediaries to transform the debt issued by the workers
into quasi-safe assets, mit, held by investors. In this baseline set-up, the earnings per unit
of finance are thus assumed to be constant.
Lump-sum taxes, T
(W)
t
, are paid by the workers in order to finance the budget of the
Government. Qualitatively, by allowing taxes to be paid by the workers – the role of the
Government becomes non-Ricardian.23 From a quantitative point of view, lump-sum taxes
play a marginal role for the overall behavior of the system.
It is important to notice that workers’ willingness to issue debt stems from its costs. The
precautionary demand for quasi-safe assets by the investors leads to a reduction of the real
interest rate below the inverse of the effective discount factor – which is the shadow price
of debt for the workers under complete markets. As such, it is always advantageous for
workers to issue debt, and increase consumption at the proposed market rates.
The larger the precautionary demand for safe assets, the more the interest rates decrease, and the larger the amount of debt issued by households. However, notice that this
mechanism also allows for an increase in the leverage of the system. As debt gets progressively cheaper, lower interest rates induce workers households to raise an amount of funds
which becomes relatively bigger as a share of total income.
A soft borrowing constraint produces an upward-sloping supply curve for debt rather
than an inelastic vertically-sloped curve generated for the case of a strict threshold. An
upward sloping supply curve has the advantage of allowing both prices and quantities of
debt to move rather than fixing any of the two.24
22It is normalized to separate the direct effects of moving L with respect to the unit cost λ.
23If investors were to be taxed instead of workers, rich households would earn an interest on safe assets. Therefore, if
they were to be taxed by the same amount the two effects would cancel out.
24Without constraints, the price schedule of debt would be flat, and interest rates would be mechanically independent
from the amount of debt in the economy.
65

Government balanced budget constraint. The Government imposes a lump-sum tax on
the workers and issues debt up to a borrowing limit, ¯b, in each period.
T
(W)
t + qBtbt+1 − bt = 0 sub bt ≤ ¯bt ∀t (PG)
Financial technology. To maintain a parsimonious structure, the implicit financial sector
transforms the debt of the workers into quasi-safe assets for the investors according to a
simple linear technology lt+1 = mt+1 ∀t.
Market clearing. In equilibrium, the total number of capital shares has to sum to the
normalized size of the capital stock: R
i
kit di = kt = 1, ∀t. Markets for the shares of
capital clear endogenously at price pKt. The market for safe assets is cleared at price qBt
for the amount provided exogenously R
i
bit = ¯bt ∀t. The total amount of debt issued by
the workers is equalized to the amount of quasi-safe assets invested by the investors, thus
R
i mit = lt
, and the market-clearing prices are qM t = qLt, ∀t. The labor force is constant,
and labor is inelastically supplied: Nt = 1, remunerated at wage wt = 1 − αt
. Capital is
rented at price dt = αt
. Final output is valued at price equal to one by construction (the
numéraire), and consumed by both households. See Section B.2 in the Appendix for a
complete characterization of the competitive equilibrium.
To help visualize the model structure, Figure 2.4 provides a representation of the
model.
2.3.1 Sensitivity analysis
In order to clarify the properties of the model, Figures 2.5 and 2.6 present a series of
sensitivity analyses for the cases in which income volatility and the capital income share
share change, respectively. Without loss of generality, for the simulation results only —
I further simplify the model structure by assuming that the government does not issue
66

Figure 2.4: Visual representation of the model
Firms
Output production
Baseline inequality
Labor owners
Borrowers
Capital owners
Investors
L K
↓ L share ↑ K share
↑ Inequality
Workers
↓ earnings
Investors
↑ savings
↑ Financial intermediation
↓ Real interest rates
↑ debt
demand
↑ loan
provision
↑ safe
assets supply
↑ hedging
demand
↑ Indebtedness,
↑ Leverage
Note: Placeholders in yellow stand for the trends the model captures. Placeholders in blue stand for market participants in the economy. Text outside of placeholders helps motivating the driving forces and initial set-up.
any public bonds (¯b = 0), and that the private safe assets produced by the financial sector are perfectly safe. All the properties would go through if we were to generalize such
conditions, as in the exposition provided in the previous section.
Panel (a) shows that investors demand more safe assets when capital income becomes
more volatile as a result of precautionary motives. Investors demand a larger amount of
privately produced safe assets to hedge their idiosyncratic risks.
In Panel (b) it is possible to see that the returns decline as a result of such higher safe
assets demand. This result stems from the upward sloping supply curve of debt coming
from the soft budget constraint of the workers. If the supply of debt is not perfectly elastic,
67

then returns decline when the demand increases.
Panel (c) provides a description of the portfolio tilting happening as a result of variations in volatility. When volatility rises, not only the absolute amount of safe assets increases, but also the relative share of safe vs. risky tilts more towards safety (i.e. φ surges).
In a traditional Merton setting, portfolio shares would not adjust because interest rates are
taken in partial equilibrium. However, when interest rates do decline – as in the current
general equilibrium framework – this has an effect on the portfolio composition. On the
one hand, an increase in capital risk pushes the allocation more towards safe assets to
obtain a greater hedge; however, such effects are partially reduced by the fact that safe
assets generate a lower return. Rather sticky portfolio shares emerge as a result (although
portfolio shares cannot stay constant in equilibrium).
In Panel (d), I show that an increase in income volatility pushes up the leverage of the
workers as a result of debt becoming cheaper. If real interest rates decline, the costs of
repayment decline, which incentivizes a larger fraction of debt to be issued.25
Figure 2.6 presents the results for a constant volatility of the income shocks when the
capital share, α, rises. Many of the same qualitative result are similar to what seen before
except for the portfolio composition. The amount of safe assets surges when the capital
share increases, as shown in Panel (a). This is an important feature of the model. A higher
capital share induces larger expected dividends but also larger potential absolute portfolio
losses coming from a greater amount of wealth under management. As a result, investors
contemporaneously increase both the demand for risky and safe assets. Panel (b) shows
that interest rates decrease when that is the case, which in turns pushes leverage upwards
for the reasons expressed above. See Panel (d).
The safe assets share declines in this case, as illustrated in Panel (c). Even though
investors demand more safe assets in absolute terms, their relative portfolio composition
tends more towards holding risky assets because of the extra returns that can be gained
25Households leverage is computed as the amount of debt issued, qMtmt+1, divided by the net disposable income after
paying the interests on debt, wt.
68

Figure 2.5: Sensitivity analysis under different volatility scenarios
(a) (b)
(c) (d)
Note: The model parameters for the simulations are: β = 0.96, δ = 0.94, λ = 0.02, L = 0.01, α = 0.2. All shocks are
assumed to be drawn from iid uniform distributions U(1 − ∆, 1 + ∆).
under the new scenarios.
I move now to the calibration exercises needed to bring the model to the data.
2.4 Calibration
The model features some parameters estimated in the data, and a few free parameters
internally calibrated jointly to match the moments of interest. Table 2.1 provides the overall list of calibrated parameters. Tables B.5 and B.6 in the chapter’s appendix explain in
detail the data sources and construction.
The levels of wealth inequality of the top 5 per cent share are obtained from the World
69

Figure 2.6: Sensitivity analysis under different capital share scenarios
(a) (b)
(c) (d)
Note: The model parameters for the simulations are: β = 0.96, δ = 0.94, λ = 0.02, L = 0.01, ∆ = 0.4. All shocks
are assumed to be drawn from iid uniform distributions U(1 − ∆, 1 + ∆).
Inequality Database (WID.world). Such measure is a mildly conservative figure. Wealth
inequality has rose mostly for the top 10 per cent of the distribution but the gains have
been very skewed in favor of the top fractiles of the top 1 per cent.
The capital share is computed as (1− lt), where lt
is the labor income share. Figure 2.3
represents the time series of (1 − lt). The labor income share is obtained from Table 2 of
the National Income and Product Accounts (NIPA) as compensation of employees divided
by personal income plus subsidies minus taxes. As such, the capital share is intended as
non-labor share.26
26See Tables B.5 and B.6 in the Appendix for the specific NIPA tables lines used for computation.
70

The construction of idiosyncratic volatility (σ) is based on daily stock returns available from CRSP; it follows the asset pricing literature on the matter (Fu, 2009), and it is
explained in detail in Section B.5 of the Appendix.
The levels of shadow banking financial intermediation refer to the "net" measure, i.e.,
the claims held in the hands of households, after stripping away the interbank holdings. To
do so, I use the data from the Financial Accounts of the Federal Reserve. The final measure
takes into account a "direct" and an "indirect" shadow component in order to avoid double
counting or defining as "shadow banking" elements that could be seen as spurious. The
direct portfolio holdings I can read directly from households balance sheets are given by:
Money market mutual funds and GSEs. For other financial intermediaries that are not
clearly definable as either inside or outside the such perimeter, I need to obtain a weighted
average of how "shadow" they are. These mixed intermediaries are: mutual funds, pension
funds, and life insurance funds. In this case, the indirect portfolio holdings are obtained
by looking at the amount of shadow banking assets (money market funds, repurchase
agreements, commercial paper, and GSEs) implicitly held by households through these
mixed financial intermediaries. In case there is a further nest, e.g., RePos held by mutual
funds, which in turns are held by pension funds, I operate a double weighted average by
looking at how sizable each component is with respect to the total assets of the financial
intermediary considered.
The real interest is calibrated by subtracting the personal consumption expenditures
inflation (PCE) to the Moody’s AAA Corporate Bonds Yield. By doing so, I aim to mitigate the potential confounding factors stemming from the foreign demand of Government
securities. The series on AAA Corporate Bonds Yield and 10 years Treasuries rates correlates at over 98 per cent for the 1970-2019 period.
The share of publicly available safe assets qBtbt+1/yt
is obtained from the U.S. Financial Accounts as the sum of Treasuries, and municipal bonds, and checking deposits and
71

currency held by households as a fraction of nominal output.27
The model predicts a small fraction of the population to have negative asset values, in
this respect, I use to the estimates from the Survey of Consumer Finances.
Table 2.1: Calibrated parameters
Parameters Value Source
Debt issuance variable cost (λ) 0.0092 Internal calibration
Debt issuance fixed cost (L) -0.0718 Internal calibration
Survival rate (δ) 0.9823 Internal calibration
Discount factor (β˜) 0.9174 Internal calibration
Quasi-safe shock (ζit) 0.050 Internal calibration
Moments baseline
Capital share 1970-79 (α1) 0.279 NIPA Tables
Idiosyncratic variance 1960-72 (σ) 0.543 CRSP
Public safe assets to GDP, 1970-79 (qBb1/y) 0.200 Fed Financial Accounts
Moments experiment
Capital share 2010-2019 (α2) 0.327 NIPA Tables
Idiosyncratic variance 2010-19 (σ) 0.570 CRSP
Public safe assets/GDP, 2010-2019(qBb2/y) 0.220 Fed Financial Accounts
The other parameters left are obtained by joint internal calibration to result in the desired initial steady state values for wealth inequality, financial intermediation and real
interest rate. In this respect, one needs to calibrate the survival rate, δ, the discount rate,
β˜, the debt issuance variable cost for workers, λ, the debt issuance fixed cost, L, and the
maximum gains or loss on the quasi-safe assets, ζM.
28
The initial steady state is solved by using the aforementioned policy function conditions (Equations (2.1)-(2.4) in the previous section), the endogenous portfolio shares in
Equations (2.6) and (2.7), the budget constraint and the Euler equation of the workers
emerging from (PW ), the government budget constraint, and the market clearing condi27The Financial Accounts of the Fed do not allow to separate currency from checking deposits. However, given that a
large fraction of checking deposits is Government-guaranteed through the FDIC, the assumption is consistent with
the modeling choice.
28If the chosen values of the parameters do not allow to match the initial level of inequality, they need to be fine-tuned
until inequality and the two other moments are matched.
72

tions. Given aggregation, the equilibrium values of {φ
∗
1
, φ∗
2
, p∗
, m∗
, b∗
, k∗} are then used to
compute the income distribution according to the law of motion in Equation (2.8).
As part of the baseline quantitative exercise, I keep all parameters fixed except the rise
of the capital share, α, used to generate higher inequality, the related estimated change in
idiosyncratic volatility, σ
, and the safe assets share held by the households.
2.5 Quantitative results and policy experiments
The model performance is tested by looking at its predictions across steady states:
1970-79 vs. 2010-2019. I first target the level of inequality, financial intermediation as a
share of GDP, and real interest rates over the period 1970-1979 consistent with the capital
share that I see in the data. This is taken as a first steady state. Subsequently, I introduce an
exogenous structural change in αt (the model reduced-form parameter leading to higher
capital share) consistently with the estimates from the NIPA tables.29
Even though the theoretical approach is more interested in the effects of inequality
(overall, regardless of the source) on the rise of finance, I follow a conservative approach
that generates only a fraction of the overall inequality seen in the data in order to understand how far the current micro-foundations can bring the model performance, and unveil
a new macro-finance relationship. The performance of the model is assessed by looking
at how well other dimensions of interest can be predicted, especially the rise of the financial sector and the decline of the real interest rates. A few further moments endogenously
generated by the model are provided such as the share of risky assets, and the price level
of capital.
Quantitative analyses based on the calibration provided above are carried out to test
the model performance: I call this the baseline scenario. A set of counterfactuals is then
provided to assess the behavior of the system under different policy regimes.
29In order to be model-consistent, I also need to estimate the idiosyncratic volatility of capital in the two steady states. I
do so by using CRSP data, and I follow the approach by Fu (2009). Section B.5 this chapter’s appendix describes the
process in detail.
73

Table 2.2: Quantitative results for the baseline model
1970-79 2010-2019
Targeted moment Model Data Model Data
Top 5% wealth share 0.508 0.508 0.523 0.582
Shadow Banking holdings (qM tmt+1/yt) 0.026 0.026 0.161 0.212
Real interest rate (RM t)
† 0.030 0.030 0.028 0.025
Additional moments Model Data Model Data
Equity Premium† 0.088 0.055 0.093 0.081
Risky assets share (1 − φ1 − φ2) 0.914 0.652 0.878 0.654
Notes:
† Results refer to the period 1960-1972 for the real interest rates to avoid the oil shocks to play a role.
2.5.1 Baseline scenario
Table 2.2 presents the results for the quantitative exercise conducted comparing the
1970-1979 period to the following one spanning over 2010-2019, as a result of the structural
change in the technology parameter, α. Qualitatively, the model correctly predicts all the
directions of the moments of interest: Higher inequality, higher financial intermediation
as a share of output, lower real interest rates, and a rather stable portfolio composition.30
The increase in the capital share is able to deliver 20 per cent of the increase in inequality seen in the data. The rise of financial intermediation responds by explaining 73 per
cent of the increase seen in the data; while the real interest rates decrease from 3.0 to 2.8
per cent (40 per cent of the variation). In this respect, it is more than plausible that other
macroeconomic factors are at play to exacerbate the safe asset shortage and the reduction
in interest rates.
In terms of non-targeted moments, the risky asset portfolio shares is not in line with
the one in the data. Given that this is not a moment the model tries neither to target nor to
predict, it is not particularly worrisome. On the other hand, the model correctly predicts
30A perfect stability in the portfolio composition share – as seen in the data – is impossible to attain in the current model
by construction.
74

Table 2.3: Policy experiments results
Baseline Counterfactuals
Moments 1970-79 2010-19 τ = 0.10 τ = 0.015,
bt = 0.378
Top 5% wealth share 0.508 0.523 0.521 0.519
Shadow Banking holdings (qM tmt+1/yt) 0.026 0.161 0.129 0.027
Real interest rate (RM t) 0.030 0.028 0.029 0.030
Risky assets share (1 − φ1 − φ2) 0.914 0.878 0.876 0.870
Equity premium 0.088 0.093 0.106 0.094
the rise in the equity premium albeit less than in the data. In the model, it rises from 8.8
to 9.3 per cent, while in the data I see a rise from 5.5 to 8.1 per cent.
2.5.2 Dividends taxes
As a first counterfactual experiment, I allow the government to tax also investors on
their dividends gains with a proportional tax, τt
. In this first scenario, I do not allow
the government to issue more public bonds as a result of higher taxation but, rather, to
effectively redistribute its revenues to workers – which now will endogenously receive a
negative tax (a subsidy), T
(W)
. The modified budget constraint of the investors’ problem
(PI ) can be re-written as in Equation (2.9):
c
(I)
it + pKtki,t+1 + qBtbi,t+1 + qM tmi,t+1 = [(pKt(1 + it) + dt(1 − τt))] kit + bit + (1 + ζt)mit
(2.9)
while the modified Government budget constraint is represented as follows:
T
(W)
t + τtdt − bt + qBtbt+1 = 0 sub bt ≤ ¯b (2.10)
The first column of the counterfactuals in Table 2.3 shows the results of such policy for
a tax of 10 percent on the dividends gains after the change in α has occurred. As can be
75

seen, the share of income held by the top 5 per cent increases up to 0.521 rather than 0.523.
In other words, inequality would have increased only slightly less than in the baseline
scenario. The effect of subsidies in this model is not very strong because most of the rise of
inequality happens within investors. As a consequence of a smaller increase in inequality,
the financial sector would have increased up to 0.129 rather than 0.161 documented before.
Interest rates would have been almost completely untouched by such shift going from 3.0
to 2.9 per cent. The portfolio share in risky assets would have been overall similar as a
result of still substantial gains (0.876 in the counterfactual vis-à-vis 0.878 found before).
Overall, the lower dividend gains produce an expected decrease in inequality (by construction), and a reduction in the amount of loanable funds, which decrease the surge of
finance, and put a lower pressure on real interest rates.
2.5.3 Unconstrained public safe assets supply
As a follow up exercise, I reduce the dividend tax to τ = 0.015, but I allow the
government to issue as much debt as demanded by the investors in the form of public
safe assets (up to fiscal capacity). To help comparing the results with the 2010-19 baseline scenario, I fix the lump-sum taxes that were imposed on the workers in the baseline scenario, T
(W)
. Thus, the new Government budget constraint can be represented as:
T
(W)
+ τtdt + qBtbi,t+1 − bit = 0. Results are provided in the last column on the right of
Table 2.3.
The first key insight is that — when allowing the government to issue debt up to fiscal
capacity —, the supply expands dramatically: from 0.220 to 0.378 times the national output.31 By construction, such a large increase in quantities is possible because of a flattening
of the (previously rigid) supply of public bonds. As a result, the model now predicts the
real interest rates to be virtually unchanged. However, what is more interesting is that
such an expansion of public safe assets strongly reduces the increase of inequality allow31In this case, the ceteris paribus assumption is crucial. I am allowing for all the extra supply to be taken up by the U.S.
households rather than by foreigners.
76

ing it to decrease from 0.523 to 0.519. This stark result is the by-product of safe assets
becoming more attractive (because of higher interest rates paid on debt), which reduces
the "search for yield" for investors and the increase in capital assets valuations, pKt. As a
result, the feedback effects through capital gains is dampened. In other words, if the real
interest rates do not change much, then the incentive to move away from them is strongly
reduced, and the feedback effect derived from capital gains is diminished, which prevents
investors to become wealthier.
The shadow banking sector is left virtually unchanged as a result of fewer loanable
funds to be privately intermediated. Importantly, this happens because funds are now
being diverted towards public bonds — and it is consistent with literature on crowdingout of public vs. private safe assets (Krishnamurthy and Vissing-Jorgensen, 2015).
An important note of caution should be used when interpreting these results. The
model does not speak to the importance of removing all fiscal discipline, thereby satisfying investors demands. Rather, it speaks to the feedback effects of the "reach for yield"
and the large opportunity costs of holding safe assets when money yields are exceptionally
compressed.32
2.6 Empirical analyses
In the empirical section, I study the predictions of the theoretical framework in two
different ways. In Section 2.6.1, I investigate the extent to which finance and inequality
have been intertwined in the United States over the very long run by using data for the
longest time span available for this type of exercise, i.e., from 1913 to 2019. In this respect,
I use the novel econometric techniques of "co-variability" proposed by Müller and Watson
(2018).
Subsequently, Section 2.6.2 tests the model mechanism itself behind the rise of finance.
32See Acharya and Dogra (2022) on the welfare effects of increasing the public safe assets amount in a context where
investments are allowed and monetary policy is constrained by the zero lower bound.
77

The model implies the rise of finance happened through market-based banking mechanisms, not traditional banking instruments. Given that banking structures can be very
different across countries, I lever on the identifying mechanism technique proposed by
Rancière, Tornell, and Westermann (2008) to test whether the industrial organization of
banking plays a crucial role to generate an expansion of credit. Results are consistent with
this interpretation in terms of feedback effects.
2.6.1 The long-run co-variability of inequality and finance
In the this subsection, I investigate the extent to which the inequality and financial time
series have proportionally co-moved over the long-run. As noted by Chudik, Pesaran,
and Smith (2023), over the very long-run, major events such as wars, pandemics, or other
regime-shifting episodes may induce problems when estimating the long-run coefficients
across time series even when the series are co-integrated. To this extent, having a clear
understanding about the nature of the historical process connected to Bretton Woods (seen
as a double regime shift), I tranche the over hundred years period of data in three subperiods. The long-run co-variability analysis of Müller and Watson (2018) is applied for
the sample periods: 1913-1943, 1947-1972, and 1973-2019, separately.33
Given an integrated process zt ∼ I(d), where d is the degree of integration, Müller and
Watson’s approach allow for any degree of integration conditional on d ∈ (−0.5, 1.5). To
tie my hands in the most conservative way, I take the inequality process and the financial
assets as a share of GDP in rates of change — de facto studying the stationary, I(0), processes. In this way, I study the extent to which a given per cent increase in inequality is
linked to a given per cent of the financial sector over time.
The exponential smoothing filter to isolate the long-run variation from cycles is chosen
33The data between 1880-1913 cannot be utilized here because they rely on linear interpolations at decade level, which
would spuriously affect the results. The second period starts from 1947 to use internally-consistent data from the
Federal Reserve Financial Accounts, which extends to today, rather than the one by Philippon (2015), which stops in
2010. The latter series is used only for the first sample of the analysis as it represents the best historical quantification
going back in time.
78

to be 7-years in the post-WWII world, while a longer 11-years cyclicality is used to account
for the Great Depression event in the first part of the sample. In this case, the choice
is consistent with the choice of the authors in their paper. The long-run projections are
presented in Figure 2.7.
34
Figure 2.7: Long Run Projections
It seems rather evident that the timing of the cycles in the pre-/post-Bretton Woods
world is identical with the one of the financial sector, although the latter reacts more than
proportionally to the former in the pre-Bretton Woods era. A result that is clearly driven
also from the credit boom effect before 1929, and the subsequent dramatic collapse. Yet,
even accounting for such spectacular rise and fall of finance, the two similarly co-move.
The series become much more independent during the financial repression and strong
re-distributive period from the 1940s to the early 1970s, before going back to move synchronously in the following period until today.
Table 2.4 presents the results for the ρ coefficients, referring to the long-run correlation
coefficients, and the βˆ coefficients, referring to the long-run best linear prediction of the
long-run projections of finance growth by the long-run projection of inequality growth.
34For matters of the picture purpose only, the initial period is extended to 1946 using Philippon (2015) data. Results
would be extremely similar by removing the last three years, and leaving a discontinuity between 1944-1946.
79

Table 2.4: Long-run co-variability results over the three relevant sub-periods
1913-1943
ρ β σy|x
Point estimate 0.733 1.137 0.043
67% Conf. Interval (0.470, 0.899) (0.750, 1.503) (0.032, 0.062)
90% Conf. Interval (0.160, 0.945) (0.428, 1.947) (0.027, 0.086)
1947-1972
ρ β σy|x
Point estimate -0.079 -0.232 0.015
67% Conf. Interval (-0.512, 0.311) (-1.024, 0.506) (0.010, 0.024)
90% Conf. Interval (-0.750, 0.700) (-1.847, 1.546) (0.008, 0.039)
1973-2019
ρ β σy|x
Point estimate 0.714 0.592 0.013
67% Conf. Interval (0.490, 0.870) (0.407, 0.785) (0.010, 0.017)
90% Conf. Interval (0.273, 0.923) (0.166, 1.143 ) (0.008, 0.026)
Note: ρ refers to long run correlation coefficient, βˆ refers to the long run regression coefficient of finance responding to
changes in inequality, and σy|x is the related standard errors of the regressions.
The σy|x coefficient refer to the average variance of the prediction error (with y being the
finance to GDP ratio, and x the effect of inequality). As evident, for the post Bretton Woods
world the relationship is very strong and significant: βˆ = 0.592. In other words, the size
of the financial sector responded to the increase in the top 1 per cent income share almost
with the same elasticity as GDP growth responded to TFP growth documented by the
authors in the original study. Such coefficient becomes largely non-statistically significant
for the 1947-1972 period (as expected), while it suggests an overshooting over the period
1913-1943 (βˆ = 1.137) — probably as a result of the large shock due to the pre-1929 boom
and subsequent devastating Great Depression.
The long-run correlation coefficients are also high for the periods of laissez-faire (0.733
pre-Bretton Woods, and 0.714 post-Bretton Woods), and as strong as the relationship between TFP growth and output growth found in other long-run econometrics works. See
Müller and Watson (2018) empirical section itself.
80

2.6.2 Testing for the identifying mechanism across advanced economies
The previous subsection provided evidence on the degree of long-run co-variability
between inequality and finance. Figure 2.8 looks at the correlation between loans issued
and inequality over the very long run, and suggests that the relationship is there for a host
of advanced economies for the period before and after the Bretton Woods regime (1912-
1943, 1974-2019), but not during the Bretton Wood era. However, the model provides
sharper predictions that can be tested by levering on a panel of countries.
By taking advantage of the MacroHistory database compiled by Jordà, Schularick, and
Taylor (2017), I can study whether the impact of inequality on credit holds more in general
outside the United States. Inequality has increased in a rather pervasive but not uniform
fashion across advanced economies, which ensures enough variation to test the model
prediction of higher inequality leading to a larger financial sector.35
Furthermore, the model predicts that a feedback effect is at play, but that this holds
true because of an increase in asset prices valuations. In order for that to be true, contracts
need to be continuously priced as in a market-based economy. However, credit systems
around the developed world fall into two broad categories: Bank-based – more typical of
continental Europe and Japan –, and market-based financial systems – more common in
Anglo-Saxon countries.36 In bank-based economies, the asset pricing valuation effect is
much less pronounced as contracts are not traded at market prices, while this tends to be
the case by construction in a market-based economy. In order to address this point more
concretely, I take advantage of the identifying mechanism approach proposed by Rancière, Tornell, and Westermann (2008), who test the relationship between financial crises
and growth by exploiting the interaction of skewness effects typical of financial crises, and
politico-economic proxies. I proceed in a similar spirit by interacting the dependent vari35Countries such as Belgium, France, and Spain have experienced almost no increase in inequality over time. Conversely,
Denmark, Germany, and Italy went through impressive distributional changes. Other countries have followed yet other
patterns – almost proceeding in waves like Austria or the Netherlands.
36See Levine (2002) for a classic reference on the matter.
81

Figure 2.8: Total loans to GDP ratio, and top 5 per cent income share for a panel of 18 advanced
economies over the period 1912-2019
Notes: Each dot in the figure represents results for one year. The (x,y)-coordinates are obtained by regressing the variables of interest (inequality and credit) on time-year fixed effects, and country fixed effects. More specifically, I run
separately the regression: yit = ci + κt + it, with yit being: first, log of the top 5 per cent income share, and then, the
log of total loans to GDP ratios. The figure reports the year fixed effects for the two regressions. Standard errors are
robust, clustered at country level. The pre-Bretton Woods period is 1912-1943; the Bretton Woods period is 1944-1973;
and the post-Bretton Woods period is 1974-2019. The blue and red lines represent linear best fits over the periods considered. The sample of countries is composed by: Australia, Belgium, Canada, Denmark, Finland, France, Germany,
Ireland, Italy, Japan, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom, and the
United States. The actual values for loans to GDP and top 5 per cent income inequality corresponding to the specific
time-year fixed effects coefficients have been used on the axes to facilitate the understanding of the true levels of the
variables. Sources: Loans and nominal GDP data are from the the macro-history data base by Jordà, Schularick, and
Taylor (2017). The top 5 per cent income share is the pre-tax and transfers measure from the World Inequality database.
ables of interest with a dummy variable that accounts for the fact that a country may have
a more financialized structure.
In line with the literature on growth equations, I run the regressions provided in Equations (2.11) and (2.12) through ARDL estimation allowing for time and country fixed effects besides a host of control variables. More specifically, I run:
∆yit = κt + β0i + β1∆yi,t−1 +
X
5
s=2
βs∆xi,t−s + γ
0Xit + it, (2.11)
∆yit = κt + β0i + β1∆yi,t−1 +
X
5
s=2
βs∆xi,t−s +
X
5
s=2
βs∆xi,t−s1(i ∈ M) + γ
0Xit + it, (2.12)
82

where κt
is the time fixed effect, β0i
is the country fixed effect, β1 pins down the coefficient of the auto-regressive component, the coefficients β2 − β5 are the estimate of the
cumulated impact of the main dependent variable, Xit is the matrix of domestic and international control variables. For the regressions studying the direct effect of inequality on
finance, the variables {yit, xit}, are the growth of lending activity and the top 5 per cent
of the income share, respectively; they switch when studying the feedback mechanism.
This structure can be effectively thought as an augmented Granger non-causality test in a
panel setting. It follows that the dependent variables of interest are not taken contemporaneously to avoid spurious effects. The number of lags is chosen following the suggestion
by Hamilton (2018) in order for relatively slow-moving variables to fully manifest their
effect, but it is robust to adjacent lag orders. In the latter equation, M is the set of the most
financialized countries, as emerging from IMF Financial Development Index. Therefore,
the dummy variable allows to accommodate for financial market structure differences. In
this case, M includes: Australia, Canada, Switzerland, the Netherlands, the United Kingdom, and the United States.
In Table 2.5a, it is possible to notice that an increase in inequality is a predictor of a size
in credit activities. Such estimates are very robust to a number of different specifications
with domestic and international variables used as controls. The elasticity is systematically
around 0.35 for the cumulated long run effect; namely, an increase of one percent of the
top 5 per cent income share happening over the previous 5 years induces a 0.35 per cent
increase in the amount of loans issued. The coefficients are stable and strong in terms
of statistical significance. Interestingly, the fact that economies are bank- or market-based
does not matter for the direct channel. In other words, when system become more unequal
they tend to generate a larger amount of credit activity regardless of the characteristics of
the intermediation sector. The results are consistent with the idea that higher inequality
generates an increase in the amount of savings to intermediate by financial intermediaries.
Table 2.5b looks at the feedback mechanism – and the identifying mechanism for the
83

Table 2.5a: Regression results for the effects of inequality on total loans issuance for a host of 18
economies over the period 1974-2019
Total loans
(1) (2) (3) (4) (5) (6) (7) (8)
Top 5 income share 0.593*** 0.356*** 0.309** 0.315** 0.663*** 0.414*** 0.359*** 0.336**
(LR effect) (0.167) (0.134) (0.130) (0.135) (0.189) (0.138) (0.136) (0.138)
Top 5 income share × -0.348 -0.306 -0.260 -0.117
Mkt-based dummy (LR) (0.311) (0.285) (0.283) (0.254)
Time fixed effect
Domestic controls
Globalization controls
USA excluded
R
2
0.588 0.636 0.649 0.650 0.591 0.637 0.650 0.650
Countries/Obs. 18/674 18/674 18/670 17/621 18/674 18/674 18/670 17/621
Note: The long run effect, θ, is estimated as: θ =
P5
s=2 βs/(1 − β1). See Chapter 6 of Pesaran (2015). The top 5 per
cent income share is the pre-tax national income of adults (households evaluated as equal-split), and retrieved from the
World Inequality Database following the distributional national accounts (variable code: sptinc992j). The post-tax and
transfers measure would result in the loss of most observations for most countries; hence, it has been avoided on small
sample size grounds. The market-based bank dummy equals one for: Australia, Canada, Switzerland, the Netherlands,
the United Kingdom, and the United States. The remaining countries are: Belgium, Denmark, Finland, France, Germany,
Ireland, Italy, Japan, Norway, Portugal, Spain, Sweden. Domestic controls include: old dependency ratio rate of change,
gross domestic product growth, population growth, money supply growth, government expenditures rate of change,
and inflation. Globalization controls include the rate of change of: trade balance, and debt liabilities constructed by
Milesi-Ferretti (2022). Debt liabilities include the sum of the stocks of portfolio debt liabilities and other investment
liabilities in the hands of nonresidents. All variables are at annual frequency, and taken in real terms after deflating for
inflation. All the domestic and trade controls are retrieved from Jordà, Schularick, and Taylor (2017). ∗∗∗p < 0.01,
∗∗ p <
0.05,
∗
p < 0.1. Standard errors are clustered at country level.
Table 2.5b: Regression results for the effects of total loans on inequality for a host of 18 economies
over the period 1974-2019
Top 5 income share
(1) (2) (3) (4) (5) (6) (7) (8)
Total loans -0.071* -0.029 -0.027 -0.032 -0.116*** -0.065* -0.065* -0.064*
(LR effect) (0.040) (0.032) (0.033) (0.034) (0.040) (0.055) (0.036) (0.036)
Total loans × 0.200*** 0.150*** 0.138*** 0.132***
Mkt-based dummy (LR) (0.055) (0.049) (0.051) (0.051)
Time fixed effect
Domestic controls
Globalization controls
USA excluded
R
2
0.185 0.263 0.272 0.274 0.215 0.290 0.299 0.302
Countries/Obs. 18/732 18/732 18/728 17/679 18/732 18/732 18/728 17/679
Note: See notes in Table 2.5a.
84

feedback effect itself. As predicted in the model, an increase in the credit sector generates
higher inequality only if these securities can be continuously priced, and thus appreciate
and generate a wealth effect for its owners. The results are in line with such prediction.
On average, when pooling all countries together there is no feedback mechanism. In this
respect, this is similar to what most literature has found: the causality link from finance to
inequality is elusive at best. See Demirgüç-Kunt and Levine (2009). However, when interacting the credit variable with the dummy variable capturing the market-based banking
structure, results become strongly significant and extremely stable across specifications.
A result that corroborates the theoretical prediction, and that may be further tested in the
future by the literature.
2.7 Concluding remarks
In this paper, I bring back to the fore the question about what generated the massive
increase in the financial sector intermediation that has occurred in the United States from
the early 1980s. I claim that the growth in size needs to be understood in conjunction with
the endogenous rise of the shadow banking sector.
The paper claims that the rise in inequality observed over the same period may be responsible for such increase, and studies such relationship when the production technology
of the economy changes. The paper builds the first connection, to my knowledge, between
a change in the technological structure and a change in the size and composition of the
financial sector. In the theoretical mechanism, a rise in the capital share generates higher
inequality that accounts not only for the phenomena aforementioned but also for parallel
macro-financial trends: Lower interest rates, higher households indebtedness, and higher
leverage in the system. The model features also a feedback effect that goes from higher
asset prices valuations to higher wealth inequality.
The stylized facts provide descriptive evidence of such joint behaviors. In this respect,
85

the levels of correlation seem to be particularly high and persistent in the very long run
except for periods of structural change such as Bretton Woods.
With that in hand, I build a macro-finance model in which a rise in the share of income
and wealth held by the top 5 per cent share of the population generates an increase in
the funds to invest and allocate across both risky and safe assets. When safe assets are not
abundant, precautionary motives lead to compressed real interest rates. A low interest rate
environment decreases the costs of issuing debt, thus, facilitating higher indebtedness of
the workers. In this scenario, financial intermediaries can step in to manufacture privatelyproduced safe assets that the government is not able to supply.
Quantitatively, I study the rise of inequality as resulting merely from a change in the
technological structure of the economy, and look at how far the results can go. Through
the lenses of the model, the rise of the capital share accounts only for a fraction of the
overall rise in inequality (20 per cent), but it is still able to explain more than 70 per cent
of the overall growth of finance. The model can explain up to 40 per cent of the decline in
real interest rates. A broader set of structural changes in the economy consistent with the
full rise of inequality (such as college premium) may lead to much stronger results.
Finally, the empirical section corroborates both the long-run co-variability between inequality and finance in the post-1970s world, and the importance of the market-based
mechanisms to induce or dampen the effect of inequality on credit production.
Although the paper is not structured to delve into welfare analyses of the problem at
stake, the good news arising from it is that there is room for policy to potentially affect the
macroeconomic consequences of banking and its feedback effects.
86

Chapter 3
The Political Economy of Banking and
Shadow Banking Competition
3.1 Introduction
From a macroeconomic perspective, the importance of financial intermediation cannot be understated. On the bright side, financial intermediaries create the conditions to
generate growth in the so-called "real" sector by screening and monitoring productive entrepreneurial projects, and by providing liquidity and stable means of payments to the
overall economy. However, the dynamics of credit creation can also generate excesses and
crises (Jordà, Schularick, and Taylor, 2017). To this extent, the modifications happened
in the U.S. banking sector from the 1980s appear to have far-reaching consequences, and
deserve close scrutiny.
Following WWII, the banking sector of the U.S. has gone through thirty years of rather
stable and unchanged patterns in the organization of its business and operations. In environments such as the one of Savings & Loans institutions, the utmost stability of the
industry is exemplified by the legendary ‘3-6-3 rule’: Bankers would borrow at three per87

cent, lend at six, and be on the golf court by 3 p.m..1 Since then, however, the financial sector has gone through major transformations. Between 1971 and 1979, exogenous events
– such as two major oil shocks and the exit of the world from Bretton Woods – pushed
the U.S. and the international financial infrastructures to re-think themselves and adapt
to conditions vastly mutating over a relatively short period of time. These events proved
regulatory requirements often inadequate for banks to cope with the new scenarios,2 hit
traditional banks in different ways, and paved the way to embolden an emerging niche of
financial institutions that operated like banks while not being so, i.e., the shadow banking
sector. Also, the previous chapter established the importance of rising inequality from
the 1980s.3 The concomitant external shocks put new non-bank financial intermediaries
de facto in a position of competitive advantage (whether because of regulatory arbitrage
or technological superiority). Therefore, as partly highlighted before, the credit system
gradually moved away from traditional banking activities towards a market-based financial system, while the overall volume of intermediated assets has increased exponentially
since then.4
Given the incremental push for financial liberalization, and the elimination of some
of the most important elements of the previous legal framework, economists have tended
to focus on such freedom as the motive behind the renewed prominence of finance and,
potentially, the occurrence of crises (Kroszner and Strahan, 2014; Igan and Mishra, 2014;
Philippon, 2015; Mian, Sufi, and Verner, 2017; Chu, 2018). This paper tries to describe
from a positive angle the emergence of financial deregulation and innovation stemming
from competition between banks and shadow banks.5
After a series of shocks favored the emergence of (unforeseen) stiff competition in
1
In the academic literature, Koetter, Kolari, and Spierdijk (2012) document statistical evidence of the so-called "enjoy
the quiet life" hypothesis in the banking sector at the time. They build a framework to show that minimum effort –
rather than profit maximization – seems to be the objective function of market players at the time.
2A remarkable example is Regulation Q cap on interests payable on deposits, see Drechsler, Savov, and Schnabl (2020).
3
In this respect, this chapter can be seen as nested into the previous.
4
Similarly, the spectrum of products provided has ballooned.
5Due to the way this work was originally structured, the first part of theoretical framework is more focused on the rise
of financial claims overall. This may be seen now as being supplanted by the analysis described in Chapter 2.
88

the traditional banking sector from the 1980s, non-bank financial institutions were able
to grow faster thanks to their competitive advantage. The rents built in the law that traditional banks enjoyed until the 1970s became increasingly challenged. Such competitive process induced a migration of flows away from traditional depository institutions,
thereby leading to higher pressure on banks margins.
As such, the long 20 years wave of deregulation that lasted from the 1980s to the early
2000s can find a root cause explanation in the incentives that banks were facing in order
to survive. Hence, the rush from banks to lobby policymakers can be rationalized as the
effort to reduce the asymmetries by expanding the scope of operations, and leveling the
playing field. From this perspective, both the deregulation phase and the spurt of financial
innovation in traditional banking could be paired and seen as the by-product of the same
force: Stiffer banking competition. To put it differently, the structural changes in the industrial organization of financial intermediation have created crucial incentives for market
participants to push harder for liberalization, while producing more innovative and engineered instruments to deliver profits and meet investors’ expectations. It is interesting to
notice that in this case the efforts of legislators in the 1930s to create a safe system by preventing banking competition from generating the type of excesses seen before 1929 might
have paradoxically engendered the very conditions for regulatory arbitrage and crises.
In order to rationalize the previous argument, I build a macro-finance dynamic general
equilibrium model with multiple sectors. The modeling exercise allows to make explicit
the effects of different access to the technology frontier and the role of regulatory arbitrage
to express the increasing predominance of shadow banks vs. traditional banks. The model
can be thought of as a nested structural transformation model that features a real sector
that becomes progressively smaller vis-à-vis the financial sector. In turn, in the financial
sector depository institutions shrink in size with respect to other non-bank financial institutions. The model is subsequently extended to allow banks to “react" to such forces by
investing in lobbying activities and the creating of equally innovative products. The main
89

forces are supported both in terms of stylized facts, and in terms of time series analysis.
Related literature. This work stands at the intersection of different strands of literature,
such as (excessive) liquidity creation, IO of banking, or financial innovation.
Previous works all shed new light on the importance of financial contracts and private
credit creation to explain credit booms, crises, or effectiveness of monetary policy (Lorenzoni, 2008; Luck and Schempp, 2014; Sunderam, 2015; Li, 2017; Dávila and Korinek, 2018;
Piazzesi and Schneider, 2018; Benigno and Robatto, 2019).
The competitive environment in banking is certainly a crucial aspect to model. A
number of new macro-economic implications have been drawn from dynamic models of
banking: Corbae and D’Erasmo (2013) and Corbae and D’Erasmo (2021) have theorized
an oligopolistic banking structure in which incumbents are challenged by a competitive
fringe, which results in an endogenous distribution of financial actors. Diallo and Koch
(2018) have stressed the importance of bank concentration in a model of growth by using
a Cournot model of banking competition. Faia, Laffitte, and Ottaviano (2019) construct a
model with imperfectly competitive (global) banks in which the effect of competition on
risk-taking is ambiguous because – on the one hand – there is a scale effect: increase in
deposits raised and loans extended; on the other hand – a larger asset size may imply a
higher vulnerability to bank runs on the liability side. Shin (2009) is the only paper I am
aware that tried to push the idea that credit creation is endogenous, and that an imperative asset expansion can lead to financial fragility. Nonetheless, the reason to explain why
such imperative expansion exists is lacking, and competitive dynamics are only based on
Value-at-Risk considerations. Thakor (2012), on the other hand, is the most similar for its
view of competition as an important driver for innovation although meant to generate endogenous crises as a result of disagreement among investors about the value of new loans
being issued.
Alternatively, financial intermediaries have been modeled to: accommodate higher
90

demand for new products and services from entrepreneurs, or exploit behavioral biases
and/or regulatory arbitrage (Merton, 1995; Acharya et al., 2013, among others). Herrera
and Schroth (2003), instead, provide an interesting account of innovation in securities
markets by developing a corporate finance environment with Bayesian learning. In this
case, costly innovation can be justified by the first-mover advantage of the innovators even
though copying is costless, the learning curve provides the incentives to stay ahead of the
process and keep innovation incentives valid. Awrey (2013), instead, provides a narrative
on the need to study the supply-side incentives embedded in the competitive structure of
financial sectors to deepen the understanding of financial innovation.
I am not aware of studies trying to address the joint trends of financial-deepening,
capital-deepening, structural transformation of banking for regulatory motives, and increase in deregulation itself as the by-product of such phenomenon.
The paper is structured as follows. Section 3.2 motivates the study and corroborates
the previous narrative by means of descriptive statistics. Section 3.3 provides empirical
evidence to support the relevance of the claims. In Section 3.4, I build and explain my
model and its extensions. Section 3.5 concludes, and draws trajectories on how to expand
on the current work for future research avenues. Further proofs, analyses and robustness
checks are available in the Appendix.
3.2 Stylized facts
Figure 3.1 plots the ratio of total depository institutions assets as a fraction of non-bank
financial institutions’ loans. It is noticeable a drastic change in the structural path after the
1970s. Shadow banks began to earn a growing and systemic role in the financial sector
thereby eroding the position that banks used to have. More interestingly, such trend did
not stop at some random point in time but at the end of the 1990s when the Gramm-LeachBliley, also known as the Financial Services Modernization Act, of 1999 was passed. This
91

date completed a gradual and continued twenty years process of financial liberalization,
which had begun between the end of the 1970s and the beginning of the 1980s when interest rate ceilings were levied nation-wide allowing banks to adjust their strategies. The
pattern is suggestive with respect to the previous narrative. Banks managed to grow at
the same rate as shadow banks (constant ratio) only after deregulation was completed
and depository institutions were allowed to re-gain a competitive edge with respect to
non-banks. The collateral effect of such practices was the excess credit creation that inflated the mortgage bubble of 2008. It is worth mentioning that the rise of banks’ assets
per se has also been quite stark. Over the period 1960 - 2020, assets grew by a factor of
8.64 in real terms, while real output grew by 4.87 times. However, non-bank financial institutions rose by more than 33 times in real terms over the same period of the time; also
the massive increase happened not from 1960 to 1980 but in the following decades. The
combination of depository and shadow banks was about the same size of real output in
1960, sixty years later the relationship between the two had grown to more than 4:1.
In the plot on the right side of Figure 3.2, it is possible to further assess the loss of dominant position in the funding markets with respect to the behavior of deregulation. The
competition on the funding side, as represented by the ratio of U.S. chartered institutions
deposits vs. money market funds’ shares, mimics what seen in terms of overall total assets although slightly more volatile due to the different type of data.6 While the amount
of deposits represented an amount 12 times as big as the competitors in 1983, it shrank to
just about 2 times as much before the 2008 crisis. This feature created increasing burdens
for banks in the way they had to attract the major fuel for their operations, and eventually
lead to shift themselves towards more debt instruments and so-called non-core funding
(Barattieri, Moretti, and Quadrini, 2021). Similar to what has been seen on the asset side,
also on the liability side a change in the pattern has emerged since the Great Recession.
6Deposits were virtually the only form of financing for banks after WWII, and since then the fall has been dramatic. I
cut the series of the ratio for the first years because MMMF were not even existent until 1974 and therefore the very
small denominator would not allow to see the major dynamics happening in a continuous fashion all the way until the
2008.
92

Figure 3.1: Traditional bank assets relative to the total amount of assets issued by other non-bank
financial institutions in the United States over the period 1960q1 – 2019q4
Notes: Banks assets stand for the total on-balance sheet loans of U.S. chartered depository institutions. Other financial
institutions assets is obtained as the sum of all the major non-bank financial institutions as per classification of the
Financial Stability Board. See Table C.3 in the chapter’s appendix. All data are at quarterly frequency, and deflated
by GDP implicit price deflator.
Deposits amounts seems to have gained new traction doubling their share with respect to
money market instruments in less than ten years.
This higher degree of competition seems to systematically anticipate the successive
deregulation waves picked by the index built by Philippon and Reshef (2012), and extended to 2010.7 This feature speaks to the ability of banks to cope with other financial
institutions once the regulatory arbitrage got extremely narrow. The deregulation here is
only expressed in terms of passed legislation, and not actually implementation therefore
giving a better sense on the actual moments in which laws (like the Glass-Steagall Act,
above all) were amended.
7
If it were possible to extend it further, it would certainly go up again as a result of the new regulation put in place by
the Obama administration from 2010q3 on with the Dodd-Frank Act of July 2010.
93

Figure 3.2: Banks off-balance sheet assets vis-à-vis non-bank financial institution assets (left),
and funding competition vis-à-vis deregulation index (right)
Notes: The measure of off-balance item is composed of the sum of total unused commitments (Revolving open-end
lines secured by 1-4 family residential property, credit card lines, construction loan commitments, other unused commitments), and letters of credit. Non-bank financial institution assets are the sum of total assets (end of period) for:
Pensions funds, insurance companies, other financial Business. Monetary authority’s assets are always excluded. Data
are normalized to 100 in 1990. The deregulation index from Philippon (2015) is a weighted average of four components:
interest rate ceilings, possibility to have multiple branches, possibility to combine commercial and investment banking,
possibility to combine insurance and financial activities. Chartered deposits is the sum of: U.S.-Chartered Depository
Institutions checkable, savings and time deposits in levels. MMMF total shares are Money Market Funds total shares
outstanding (liabilities) in levels. All data are deflated using CPI (total all items). Data is reported at annual frequency.
Sources: Off-balance sheet items form Enhanced Financial Accounts of the Federal Reserve Board. Non-bank financial
institution assets are from Integrated Macroeconomic Accounts of the Fed, S.63.a S.65.a tables. Deregulation index is
from Thomas Philippon’s website. Chartered deposits and MMMF series are from tables L.111 and L.121 of the Financial
Accounts of the Federal Reserve. CPI from OECD data retrieved through FRED.
In the same picture, it is also possible to see on the left hand side that banks have
massively increased the amount of off-balance sheet operations to stay ahead of competition. The picture takes advantage of consolidated off-balance sheet operations, and shows
that banks made great use of these items. The off-balance sheet items are composed of five
main groups: unused commitments, letters of credit, credit derivatives (e.g. credit default
swaps), interest rate derivatives (e.g. interest rate swaps), and other derivatives. The latter
three categories are the ones typically associated with risk-sharing arguments and hedging strategies, and thus excluded.8 The former two, however, include: credit card lines,
8The system as a whole works under something similar to matched books in notional amounts, therefore – as well known
in the literature – it is impossible to tell the amount of speculative vs. hedging positions, especially from aggregate
statistics.
94

construction loan commitments, etc and are crucial for liquidity provision. The special
status of these items can be understood in light of the fact that they do not occupy space
on banks’ balance sheets, thus relaxing de facto the regulatory constraints, and allowing
to expand the revenue stream without engaging in extra reserves or equity allocations.9
Such liquidity provision component grew at even faster rates than the overall growth of
non-bank financial institutions. Data are available in a harmonized fashion from 1990 onward, when the overall outstanding amount was about $2 Tn in notional amount. In the
following seventeen years, however, the growth was exponential and – after having deflated the series – the amount was more that four time as big, outpacing even the amount
of liquidity produced by shadow banks by a third.
To conclude, there seems to be descriptive evidence in favor of the idea that, by being more constrained on the traditional side, banks took advantage of innovative ways to
cope with the rise of other intermediaries. Sometimes by pushing for more deregulation,
sometimes by acting themselves in innovative ways sidestepping the main provisions of
regulation. Although such results can be only taken on a correlational level, they appear
rather suggestive of the timing of events that could help explain the evolution, and if anything do not dismiss the proposed explanation: Deregulation has systematically followed
the decrease in traditional banks assets, and once the deregulation waves were completed
such trend stopped. Furthermore, banks ended up massively shifting their operations offbalance sheet as a way to save on costly reserves and collateral. The next session assess
the quantitative implication of such chronological structure in a more detailed fashion.
3.3 Empirical assessment
In this section, I test empirically the narrative just mentioned. Although I cannot account for a precise identification strategy, I propose to begin by assessing the phenomenon
through a lag-lead type of structure, which builds on arguments of weak exogeneity and
9Berger and Bouwman (2015) for a thorough discussion.
95

takes advantage of insights from Granger non-causality tests. I first look at the effect of
competition from non-bank financial intermediaries on the liability side (given its importance for banks revenues).10 I proxy such competition from other wholesale funding
institutions as the proportion of chartered deposits to money market funds shares (the direct competitor, although other money market instruments as Asset Backed Commercial
Papers, and other repurchase agreements – RePos – could be added). Shadow banking
only works through money market funding of capital market lending, and – as such – the
size of money markets is an instrument to capture the overall growth of shadow banking. Deposits, on the other hand, are a distinctive feature of banks, and still represent the
overwhelming majority of banks funding as of 2020 (beyond 80 per cent).
In Table 3.1, I evaluate the effect of a percentage change in the funding competition
index vis-à-vis the subsequent change in deregulation.11 I lag backward the independent
variable to an arbitrary number of periods in order to discover the time length necessary
to generate a change in deregulation. I also introduce an autoregressive component and
additional controls, so that a full-fledged ARDL model is considered, and the long-run
effects from partial adjustment can be evaluated, as expressed in Equation (3.1).
∆Dt+4 = α + β∆xt−i + θ∆Dt + γ
0Xt + t i = 0, 1, . . . 8; (3.1)
Stiffer competition seems to deliver a stronger push for laxer regulation already four
quarters ahead, and reach a peak in intensity and significance the following quarter.12 The
indicator keeps being significant for up to two years and ceases in intensity after that (additional lags reflect the non-significant results provided in the last column). The long-run
effect is also systematically important, and follows the same pattern over time corroborat10The same kind of analysis can be carried out for the lending amounts on the assets’ side, which leads to similar results.
11The effect of deregulation is taken as a year-to-year change rather than quarter-to-quarter given its slow moving nature.
The index is non-serially correlated after taking the absolute difference; I correct for autocorrelation following NeweyWest.
12Table C.2 in the Appendix shows that such results are robust to introduction of a relatively large number of macrofinancial control variables.
96

ing what was the thirty years evidence highlighted before. Notice that none of the control variables here nor in the appendix appears to be systematically significant to explain
deregulation.
With that being said, one needs to be careful about reverse causality, namely, to what
extent an increase in deregulation did not actually favor shadow banks and increase competition. Equation (3.2) deals with this issue, and Table 3.2 shows the results for this
channel.
∆xt = α + β1∆Dt−i + β2∆Dt−i−1 + θ∆xt−1 + γ
0Xt + t i = 1, 2 . . . 22; (3.2)
I compress the effect of two subsequent quarters together to convey the message in one
table, but results are always significant also when considering one quarter in isolation. As
shown in this case, for the first two years after liberalization measures were passed, the
impact on banks and shadow banks does not seem to be statistically significant. This is
probably also due to the fact that deregulation takes time to be actually implemented since
the moment it is passed. After two years, however, the coefficient starts to be significant
and negative in sign – and it does so for the following three years (lags after that replicate the non-significant coefficients in Column 9). Furthermore, notice that coefficients
are negative, i.e. higher deregulation does benefit banks more than shadow banks, and
allows them to grow stronger than other financial intermediaries. The effects are robust
to different specifications, and the long-run effects mimic what observed in the short run.
With that in hand, I proceed to carry out a similar type of analysis by means of local
projections (Jordà, 2005).
∆Dt+h = α + β∆xt + θ∆Dt+h−4 + γ
0Xt + t h = 1, 2, . . . 20; (3.3)
As shown in Equation (3.3), the opposite type of lag structure is assessed. The independent variables are kept fixed (except for the autoregressive component), and the
97

dependent variables are forecast in the future. As such, the expression in Equation (3.3)
can be thought of as following the similar idea as the one in Equation (3.1). Figure 3.3
plots the impulse response functions derived for this case. It is interesting to notice that,
not only the pattern of the impulse response follows the one shown of the coefficients
of the ARDL analysis, but also that the IRF follows almost an ’M’ with a first wave being
eight quarters ahead (i.e. two years), and the following sixteen (i.e. four years). Although
purely suggestive, it is interesting to notice how these two horizons coincide with midterm
elections periods, and a full mandate in the U.S. political cycle.
The bottom panel of Figure 3.3 is the by-product of the impulse response from Equation
3.4, and it replicates the same idea expressed in Equation 3.2.
∆xt+h = α + β∆Dt−1 + θ∆xt+h−1 + γ
0Xt + th = 1, 2, . . . 23; (3.4)
Again, we can see that for the first two years the effects are rather noisy and nonsignificant before dipping into negative for the following three years. After five years
from the implementation of deregulation, the coefficients cease to matter for the benefits
of banks vis-à-vis non-banks financial institutions.
One final point of interest is the extent to which funding competition also has an impact on- the off-balance sheet operations. Equation (3.5) depicts the set of regressions
conducted in Table 3.3. Ot here refers to the ratio of off-balance sheet assets as a fraction
of the overall consolidated assets.13
∆Ot = α + β∆xt−2 + θ∆Ot−1 + Xtγ + t (3.5)
As aforementioned, off-balance operations are utilized to save on regulatory requirements imposed on banks, and are associated with entering into financially innovative con13Off-balance sheet operations are harmonized on a balance-sheet consolidate fashion, and therefore need to be accordingly evaluated with respect to the overall size of assets, as specifically indicated by the Federal Reserve Board
explanation of the Enhanced Financial Accounts.
98

Table 3.1: Quarterly estimates of the effect of funding competition on deregulation over the period 1980q1-2008q4
Deregulation index (∆Dt+4)
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Funding competition 0.782* 0.840** 0.781* 0.765** 0.619** 0.518** 0.364* 0.357** 0.328
(∆xt−i) (0.419) (0.410) (0.424) (0.361) (0.311) (0.253) (0.199) (0.179) (0.213)
Fund competition – LT effect 1.196* 1.253** 1.160* 1.119** 0.902** 0.753** 0.531* 0.519** 0.480
β/(1 − θ) (0.648) (0.621) (0.628) (0.526) (0.443) (0.355) (0.286) (0.255) (0.305)
Lag=0 (∆xt) yes
Lag=1 (∆xt−1) yes
Lag=2 (∆xt−2) yes
Lag=3 (∆xt−3) yes
Lag=4 (∆xt−4) yes
Lag=5 (∆xt−5) yes
Lag=6 (∆xt−6) yes
Lag=7 (∆xt−7) yes
Lag=8 (∆xt−8) yes
Deregulation index (∆Dt) 0.346*** 0.329*** 0.327*** 0.316*** 0.314*** 0.311*** 0.315*** 0.313*** 0.317***
(0.088) (0.085) (0.084) (0.077) (0.080) (0.080) (0.082) (0.081) (0.082)
Output growth (∆yt) -0.232 2.452 1.144 0.642 1.859 -0.424 -0.703 -1.465 -0.907
(3.386) (3.532) (3.627) (3.102) (3.700) (3.145) (3.016) (2.727) (2.939)
Inflation (∆pt) -0.805 -0.010 -0.518 -0.919 -0.463 -0.522 0.845 0.972 1.170
(5.350) (5.140) (5.463) (5.232) (5.025) (4.647) (4.739) (4.574) (4.507)
U.S. real equity price (∆eqUS t ) 0.630* 0.433 0.490* 0.560* 0.417 0.548* 0.521* 0.570* 0.556*
(0.342) (0.267) (0.288) (0.298) (0.272) (0.292) (0.298) (0.316) (0.314)
U.S. long term i rate (∆lrUS t ) 29.859* 20.673 22.946 28.822 18.996 30.520* 26.152 27.551 22.021
(17.195) (15.768) (15.456) (17.510) (15.630) (17.317) (16.563) (17.130) (16.673)
R-squared 0.235 0.240 0.240 0.246 0.229 0.223 0.206 0.207 0.204
Notes: ‘Funding competition’ is the ratio between U.S. chartered depository institutions checking, savings, and time deposits and the total amount of outstanding
MMMF shares. The deregulation index is taken as absolute difference with respect to four quarters before. Long term interest rates are quarterly absolute first
difference. All other variables are first difference of the natural logarithm one quarter before. The deregulation quarterly variables are obtained by linearly interpolating the original annual indicator. The long term effect of funding competition is computed from the specification of an ARDL with partial adjustment. Errors
are corrected following Newey-West. All data are quarterly. Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Sources: U.S. deposits and MMMF
shares: Financial Account of the Fed, Z.1 tables (quarterly frequency). Deregulation index: Philippon and Reshef (2012). Output growth: U.S. Bureau of Economic
Analysis. CPI: OECD accessed through FRED. U.S. real equity price, and long term i rate: GVAR dataset, Mohaddes and Raissi (2020).
99

Table 3.2: Quarterly estimates of the effect of deregulation on funding competition over the period 1980q1-2008q4
Funding competition (∆xt)
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Deregulation index -0.035 -0.062** -0.068*** -0.051** -0.050** -0.061*** -0.063*** -0.042** -0.022
(β1 + β2) (0.030) (0.026) (0.022) (0.020) (0.020) (0.019) (0.021) (0.019) (0.016)
Deregulation – LT effect -0.039 -0.067** -0.074*** -0.056*** -0.055*** -0.066*** -0.068*** -0.045** -0.024
(β1 + β2)/(1 − θ) (0.033) (0.027) (0.020) (0.020) (0.019) (0.016) (0.017) (0.018) (0.016)
Lag=6, 7 (∆Dt−6, ∆Dt−7) yes
Lag=8, 9 (∆Dt−8, ∆Dt−9) yes
Lag=10, 11 (∆Dt−10, ∆Dt−11) yes
Lag=12, 13 (∆Dt−12, ∆Dt−13) yes
Lag=14, 15 (∆Dt−14, ∆Dt−15) yes
Lag=16, 17 (∆Dt−16, ∆Dt−17) yes
Lag=18, 19 (∆Dt−18,∆Dt−19) yes
Lag=20, 21 (∆Dt−20, ∆Dt−21) yes
Lag=22, 23 (∆Dt−22, ∆Dt−23) yes
Funding competition (∆xt−1) 0.113 0.081 0.082 0.088 0.088 0.072 0.072 0.086 0.103
(0.117) (0.118) (0.118) (0.119) (0.115) (0.116) (0.117) (0.119) (0.121)
Output growth -1.099 -0.995 -0.976 -1.124 -1.239 -1.239 -1.203 -1.198 -1.235
(0.969) (0.983) (0.977) (0.982) (0.931) (0.918) (0.918) (0.939) (0.948)
Inflation 4.051** 4.287** 4.261** 4.266** 4.086** 4.233** 4.266** 4.262** 4.227**
(1.867) (1.835) (1.794) (1.871) (1.877) (1.876) (1.859) (1.873) (1.887)
US real equity price -0.119 -0.151 -0.153 -0.131 -0.081 -0.093 -0.101 -0.106 -0.099
(0.128) (0.127) (0.129) (0.137) (0.129) (0.128) (0.123) (0.126) (0.127)
US long term i rate -4.796 -6.172 -6.330 -5.750 -4.176 -4.400 -4.766 -4.692 -4.322
(5.874) (6.226) (6.081) (6.395) (6.313) (6.289) (6.304) (6.244) (6.148)
R-squared 0.287 0.295 0.311 0.296 0.288 0.298 0.305 0.288 0.276
Notes: ‘Funding competition’ is the ratio between U.S. chartered depository institutions checking, savings, and time deposits and the total amount of outstanding
MMMF shares. The deregulation index is taken as absolute difference with respect to four quarters before. Long term interest rates are quarterly absolute first
difference. All other variables are first difference of the natural logarithm one quarter before. The deregulation quarterly variables are obtained by linearly interpolating the original annual indicator. The long term effect of funding competition is computed from the specification of an ARDL with partial adjustment. Errors
are corrected following Newey-West. All data are quarterly. Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Sources: U.S. deposits and MMMF
shares: Financial Account of the Fed, Z.1 tables (quarterly frequency). Deregulation index: Philippon and Reshef (2012). Output growth: U.S. Bureau of Economic
Analysis. CPI: OECD accessed through FRED. U.S. real equity price, and long term i rate: GVAR dataset, Mohaddes and Raissi (2020).
100

Figure 3.3: Local projections over the period 1980q1–2008q4
Notes: Top: One standard deviation shock in the funding competition index. Bottom: One standard deviation shock
in the deregulation index. ‘Funding competition’ is the ratio between U.S. chartered depository institutions checking,
savings, and time deposits and the total amount of outstanding MMMF shares. The deregulation index is taken as
absolute difference with respect to four quarters before. Long term interest rates are quarterly absolute first difference.
All other variables are first difference of the natural logarithm one quarter before. The deregulation quarterly variables
are obtained by linearly interpolating the original annual indicator. The long term effect of funding competition is
computed from the specification of an ARDL with partial adjustment. Errors are corrected following Newey-West.
All data are quarterly. Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Sources: U.S. deposits and
MMMF shares: Financial Account of the Fed, Z.1 tables (quarterly frequency). Deregulation index: Philippon and
Reshef (2012). Output growth: U.S. Bureau of Economic Analysis. CPI: OECD accessed through FRED. U.S. real
equity price, and long term i rate: GVAR dataset, Mohaddes and Raissi (2020).
101

Table 3.3: Effect of funding competition on off-balance items as a fraction of total assets. Sample period 1990q1-2007q4
Off-balance sheet ratio (∆Ot)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Funding competition 0.162** 0.163** 0.149** 0.152** 0.162* 0.166** 0.166** 0.152** 0.161** 0.164*
(∆xt−2) (0.072) (0.076) (0.065) (0.074) (0.092) (0.073) (0.076) (0.065) (0.079) (0.094)
Funding competition – LT effect 0.194*** 0.194*** 0.179*** 0.182** 0.188** 0.197*** 0.197*** 0.183*** 0.191** 0.189**
β/(1 − θ) 0.0728 0.0744 0.0668 0.0744 0.0910 0.0735 0.0748 0.0675 0.0781 0.0928
Off-balance sheet ratio (∆Ot−1) 0.161 0.161 0.170** 0.165* 0.136 0.158 0.158 0.166** 0.154* 0.132
(0.107) (0.118) (0.073) (0.084) (0.084) (0.107) (0.118) (0.071) (0.085) (0.087)
Output growth (∆yt) 0.018 0.098 0.217 0.601 -0.000 0.085 0.246 0.615
(0.602) (0.493) (0.629) (0.647) (0.605) (0.496) (0.632) (0.667)
Inflation (∆pt) 0.445 0.474 0.593 0.485 0.554 0.634
(0.914) (0.953) (0.852) (0.933) (0.977) (0.910)
Oil price (∆poil t ) -0.063*** -0.063** -0.064** -0.063*** -0.066** -0.065**
(0.021) (0.025) (0.025) (0.022) (0.025) (0.025)
US real equity price (∆eqUS t ) -0.014 -0.080 -0.026 -0.082
(0.057) (0.091) (0.062) (0.093)
US long term i rate (∆lrUS t ) -1.158 -4.880 -1.290 -4.998
(3.077) (7.144) (3.120) (7.221)
Global realized volatility (grvt) -0.265*** -0.258***
(0.096) (0.095)
Global real equity price (∆eqUS t ) 0.014 0.014
(0.099) (0.100)
Global long term i rate (∆glrt) 0.041 0.044
(0.188) (0.189)
Global output growth (∆y∗t ) -1.023 -1.038
(0.709) (0.726)
Deregulation index (∆Dt−2 ) -0.003 -0.003 -0.004 -0.006 -0.002
(0.006) (0.006) (0.005) (0.006) (0.006)
R-squared 0.143 0.143 0.232 0.235 0.302 0.145 0.145 0.234 0.238 0.303
Notes: ‘Off-balance sheet ratio’ is the fraction of Unused commitments and letters of credit out of consolidated U.S. depository institutions’ assets. ‘Funding
competition’ is the ratio between U.S. chartered depository institutions checking, savings, and time deposits and the total amount of outstanding MMMF shares.
The deregulation index is taken as absolute difference with respect to four quarters before. Long term interest rates are quarterly absolute first difference. All
other variables are first difference of the natural logarithm one quarter before. The deregulation quarterly variables are obtained by linearly interpolating the
original annual indicator. The long term effect of funding competition is computed from the specification of an ARDL with partial adjustment. Errors are corrected
following Newey-West. All data are quarterly. Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Sources: Off-balance sheet items, and consolidated
assets: Enhanced Financial Accounts, Federal Reserve Board. U.S. deposits and MMMF shares: Financial Account of the Fed, Z.1 tables (quarterly frequency).
Deregulation index: Philippon and Reshef (2012). Output growth: U.S. Bureau of Economic Analysis. CPI: OECD accessed through FRED. U.S. real equity price,
and long term i rate: GVAR dataset, Mohaddes and Raissi (2020).
102

tracts in the literature. Thus, I use this as a proxy to estimate the likelihood of seeing
a higher push for securitization and financial engineering, as a result of stiffer competition. The previous ARDL structure is maintained by adding a number of controls and an
autoregressive component which parses out additional autocorrelation left after having
taken the ratios in first difference and having corrected for Newey-West standard errors.
As shown in Table 3.3, the effects are systematically positive and significant regardless
of the choice of controls used. Furthermore, the coefficients are highly stable both in the
short- and long-run. A percentage point of increase in MMMF with respect to deposits
leads to 0.16 per cent growth of off-balance sheet operations with respect to the overall
asset growth. Once again, the results lean in the direction that these channels may be selectively used over time according to the market environment. Other controls seems to be
non-significant. One important remark is that the time lag of the explanatory variable xt
is rather important here. The effects seem to be not significant before and after that. However, this can be thought as a reasonable reaction time, the first quarter may be needed to
actually assess the stress, while it may need no more than between three and six months
to respond to heightened competitive pressure.
To conclude, the empirical evidence, although not accounting for a full-fledged exogenous structure, remarks the overall stream of reasoning and descriptive evidence highlighted in the previous two sections. Namely, higher deregulation is systematically observed between two and three years of increased pressure on banks, such deregulation
seems to make banks grow larger than shadow banks with a lag comprised between two
and five years, and in the short term banks push their assets off-balance sheet with a lag
of about six months as a result of higher shadow banks growth – in this case, deregulation
is not important at any horizon considered.
103

3.4 Theoretical framework
I proceed to build a general equilibrium macro-finance model that can rationalize the
patterns seen before. I propose a nested structural model in which there is both capitaland financial-deepening. Capital deepening represents an increase in the capital-output
ratio, which appears to be in line with data for the last forty years. Financial-deepening, on
the other hand, remarks the original goal of the paper – which was to explain an increase
in the share of finance over output. At the same time, the model allows the financial sector
to go through structural transformation by means of higher growth of the shadow banking sector vis-à-vis the traditional one. The higher growth of the shadow banking system
is going to be pinned down by a larger access to products and lack of reserve requirements
– which, on the other hand, are not features of the traditional banking system. Such difference in terms of access to different regulatory settings leads to a movement of the funds
away from traditional banks to shadow banks. As a consequence, the relative shares of
each sector changes over time. In fact, the baseline model predicts the disappearance of
traditional banks. To address such counterfactual prediction, I build two extensions in
which traditional banks can respond by investing in deregulation and financial innovation (in the form of off-balance sheet operations).
In terms of model environment, the model is set in discrete time, it features no uncertainty, and no population growth. The economy is populated by: A representative household, two representative firms with consumption-specific and capital-specific sector technologies, respectively. The traditional banking sector is populated by heterogeneous retail
branches (which benefit from market power on deposits), and wholesale branches that
aggregate funds obtained from deposits into financial investment products. The shadow
banking sector is similarly composed of a capital market side, which produces investment
products financed on money markets (which are in direct competition with bank deposits
in terms of returns).
104

Households provide labor and physical capital to the first two industries, while investing the remaining part of their income in financial assets. Both real and financial capital
are reproducible. The retail units of banks take on deposits and earn a spread on them
due to their monopsonistic power. They also set aside an amount of reserve proportional
to their size, and pass on the funds at no cost to the wholesale unit – which is able to
repackage the different loans and sell them on the market to the households. The extent
to which these products can be sold is limited by regulation. The wholesale banks earn
no direct spread on manufacturing such claims, although the model can easily be relaxed
along this dimension.
The shadow banking system competes with the traditional banking activities in the following sense. On the liability side, money market funds do not earn a direct spread from
households on their shares, but they are still profitable thanks to the demand of products
coming from the financial firms populating capital markets. Also, they are not compelled
to set aside reserves. Capital markets agents fund themselves on money markets to transform and repackage the loans that are eventually sold as financial investment products
to households. Without loss of generality, this is assumed to happen with no extra markup. Given the highly predominant role of mortgages, consumer credit, auto and student
loans, paired with a limited role of excessive loans to non-financial business, I limit the
amount of outstanding credit by banks to the needs of households. One can generalize it
to a broader set of instruments for non-financial business.
In order to explain the overall dynamics up to the financial crisis, the model abstracts
from risk for now, and it does not investigate the direct effects on financial stability. This
remains a future modification of the model to potentially account for.
3.4.1 Baseline set-up
Households. A representative household maximizes its utility over consumption Ct
, and
discounts its stream of Ct at rate β. It earns a wage wt by inelastically supplying labor, and
105

from three different types of interest income by renting its physical capital, and holding
deposits and money market shares – {RK
t
, RBK
t
, RSB
t
} represent the returns from such
holdings, respectively. Households intertemporal optimization is expressed in Problem
(3.6).
max
{kt+1, FBK,t
FSB,t, Ct}
X∞
t=0
β
t
log(Ct) (3.6)
sub



Ct + p
K
t
IK,t + p
SB
t
ISB,t + p
BK
t
IBK,t = RK
t kt + RSB
t FSB,t + RBK
t FBK,t + wtL
IK,t = kt+1 − (1 − δ)kt
IBK,t = FBK,t+1 − FBK,t
ISB,t = FSB,t+1 − FSB,t
k0 > 0, C0 > 0
Households face customary consumption-savings decisions. Investments in physical
capital are represented by IK,t, the ones in traditional banks products by IBK,t, and the
ones in shadow banking assets as ISB,t. Physical capital is denoted by kt
, and it depreciates at an exogenous rate δ. It gets accumulated by the law of motion expressed by the
second constraint. Assets provided to banks and shadow banks are described by variables
FBK,t, FSB,t, which increase by investing IBK,t, ISB,t in financial assets. Financial capital
does not depreciate. In order to obtain such products, the household pays a tuple of prices
{p
K
t
, pBK
t
, pSB
t }, while the price of consumption goods is normalized to one, and used as
numéraire. The system features three Euler equations, and three no-arbitrage conditions
between assets returns.
Firms. There are two distinct goods on the "real" side of the economy: consumption and
106

capital goods. This side of the economy is highly stylized and replicates models in an AK
fashion in order to obtain endogenous growth in a tractable form. There is a measure one
of risk-neutral consumption good firms which operate under perfect competition. Firms
rent labor and a fraction, φ
K
t
, of physical capital from households, and produce an amount
of final goods Ct
, with a constant Hicks neutral productivity, AC. Factor shares {α, 1 − α}
accrue to real capital and labor, respectively.
max
{(φK
t
k),L}
AC(φ
K
t kt)
αL
1−α
| {z }
,Ct
−R
K
t φ
K
t kt − wtL (3.7)
Investments in physical capital also follow a perfect competition benchmark, by renting
the remaining fraction of capital (1 − φt)kt at rate RK
t
, and transforming it into real investments goods with a production function exhibiting linear technology with AK being the
technology-shifter. The problem is simply posed as:
max
(1−φK
t
)kt
p
K
t AK(1 − φ
K
t
)kt
| {z }
,IK
t
−R
K
t
(1 − φ
K
t
)kt (3.8)
As aforementioned, such characterization is a stripped down version of more complex
settings, and allows for consumption and investment sectors to reach a stationary equilibrium in growth rates rather than in levels while the financial sector will keep expanding.
Proposition 1. The fraction of capital allocated between consumption and physical investment
sector is constant and equal to: φ
∗
K = (1 − β)
AK+1−δ
AK
Proof. See Appendix.
Demand for deposits. When households deposit their income at their retail banks, they
face a monopolistically competitive supply of such services, which allow each bank, n, to
earn a spread. As such, the upward sloping supply curve can be written as stemming
107

from the following problem:
max
{F BK
t
(n)}
Z 1
0
R
BK
t
(n)F
BK
t
(n)dn
s.t. Z 1
0

F
BK
t
(n)
1+ν
ν dn ν
1+ν
≤ F¯BK
t
(3.9)
with ν > 1, being the elasticity parameter customary with the CES aggregator. I am
allowing a measure one of banks to exist, even though a more generic continuum N can
be allowed for. Each household earns a return, RBK
t
(n), on their deposits, F
BK
t
(n), from
bank n. RBK
t
is the price index, and F
BK
t
the overall amount of deposits channeled to
the banking system. The amount of deposits attracted by each bank is a function of their
competitive returns with respect to the weighted average of the industry. The final supply
can be written in the canonical form: F
BK
t
(n) =
R
BK
t
(n)/RBK
t
ν
F
BK
t
.
Traditional banking sector.
— Retail unit. Each bank is allowed to have a maximum number of Z
BK
t branches and/or
profitable products, but there exists a single wholesale unit for each bank. For simplicity,
I assume the fraction of deposits attracted by each bank, F
BK
t
(n), to be composed by a
uniform distribution from each retail unit. Regulation is expressed by the index 1/Dt
,
where Dt > 1. It constrains the size of the banks (or the number of products the bank
can sell). Under this representation, it is possible to write regulation as a productivity
wedge, (1 − 1/Dt), with respect to the financial innovation frontier, Zt
, such that Z
BK
t =
Zt (1 − 1/Dt).
14 For now, Zt
, Dt
, ZBK
t are assumed to be time invariant.
Each retail unit j at bank n transforms a share of deposits f
BK
t
(n, j)into financial claims,
ι
BK
t
(n, j). When doing so, it needs to set aside a fraction St(n, j) = sf BK
t
(n, j) as reserves,
14Zt can be seen as the number of products potentially existent at the frontier.
1

with s ∈ (0, 1). The problem is presented in (3.10):
max
fBK
t
(n,j)
[1 + r
BK
t
(n, j)]ι
BK
t
(n, j) + St(n, j) − [1 + R
BK
t
(n)]f
BK
t
(n, j) (3.10)
s.to



ι
BK
t
(n, j) + St(n, j) = f
BK
t
(n, j)
St = sf BK
t
(n, j) s ∈ (0, 1)
R
BK
t
(n) =
F
BK
t
(n)
F
BK
t
1
ν
R
BK
t
F
BK
t
(n) = Z ZBK
0
f
BK
t
(n, j)dj
The constraints account for: an accounting identity (with exogenous reserve requirements imposed on chartered institutions), a downward sloping demand curve for deposits coming from the households (as described before), “internal market clearing" of
funds received/allocated across units. Assuming symmetry across branches, we can say
f
BK
t =
F
BK
t
ZBK . Loans are subsequently provided to the wholesale unit that “repackages"
them and sells them on the market. The model leads to an optimal pricing condition:
r
BK
t
(n, j) = RBK
t
(n)
1+ν
ν(1−s)
, and profits: π
BK
t
(n, j) = 1−s
1+ν
r
BK
t
(n, j)f
BK
t
(n, j).
— Wholesale unit. The wholesale unit of a traditional bank takes the loans produced by
each bank, and transforms them by bundling together and selling them as new financial
contracts, I
BK
t
(n). The problem is expressed in (3.11).
max
ιBK
t
(n,j)
p
BK
t
(n)I
BK
t
(n) − r
BK
t
(n, j)(1 − s)F
BK
t
(n) (3.11)
109

sub



I
BK
t
(n) = Z ZBK
t
0

ι
BK
t
(n, j)
σ−1
σ dj! σ
σ−1
F
BK
t
(n) = Z ZBK
0
f
BK
t
(n, j)dj
ι
BK
t
(n, j) = (1 − s)f
BK
t
(n, j)
σ > 1 represents the elasticity across units. r
BK
t
(n, j) is the price "charged" to the retail
unit, and it has to be intended as a shadow price (given that the wholesale unit does not
charge an actual price on itself). On the other hand, p
BK
t
is the final price at which the
products are sold to the households. For simplicity, banks are assumed to be perfectly
competitive when selling such products, therefore, all their consolidated profits derive
from deposits.15 From the expression, it is possible to see that a fraction, (1 − s), of overall
funds F
BK
t
(n) is netted out to account for the reserves. It can be showed that:
I
BK
t
(n) = (1 − s)ABKF
BK
t
(n)
with ABK
t ,

Z
BK
t
1
σ−1
. This is an important result because it links the productivity of
the banking sector to its operational capacity: The stronger the regulatory constraint restraining Z
BK
t
, the lower the productivity that the bank will achieve. Furthermore, the
higher the complementarity of contracts and type of loans, the higher the productivity.
This speaks to the efforts of finance to create contracts that could tranche different parts
of, say, risk operations as happened when bonds risks where replicated by different products bearing credit risk, interest risk, etc.
Notice that – in its simplest form without risks – the model speaks only to the "bright
side" of financial innovation. Given that p
BK
t
(n) = rt(n, j)/ABK, we need to conclude that
the higher productivity of the banks translates into cheaper services for households. Such
15This assumption can be easily relaxed in order to have double margins on both loans and deposits.
1

statement does not seem to be true in practice, as highlighted by Philippon (2015).
Shadow banking sector.
— Capital market. The shadow banking system operates in ways that are similar to the
ones of the traditional banking system. I begin by defining the "upstream" part of it, where
capital markets repackage and produce j securities for households by financing themselves
on the (monopolistically competitive) money market funds. These units face regulatory
requirements normalized to zero, therefore they do not have operational capacity limited
by the wedge, 1 − 1/Dt
, as for the banks showed before. Hence, they can work up to the
innovation frontier: Z
SB
t = Zt
.
16 The problem can be written as:
max
ι
SB
t
(j)
p
SB
t
Z ZSB
t
0

ι
SB
t
(j)
σ−1
σ dj! σ
σ−1
| {z }
I
SB
t
−
Z ZSB
t
0
R
SB
t
(j)mSB
t
(j)dj (3.12)
s.t. mSB
t
(j) = ι
SB
t
(j)
where {ι
SB
t
(j)} is the specific set of contracts funded with mSB
t
(j) resources. I assume
σ not to differ between the traditional banking and shadow banking sector to maintain the
focus on the regulatory side. It goes without saying that elasticities can be sector-specific.
As before, it will be true that the productivity will depend on the range of products, ASB
t ,

Z
SB
t
1
σ−1 but given the same σ, the only variable to explain the divergence between the
two sector is opportunity to take advantage of the regulatory arbitrage.
— Money market. When funding themselves on the money markets, the shadow banks
on the capital markets generate a downward sloping demand curve. The money market
funds can earn a profit as a result. To maintain a slightly more parsimonious structure,
money market funds will not be able to earn a monopolistically competitive margin also on
their liabilities. They face perfect competition when offering their products to households.
16Furthermore, as shown below, they will be able to attract a larger fraction of capital in light of the lack of reserve
requirements demanded “downstream" to money market funds.
1

I assume symmetry among all j money market funds. The problem can be described as:
max
f
SB
t
(j)
R
SB
t
(j)mSB
t
(j) − R
SB
t
f
SB
t
(j) (3.13)
sub



mSB
t
(j) = f
SB
t
(j)
R
SB
t
(j) = p
SB
t

mSB
t
(j)
I
SB
t
−1/σ
F
SB
t =
Z ZSB
0
f
SB
t
(j)dj
The optimal profits condition for money market funds is: π
SB
t
(j) = σ
−1
r
SB
t
(j)f
SB
t
(j).
Notice that there is an implicit assumption about funds being symmetric: f
SB
t = F
SB
t
/ZSB.
Figure 3.4 depicts the model dynamics.
Equilibrium. Given initial prices p(0), and economy-wide initial endowments {L, k(0),
F
BK(0), F SB(0)}, a competitive equilibrium is a set of prices:
{p
C
t
, pK
t
, pBK
t
, pSB
t
, RBK
t
, RBK
t
(n), rBK
t
(n, j), RSB
t
, RSB
t
(j), rSB
t
(j), }t∈[0,∞)
, quantities
{Ct
, kt
, F BK
t
, F BK
t
(n), f BK
t
(n, j), ιBK
t
(n, j), F SB
t
, fSB
t
(j), ιSB
t
(j), mSB
t
(j)}t∈[0,∞)
, and sector allocations {φ
K
t
,(1 − φ
K
t
)}t∈[0,∞) such that:
1. Households optimize their consumption and savings/investments decisions according to Problem (3.6).
2. Consumption goods firms maximize profits according to Problem (3.7), by taking
prices as given.
3. Physical capital investment firms maximize profits according to Problem (3.8), by
taking prices as given.
4. Banks, Shadow banks and their product units branches maximize profits by choosing the optimal quantities amount, and by taking reserve constraints as given, according to Problems: (3.10), (3.11), (3.12), and (3.13).
112

5. Markets clear for commodities, labor, capital, banking services, and shadow banking
services.
Figure 3.4: Model baseline description
Solution. The details of the model’s solution and derivations are provided in the appendix.
The first result stemming from the multi-sector growth literature is that capital will
grow at faster rates than real wages and the commodity sector in real terms. This seems
to be systematically true from a large body of previous literature focusing on structural
transformation (Herrendorf, Rogerson, and Valentinyi (2014) for a review). Furthermore
(in an extension not reported here to avoid confusion), it is relatively easy to show that
if commodity firms take on a fraction of banking and shadow banking products to en113

hance production (potentially buying from both financial fringes) then commodity firms
become “financialized" themselves. When confronting this with the reality, it seems that
such a dynamic process is very much ongoing: firms have been sitting on higher and
higher amounts of cash over the past decades and used that to progressively increase their
exposure to financial assets.
Moving to the effects of regulatory arbitrage on the banking sector, it is possible to
show that the following conditions for growth rates hold: gBK = 1 + (1 − s)ABK, gSB =
ASB + 1, gK = β(1 + AK − δ). To the extent that Z
SB > ZBK, as provided by regulatory
burdens, we have ASB > ABK. But then, for ASB > (1 − s)ABK > βAK we can conclude
that gSB > gBK > gK. In other words, growth rates are constant – because of constant
productivity – but unequal.
The economy faces capital-deepening, i.e., the growth rate of consumption is lower
than the one in capital gC = g
α
K. Furthermore the model displays financial-deepening, i.e. an
increase in the share of finance in the economy to conditional on real productivity being
lower than the one in finance – which tends to be true empirically. More importantly,
this is a nested structural transformation that happens while the share of the traditional
banking sector progressively decreases vis-à-vis the shadow banking one. In this respect,
households increase their reliance on shadow banks. Asymptotically, financial-deepening
gets increasingly determined by the growth rate of the funds made available by shadow
banks. To sum the previous result – capital markets become increasingly more important
within the financial world, and the financial world becomes increasingly more important
with respect to the production in the economy.
Growth rates in nominal terms maintain the same properties, although modulated by
the extent of market power. In fact, the larger the profits that can be exploited by the
banking sector, the larger the amplification of growth in nominal terms.
With that being said, the banking system shrunk in relative terms but did not disappear – in fact it stabilized. At this point, I add a political economy element of banking
114

competition. I claim that the ability of the traditional banking sector to survive was by
means of pushing back by: lobbying for deregulation, and restructuring its operations to
become more innovative.
With that in mind, I proceed to model two relevant extensions: in the first case, banks
are able to lever their resources to obtain deregulation with some probability; in the second case, the banks are able to set up a Special Purpose Vehicle (SPV) that is able to get
around regulatory burdens. In other words, the following two extensions reflect potential strategies that banks have utilized in order to remain in the business: innovating and
lobbying to become more similar to shadow banks. As such, financial deregulation can be
seen as the by-product of competitive pressures – which is quite importantly a departure
from the interpretation of the literature.
3.4.2 Extensions
Deregulation. In order to model the process of deregulation, I allow banks to invest
a fraction of their funds in lobbying activities. Banks need to solve two subproblems:
manufacturing financial assets and lobbying. When financing the lobbying industry, I
assume them to make no profits in equilibrium.17 Once deregulation is successful, banks
are able to enjoy a larger share of products to invest into (and a faster growth rate as a
result). Let the original problems of banks (3.10, 3.11) be condensed in one layer without
loss of generality, and let ψ be the fraction of funds spent on lobbying activities by the
bank:
Subproblem 1 (Lending)
max
Dt+1,ψtFBK,t
p
BK
t
(n)ABK(1 − s)ψtF
BK
t − R
BK
t
(n)ψtF
BK
t − p
D
t
(n)Dt+1
17Effectively, I am interpreting the lobbyists as being part of the bank itself.
115

s.to



Dt+1 = (1 + λ)Dt
Upward sloping Supply curve in (3.9)
(3.14)
where ABK = Z
x
(1 − 1/Dt)
x
, x = 1/(σ − 1), and ψt
is the fraction of original funds not
diverted to lobbying. Deregulation follows now a law of motion that is pinned down by
the size of deregulation jumps, λ, that the lobbying unit is able to achieve. The attempts
to deregulate are idiosyncratically uncertain, but deterministic in aggregate, and equal
regardless of the bank succeeding in such attempts. The upward sloping supply curve
follows from the Problem described by (3.9). The profits and a pricing conditions are the
same as in the baseline case up to a rescaling by a factor ψt
.
Further action happens in the second subproblem of lobbying. As aforementioned, I
assume a perfectly competitive lobbying market, which transforms the funds provided by
the banks to deliver a larger degree of deregulation in the following period.
Subproblem 2: (Lobbying)
max
(1−ψt)F BK
t
p
D
t
(n)Dt+1 − R
BK
t
(n)(1 − ψt)F
BK
t
(n)
s.to Dt+1 = ((1 − ψt)F
BK
t
)
αD
(3.15)
where the parameter αD pins down the returns to scale of the lobbying unit. As a
consequence, banks need to equalize the rate of return at the margin that allows them
to allocate the optimal amount ψtF
BK
t of funds between standard loan production, and
lobbying to obtain deregulation Dt
. Figure 3.5 provides a graphical representation of the
new forces.
The optimal condition for ψ is captured by the following expression:
ψ
∗
t =
1
1 + αD
(σ−1)[(1+λ)
t−1]
(3.16)
116

Figure 3.5: Model description: Deregulation extension.
Equation (3.16) has some interesting implications. First of all, limt→∞ ψ
∗
t = 1, i.e. the
higher the number of successful attempts to deregulate (as t grows), the lower the amount
that is spent in such activity because the banks get closer to the technology frontier. Notice
that this feature implies that the share of funds is non-constant even in equilibrium along
the growth path. Regardless of the asymptotic behavior, it can be shown that the growth
rate of ψ in the current framework is extremely slow to the point of being almost proxied
by a constant for arbitrarily large periods of time.
In addition, some comparative statics lessons can be derived. I begin with by noticing
that ∂ψ∗
t
/∂σ > 0. This is a peculiar feature that seems to be true in reality: the more productive the banks become, the higher the amount they can lever to obtain deregulation.
This characterizes a dark side of financial innovation, as it induces a feedback loop from innovation. First, it leads directly to more competitive and faster growing banks; secondly,
117

more power to push for deregulation leads to higher assets growth, which in turn allows
banks to have more funds to deregulate; as such the process of convergence is reduced
in time. If we complement this with the actual dynamics happening in the industrial organization of banking in the U.S., where an increasingly smaller number of banks have
been able to stay in business since the 1980s thus increasing concentration, this seems to
find further support in reality. Thus, if banks are able to increase profits margins either
by reducing competition directly, or making their products more opaque and less substitutable, the higher profits can – in turn – be used to increase deregulation, do more of that,
and eventually close the gap.
Abstracting from market power, the growth rate of deregulation is: gD = g
αD
(1−ψt)
g
αD
BK,t ∝
g
αD
BK,t, where the first term is an extremely slowly decaying growth rate of the expenditures
in lobbying. Differently from before, the growth rate of banks assets is not time-varying
as well. This is the by-product on a progressively higher productivity growth stemming
from higher deregulation, asymptotically: limt→∞ gBK,t = gSB. This is something that can
match the dynamics observed in Figure (C.2). After about twenty years of systematic loss
of market power, banks eventually managed to catch up with shadow banks in terms of
asset growth once the deregulation process was completed at the beginning of the 2000s.
Thus the “double feedback" effects from market power resides in the fact that higher margins, allow for higher deregulation, that in turn leads to higher assets growth, which in
turns speeds up the process of convergence, while increasing the overall level of profits
per period, and enlarging the financial sector even faster at the aggregate level.
With that being said, however, the banking sector has reacted to increasing competition not only by increasing the market power of the remaining subjects, or by pushing for
higher deregulation, but also by adopting different strategies, such as setting up SPVs that
could invest or take advantage of investments in products that could not managed in the
normal operations on balance sheet.
118

Off-Balance Sheet activities In this section I add off-balance sheet operations in the
baseline scenario removing the possibility to also lobby to deregulate. In other words,
I am not allowing for a contemporaneous combination of off-balance sheet operations and
deregulation. Let the bank now invest part of its proceeds, 1 − ξ, in an SPV, which can
generate – de facto – shadow banking activities. In this case, the SPV can invest in all
the Z products available without regulatory constraints of sort. Subsequently, the overall
quantity of financial investments sold to the market is a bundle of products built on- and
off-balance sheet. The bank substitutes one with another according to an elasticity η > 1.
When investing off-balance sheet, the bank does not have to post regulatory reserves in
such entities.18 Figure 3.6 represents graphically this scenario. The problem that the bank
faces can now be re-written as:
max
{ιBK
t
(j),ιX
t
(j)}
p
BK
t
I
BK
t −
Z ZBK
0
r
BK
t
(j)ι
BK
t
(j)dj −
Z Z
0
r
X
t
(j)ι
X
t
(j)dj
sub



I
BK
t =

˜I
BK
t
η−1
η
+

I
X
t
η−1
η
η
η−1
˜I
BK
t =
Z ZBK
0
(ι
BK
t
(j)) σ−1
σ dj σ
σ−1
I
X
t =
Z Z
0
(ι
X
t
(j)) σ−1
σ dj σ
σ−1
Given that a fraction ξ is allocated to the production of the standard sector and
1 − ξ to the production of securities off-balance sheet, we can re-write: f
BK
t
(j) =
ξF BK
t
/ZBK, f
X
t
(j) = (1 − ξ)F
BK
t
/Z.
For each contract off-balance sheet, the bank solves:
max
{fX
t
(j)}
r
X
t
(j)ι
X
t
(j) − R
BK
t
f
X
t
(j)
18I do not allow shadow banks to buy such products although that is the most realistic case, and a further avenue to
expand the current section.
11

Figure 3.6: Model description: Off-balance sheet extension.
sub



ι
X
t
(j) = f
X
t
(j)
r
X
t
(j) = p
BK
t

IX,t
IBK,t− 1
η

ιX,t
IX,t− 1
σ
By combining with the profits that can be generated at single unit level, we have a profit
margin equal to:
π
X
t
(j) = 1 − ξ
σ
R
BK
t
F
BK
t
Z
,
and:
π
BK
t
(j) = ξ
σ
R
BK
t
F
BK
t
ZBK
.
120

ξ is endogenous and can be computed from the optimal pricing conditions:
ξ
∗ =
1
1 + (1 − s)
−η+1
AX,t
ABK,tη−1
=
1
1 + 1
(1−s)
η−1

1 −
1
Dt
1−η
σ−1
(3.17)
There are a number of interesting features that can be noticed from this framework.
First of all, notice that ∂ξ/∂AX,t < 0, ∂ξ/∂ABK,t > 0. This implies that banks attempt to
bring off balance sheet the larger amount possible of loans when the growth rate of shadow
banks is larger than the one of traditional banking activities. The larger the distance between the technology of shadow banks vis-à-vis the one of traditional banks, AX,t/ABK,t,
the larger the incentives to deviate from the current regulatory framework and find loopholes in the system. The same idea can also be seen from a different perspective in the second equation as: ∂ξ/∂Dt > 0. When banks manage to achieve deregulation, the push to
increase such a measure decreases. This also seems to be true in practice. From Figure 3.2,
it is almost possible to visualize that from the beginning of the 2000s, the extent of liquidity
provision through off-balance sheet has not been as strong as it used to be. To conclude the
set of comparative analyses exercises, it is important to notice that reserve requirements
are not sufficient per se to halt the growth of the financial sector, ∂ξ/∂s < 0 for η > 1.
Nevertheless, the most interesting feature of this section can be derived from the new
growth rate of banking activities:
gBK =
h
(ξABK,t)
η−1
η + ((1 − ξ)AX,t)
η−1
η
i η
η−1
+ 1. (3.18)
The reason why the new growth rate is interesting stems from the interplay of ξ and
the financial innovation contracts that underpin ABK and AX. When the regulatory constraints are particularly binding (ABK small), ξ decreases therefore increasing the loading
121

on the off-balance sheet items, which eventually lead gBK to get closer to gSB. On the
other hand, if the deregulation is largely in place, then ξ increases and the same effect
is reached: the wedge decreases and the banking sector starts to be as productive as the
shadow banking one.19
The deeper message of the previous derivations is that regulation by itself does not suffice to prevent an increase in the growth of finance. Neither the reserve requirement nor a
mere restriction of the objects that banks can or cannot use is sufficient to limit the growth
rate of finance. This can only be achieved by limiting the financial contracts and innovations created in the first place. By limiting the entry of financial innovations, it is possible
to limit the growth rate of the sector. This may be seen as a new macro-prudential policy
instrument to be included in the policy toolkit. By treating financial innovation from the
same perspective as pharmaceutical innovation,20 i.e. by setting a system and a procedure
that has to deliberate on the nature of the new contracts brought forward, a more ordered
growth rate of the system can be ensured. The previous derivations become all the more
important in light of the potential systemic risks stemming from a larger financial structure.
3.5 Concluding remarks
In this paper I investigated the overarching phenomenon of explosive growth of financial assets with respect to real output from the 1980s on, which can be defined as financial
deepening.
In this respect, the framework I proposed allowed me to tackle deregulation as a proximate rather than a root cause for the spurt in financial activities. Deregulation has been
identified as the tool to prevent the disappearance of depository institutions when faced
with stiff business stealing. As a consequence, I provided the first attempt, to my knowl19These results become even stronger if we allow the number of financial products to evolve over time. See Fostel and
Geanakoplos (2012).
20A similar point is made in Haliassos (2013).
122

edge, to characterize deregulation neither as an exogenous given evidence, nor as a behavioral process à la Minsky.
Analyses on the time series relationships among relevant variables seem to be conducive to the same story. Waves of financial liberalization have systematically followed
shrinking positions by the traditional chartered institutions even when they looked closer
to catch up the pace of shadow banks. Furthermore, such shrinking positions systematically pushed banks harder towards off-balance sheet operations.
The model, although tractable, allowed for a fairly rich set of dynamics, and helped to
rationalize and describe all the major trends characterizing the economy. When looking
at the overall long-run dynamics, the most important take-away from a macro-prudential
perspective is to pay attention not only to reserve requirements and regulatory constraints
for chartered entities but the extent to which financial innovations are allowed to be introduced. As such, I suggest to analyze the potentially pervasive effects stemming from the
introduction of new instruments, and I propose to shift macro-prudential analyses away
from a good/bad type of debate on shadow banking, while re-focusing on the micro- and
macro-implications of financial engineering.
Many more extensions are possible to enrich the current framework both from a theoretical and an empirical point of view, e.g. on the one hand, the model might be allowed
to incorporate a direct interconnection between banks and shadow banks through interbanking market (which is currently outside of the picture, yet highly relevant in practice),
while on the empirical side a more refined and identified cross-sectional event study can
further corroborate the multivariate and long-run reduced form approach followed so far.
A calibration exercise also remains a must to quantitatively estimate the dynamics taken
from a positive side.
123

Bibliography
Abadie, A., A. Diamond, and J. Hainmueller (2010). Synthetic Control Methods for Comparative Case Studies: Estimating the Effect of California’s Tobacco Control Program.
Journal of the American Statistical Association 105, 493–505.
Abadie, A. and J. Gardeazabal (2003). The Economic Costs of Conflict: A Case Study of
the Basque Country. American Economic Review 93, 113–132.
Acemoglu, D. and P. Restrepo (2022). Tasks, Automation, and the Rise in US Wage Inequality. Econometrica 90(5), 1973–2016.
Acharya, S. and K. Dogra (2022). The Side Effects of Safe Asset Creation. Journal of the
European Economic Association 20(2), 581–625.
Acharya, V. V., P. Schnabl, and G. Suarez (2013). Securitization without risk transfer. Journal of Financial Economics 107(3), 515–536.
Amuzegar, J. (1997a). Adjusting to Sanctions. Foreign Affairs 76, 31–41.
Amuzegar, J. (1997b). Iran’s economy and the US sanctions. Middle East Journal 51, 185–
199.
Andreas, P. (2005). Criminalizing Consequences of Sanctions: Embargo Busting and Its
Legacy. International Studies Quarterly 49, 335–360.
Angeletos, G.-M. (2007). Uninsured idiosyncratic investment risk and aggregate saving.
Review of Economic Dynamics 10(1), 1–30.
Arcand, J. L., E. Berkes, and U. Panizza (2015). Too much finance? Journal of Economic
Growth 20(2), 105–148.
Auclert, A. and M. Rognlie (2017). Aggregate Demand and the Top 1 Percent . American
Economic Review: Papers & Proceedings 107(5), 588–592.
Auclert, A. and M. Rognlie (2020). Inequality and Aggregate Demand . NBER Working
Paper No. 24280.
Auten, G. and D. Splinter (2024). Income Inequality in the United States: Using Tax Data
to Measure Long-Term Trends. Journal of Political Economy, forthcoming.
124

Autor, D. H., D. Dorn, and G. H. Hanson (2016). The China Shock: Learning from LaborMarket Adjustment to Large Changes in Trade. Annual Review of Economics 8, 205–240.
Awrey, D. (2013). Toward a supply-side theory of financial innovation. Journal of Comparative Economics 41(2), 401–419.
Azzimonti, M., E. de Francisco, and V. Quadrini (2014). Financial Globalization, Inequality, and the Rising Public Debt. American Economic Review 104(8), 2267–2302.
Bahmani-Oskooee, M. (1996). The Black Market Exchange Rate and Demand for Money
in Iran. Journal of Macroeconomics 18, 171–176.
Baker, S. R., N. Bloom, and S. J. Davis (2016). Measuring economic policy uncertainty.
Quarterly Journal of Economics 131, 1593–1636.
Barattieri, A., L. Moretti, and V. Quadrini (2021). Banks funding, leverage, and investment.
Journal of Financial Economics.
Barro, R. J., J. Fernández-Villaverde, O. Levintal, and A. Mollerus (2022). Safe Assets.
Economic Journal 132(646), 2075–2100.
Baumeister, C. and J. D. Hamilton (2015). Sign Restrictions, Structural Vector Autoregressions, and Useful Prior Information. Econometrica 83, 1963–1999.
Benhabib, J. and A. Bisin (2018). Skewed Wealth Distributions: Theory and Empirics.
Journal of Economic Literature 56(4), 1261–91.
Benigno, P. and R. Robatto (2019). Private money creation, liquidity crises, and government interventions. Journal of Monetary Economics 106, 42–58.
Berger, A. N. and C. Bouwman (2015). Bank liquidity creation and financial crises. Oxford
and San Diego: Academic Press.
Bonato, L. (2008). Money and Inflation in the Islamic Republic of Iran. Review of Middle
East Economics and Finance 4, Article 3.
Borszik, O. (2016). International sanctions against Iran and Tehran’s responses: political
effects on the targeted regime. Contemporary Politics 22, 20–39.
Brock, W. A. and L. J. Mirman (1972). Optimal Economic Growth and Uncertainty: The
Discounted Case. Journal of Economic Theory 4(3), 479–513.
Brunnermeier, M., D. Palia, K. A. Sastry, and C. A. Sims (2021). Feedbacks: Financial
Markets and Economic Activity. American Economic Review 111(6), 1845–79.
Brunnermeier, M. K. and Y. Sannikov (2016). The I Theory of Money. NBER Working Paper
No. 22533.
125

Caballero, R. J. and E. Farhi (2018). The Safety Trap. Review of Economic Studies 85(1),
223–274.
Caballero, R. J., E. Farhi, and P.-O. Gourinchas (2017). The Safe Assets Shortage Conundrum. Journal of Economic Perspectives 31(3), 29–46.
Carswell, R. (1981). Economic sanctions and the Iran experience. Foreign Affairs 60, 247–
265.
Celasun, O. and M. Goswami (2002). An Analysis of Money Demand and Inflation in the
Islamic Republic of Iran. IMF Working Paper No. 02/205.
Chu, Y. (2018). Banking deregulation and credit supply: Distinguishing the balance sheet
and the competition channels. Journal of Financial Intermediation 35, 102–119.
Chudik, A., K. Mohaddes, M. H. Pesaran, M. Raissi, and A. Rebucci (2020). A Counterfactual Economic Analysis of Covid-19 Using a Threshold Augmented Multi-Country
Model. National Bureau of Economic Research Working Paper No. 27855.
Chudik, A., M. H. Pesaran, and R. P. Smith (2023). Revisiting the Great Ratios Hypothesis.
Oxford Bulletin of Economics and Statistics 85(5), 1023–1047.
Chudik, A., M. H. Pesaran, and E. Tosetti (2011). Weak and strong cross-section dependence and estimation of large panels. The Econometrics Journal 14, C45–C90.
Cochrane, J. H. (2013). Finance: Function Matters, Not Size. Journal of Economic Perspectives 27(2), 29–50.
Coibion, O., Y. Gorodnichenko, M. Kudlyak, and J. Mondragon (2020). Greater Inequality
and Household Borrowing: New Evidence from Household Data. Journal of the European
Economic Association 18(6), 2922–2971.
Corbae, D. and P. D’Erasmo (2013). A quantitative model of banking industry dynamics.
Federal Reserve Bank of Philadelphia Working Paper (14-13).
Corbae, D. and P. D’Erasmo (2021). Capital buffers in a quantitative model of banking
industry dynamics. Econometrica 89(6), 2975–3023.
Dadkhah, K. and H. Zangeneh (1998). International Economic Sanctions Are Not ZeroSum Games: There Are Only Losers. Iranian Journal of Trade Studies Quarterly 1, 1–14.
Dávila, E. and A. Korinek (2018). Pecuniary externalities in economies with financial
frictions. Review of Economic Studies 85(1), 352–395.
Demirgüç-Kunt, A. and R. Levine (2009). Finance and Inequality: Theory and Evidence.
Annual Review of Financial Economics 1(1), 287–318.
Di Tella, S. (2019). Optimal Regulation of Financial Intermediaries. American Economic
Review 109(1), 271–313.
126

Diallo, B. and W. Koch (2018). Bank concentration and Schumpeterian growth: Theory
and international evidence. Review of Economics and Statistics 100(3), 489–501.
Dizaji, S. F. and P. A. G. van Bergeijk (2013). Potential early phase success and ultimate
failure of economic sanctions: A VAR approach with an application to Iran. Journal of
Peace Research 50, 721–736.
Dornbusch, R. (1976). Expectations and Exchange Rate Dynamics. Journal of Political
Economy 84, 1161–1176.
Doudchenko, N. and G. W. Imbens (2016). Balancing, Regression, Difference-inDifferences and Synthetic Control Methods: A Synthesis. National Bureau of Economic
Research Working Paper No. 22791.
Downs, E. and S. Maloney (2011). Getting China to Sanction Iran: The Chinese-Iranian
Oil Connection. Foreign Affairs 90, 15–21.
Doxey, M. P. (1996). International Sanctions in Contemporary Perspective (2nd ed.). Palgrave
Macmillan UK, London.
Drechsler, I., A. Savov, and P. Schnabl (2020). The financial origins of the rise and fall of
american inflation. NYU Stern Working Paper.
Esfahani, H. S., K. Mohaddes, and M. H. Pesaran (2013). Oil exports and the Iranian
economy. Quarterly Review of Economics and Finance 53, 221–237.
Esfahani, H. S., K. Mohaddes, and M. H. Pesaran (2014). An empirical growth model for
major oil exporters. Journal of Applied Econometrics 29, 1–21.
Fagereng, A., M. Gomez, E. Gouin-Bonenfant, M. Holm, B. Moll, and G. Natvik (2023).
Asset-Price Redistribution. mimeo.
Fagereng, A., M. B. Holm, B. Moll, and G. Natvik (2021). Saving Behavior Across the
Wealth Distribution: The Importance of Capital Gains . NBER Working Paper No. 26588.
Faia, E., S. Laffitte, and G. I. Ottaviano (2019). Foreign expansion, competition and bank
risk. Journal of International Economics 118, 179–199.
Farzanegan, M. R. (2013). Effects of International Financial and Energy Sanctions on Iran’s
Informal Economy. SAIS Review of International Affairs 33, 13–36.
Farzanegan, M. R. (2014). Military spending and economic growth: The case of Iran.
Defence and Peace Economics 25, 247–269.
Farzanegan, M. R. and B. Hayo (2019). Sanctions and the shadow economy: empirical
evidence from Iranian provinces. Applied Economics Letters 26, 501–505.
127

Farzanegan, M. R., M. M. Khabbazan, and H. Sadeghi (2016). Effects of Oil Sanctions on
Iran’s Economy and Household Welfare: New Evidence from a CGE Model. In M. R.
Farzanegan and P. Alaedini (Eds.), Economic Welfare and Inequality in Iran: Developments
since the Revolution, Chapter 8, 185–211. Palgrave Macmillan US, New York.
Farzanegan, M. R. and G. Markwardt (2009). The effects of oil price shocks on the Iranian
economy. Energy Economics 31, 134–151.
Favilukis, J. (2013). Inequality, stock market participation, and the equity premium. Journal of Financial Economics 107(3), 740–759.
Fernández-Villaverde, J., D. Laudati, L. Ohanian, and V. Quadrini (2023). Accounting for
the Duality of the Italian Economy. Review of Economic Dynamics 50, 267–290.
Fostel, A. and J. Geanakoplos (2012). Tranching, cds, and asset prices: How financial innovation can cause bubbles and crashes. American Economic Journal: Macroeconomics 4(1),
190–225.
Fu, F. (2009). Idiosyncratic risk and the cross-section of expected stock returns. Journal of
Financial Economics 91(1), 24–37.
Gardeazabal, J. and A. Vega-Bayo (2017). An empirical comparison between the synthetic
control method and Hsiao et al.’s panel data approach to program evaluation. Journal
of Applied Econometrics 32, 983–1002.
Gennaioli, N., A. Shleifer, and R. Vishny (2014). Finance and the Preservation of Wealth.
Quarterly Journal of Economics 129(3), 1221–1254.
Gharehgozli, O. (2017). An estimation of the economic cost of recent sanctions on Iran
using the synthetic control method. Economics Letters 157, 141–144.
Gomez, M. (2023). Asset Prices and Wealth Inequality. Mimeo.
Gorton, G. (2017). The History and Economics of Safe Assets. Annual Review of Economics 9, 547–586.
Gorton, G., S. Lewellen, and A. Metrick (2012). The Safe-Asset Share. American Economic
Review, Papers & Proceedings 102(3), 101–06.
Greenwood, R. and D. Scharfstein (2013). The Growth of Finance. Journal of Economic
Perspectives 27(2), 3–28.
Guvenen, F., L. Pistaferri, and G. L. Violante (2022). Global trends in income inequality
and income dynamics: New insights from GRID. Quantitative Economics 13(4), 1321–
1360.
Haidar, J. I. (2017). Sanctions and export deflection: Evidence from Iran. Economic Policy 32, 319–355.
128

Haliassos, M. (2013). Financial innovation: Too much or too little? MIT Press: Boston.
Hamilton, J. D. (2018). Why You Should Never Use the Hodrick-Prescott Filter. Review of
Economics and Statistics 100(5), 831–843.
Herrendorf, B., R. Rogerson, and A. Valentinyi (2014). Growth and structural transformation. In Aghion, P. and Durlauf, S. N. (eds.), "Handbook of Economic Growth", Volume 2,
Chapter 6, 855–941. Elsevier Science, Oxford.
Herrera, H. and E. Schroth (2003). Profitable innovation without patent protection: the
case of credit derivatives. Berkeley Center for Entrepreneurial Studies Working Paper, 01–02.
Hsiao, C., H. S. Ching, and S.-K. Wan (2012). A panel data approach for program evaluation: Measuring the benefits of political and economic integration of Hong Kong with
mainland China. Journal of Applied Econometrics 27, 705–740.
Hubmer, J., P. Krusell, and A. A. Smith (2021). Sources of US Wealth Inequality: Past,
Present, and Future. NBER Macroeconomics Annual 35, 391–455.
Hufbauer, G. C., J. J. Schott, and K. A. Elliott (1990). Economic Sanctions Reconsidered:
History and Current Policy (2nd ed.), Volume 1. Peterson Institute for International Economics, Washington, D.C.
Iacoviello, M. (2008). Household Debt and Income Inequality, 1963–2003. Journal of
Money, Credit and Banking 40(5), 929–965.
Igan, D. and P. Mishra (2014). Wall street, capitol hill, and k street: Political influence and
financial regulation. The Journal of Law and Economics 57(4), 1063–1084.
Jordà, Ò. (2005). Estimation and inference of impulse responses by local projections. American economic review 95(1), 161–182.
Jordà, Ò., K. Knoll, D. Kuvshinov, M. Schularick, and A. M. Taylor (2019). The Rate of
Return on Everything, 1870–2015. Quarterly Journal of Economics 134(3), 1225–1298.
Jordà, Ò., M. Schularick, and A. M. Taylor (2017). Macrofinancial history and the new
business cycle facts. NBER macroeconomics annual 31(1), 213–263.
Kiyotaki, N. and J. Moore (2019). Liquidity, Business Cycles, and Monetary Policy. Journal
of Political Economy 127(6), 2926–2966.
Koetter, M., J. W. Kolari, and L. Spierdijk (2012). Enjoying the quiet life under deregulation? Evidence from adjusted Lerner indices for US banks. Review of Economics and
Statistics 94(2), 462–480.
Krishnamurthy, A. and W. Li (2023). The Demand for Money, Near-Money, and Treasury
Bonds. Review of Financial Studies 36(5), 2091–2130.
129

Krishnamurthy, A. and A. Vissing-Jorgensen (2012). The Aggregate Demand for Treasury
Debt. Journal of Political Economy 120(2), 233–267.
Krishnamurthy, A. and A. Vissing-Jorgensen (2015). The impact of Treasury supply on
financial sector lending and stability. Journal of Financial Economics 118(3), 571–600.
Kroszner, R. S. and P. E. Strahan (2014). Regulation and deregulation of the US banking
industry: causes, consequences, and implications for the future. In Economic Regulation
and Its Reform: What Have We Learned?, pp. 485–543. University of Chicago Press.
Kumhof, M., R. Rancière, and P. Winant (2015). Inequality, Leverage, and Crises. American
Economic Review 105(3), 1217–45.
Lansing, K. J. (2015). Asset Pricing with Concentrated Ownership of Capital and Distribution Shocks. American Economic Journal: Macroeconomics 7(4), 67–103.
Laudati, D. and M. H. Pesaran (2021). Identifying the Effects of Sanctions on the Iranian
Economy Using Newspaper Coverage. CESifo Working Paper No. 9217.
Laudati, D. and M. H. Pesaran (2023). Identifying the effects of sanctions on the Iranian
economy using newspaper coverage. Journal of Applied Econometrics 38(3), 271–294.
Levine, R. (2002). Bank-Based or Market-Based Financial Systems: Which Is Better? Journal of Financial Intermediation 11(4), 398–428.
Li, Y. (2017). Procyclical finance: The money view. Fisher College of Business Working Paper (2017-03), 24.
Liu, O. and O. Adedeji (2000). Determinants of Inflation in the Islamic Republic of Iran:
A Macroeconomic Analysis. IMF Working Paper No. 00/127.
Lorenzoni, G. (2008). Inefficient credit booms. The Review of Economic Studies 75(3), 809–
833.
Luck, S. and P. Schempp (2014). Banks, Shadow Banking, and Fragility. ECB Working
Paper 1726.
Majidpour, M. (2013). The Unintended Consequences of US-led Sanctions on Iranian
Industries. Iranian Studies 46, 1–15.
Maloney, S. (2015). Iran’s Political Economy since the Revolution. Cambridge University
Press, Cambridge.
Markiewicz, A. and R. Raciborski (2022). Income inequality and stock market returns.
Review of Economic Dynamics 43, 286–307.
Mazarei, A. (2019). Iran Has a Slow Motion Banking Crisis. Peterson Institute for International Economics Policy Brief No. 19-8.
130

Mazarei, A. (2020). Inflation Targeting in the Time of Sanctions and Pandemic. John
Hopkins University School of Advanced International Studies report.
Merton, R. C. (1995). Financial innovation and the management and regulation of financial
institutions. Journal of Banking & Finance 19(3-4), 461–481.
Mian, A., A. Sufi, and E. Verner (2017). Household debt and business cycles worldwide.
Quarterly Journal of Economics 132(4), 1755–1817.
Mian, A. R., L. Straub, and A. Sufi (2021a). Indebted Demand. Quarterly Journal of Economics 136(4), 2243–2307.
Mian, A. R., L. Straub, and A. Sufi (2021b). The Saving Glut of the Rich. NBER Working
Paper No. 26941.
Mian, A. R., L. Straub, and A. Sufi (2021c). What Explains the Decline in r*? Rising Income
Inequality Versus Demographic Shifts. University of Chicago, Becker Friedman Institute for
Economics Working Paper No. 2021-104.
Mian, A. R. and A. Sufi (2018). Finance and Business Cycles: The Credit-Driven Household Demand Channel. Journal of Economic Perspectives 32(3), 31–58.
Milesi-Ferretti, G. M. (2022). The external wealth of nations database. The Brookings Institution (based on Lane, Philip R. and Gian Maria Milesi-Ferretti, 2018, "The External Wealth of
Nations Revisited: International Financial Integration in the Aftermath of the Global Financial
Crisis," IMF Economic Review 66, 189-222. ).
Mohaddes, K. and M. H. Pesaran (2013). One hundred years of oil income and the Iranian
economy: A curse or a blessing? In P. Alizadeh and H. Hakimian (Eds.), Iran and the
Global Economy: Petro populism, Islam and economic sanctions, Chapter 1, 12–45. Routledge,
New York.
Mohaddes, K. and M. H. Pesaran (2016). Country-specific oil supply shocks and the global
economy: A counterfactual analysis. Energy Economics 59, 382–399.
Mohaddes, K. and M. Raissi (2020). Compilation, Revision and Updating of the Global
VAR (GVAR) Database, 1979Q2-2019Q4. University of Cambridge: Judge Business School
(mimeo).
Morgan, T. C., N. Bapat, and Y. Kobayashi (2014). Threat and imposition of economic sanctions 1945–2005: Updating the TIES dataset. Conflict Management and Peace Science 31,
541–558.
Müller, U. K. and M. W. Watson (2018). Long-Run Covariability. Econometrica 86(3), 775–
804.
Naghavi, A. and G. Pignataro (2015). Theocracy and resilience against economic sanctions. Journal of Economic Behavior & Organization 111, 1–12.
131

Ordoñez, G. and F. Piguillem (2021). Macro and Financial Implications of Aging. NBER
Working Paper No. 26337.
Panageas, S. (2020). The Implications of Heterogeneity and Inequality for Asset Pricing.
Foundations and Trends in Finance 12(3), 199–275.
Pape, R. A. (1997). Why Economic Sanctions Do Not Work. International Security 22, 90–
136.
Pape, R. A. (1998). Why Economic Sanctions Still Do Not Work. International Security 23,
66–77.
Peksen, D. and A. C. Drury (2010). Coercive or Corrosive: The Negative Impact of Economic Sanctions on Democracy. International Interactions 36, 240–264.
Pesaran, M. H. (1992). The Iranian foreign exchange policy and the black market for dollars. International Journal of Middle East Studies 24, 101–125.
Pesaran, M. H. (2000). Economic Trends and Macroeconomic Policies in PostRevolutionary Iran. In P. Alizadeh (Ed.), The Economy of Iran: Dilemmas of an Islamic
State, Chapter 2, 63-99. I.B. Tauris, London.
Pesaran, M. H. (2006). Estimation and inference in large heterogeneous panels with a
multifactor error structure. Econometrica 74, 967–1012.
Pesaran, M. H. (2015). Time Series and Panel Data Econometrics. Oxford University Press,
Oxford.
Philippon, T. (2015). Has the US Finance Industry Become Less Efficient? On the Theory
and Measurement of Financial Intermediation. American Economic Review 105(4), 1408–
38.
Philippon, T. and A. Reshef (2012). Wages and Human Capital in the U.S. Finance Industry: 1909–2006. Quarterly Journal of Economics 127(4), 1551–1609.
Philippon, T. and A. Reshef (2013). An International Look at the Growth of Modern Finance. Journal of Economic Perspectives 27(2), 73–96.
Piazzesi, M. and M. Schneider (2018). Payments, credit and asset prices. BIS Working Paper
No. 734.
Piketty, T. and G. Zucman (2014). Capital is Back: Wealth-Income Ratios in Rich Countries
1700–2010. Quarterly Journal of Economics 129(3), 1255–1310.
Plante, M. (2019). OPEC in the news. Energy Economics 80, 163–172.
Popova, L. and E. Rasoulinezhad (2016). Have Sanctions Modified Iran’s Trade Policy? An Evidence of Asianization and De-Europeanization through the Gravity Model.
Economies 4, Article 24.
132

Quadrini, V. (2017). Bank liabilities channel. Journal of Monetary Economics 89, 25–44.
Rancière, R., A. Tornell, and F. Westermann (2008). Systemic Crises and Growth. Quarterly
Journal of Economics 123(1), 359–406.
Sarto, A. and O. Wang (2022). The Secular Decline in Interest Rates and the Rise of Shadow
Banks. SSRN Working Paper No. 4248374.
Shin, H. S. (2009). Securitisation and Financial Stability. Economic Journal 119(536), 309–
332.
Stansbury, A. M. and L. H. Summers (2018). Productivity and Pay: Is the link broken?
NBER Working Paper No. 24165.
Sunderam, A. (2015). Money creation and the shadow banking system. Review of Financial
Studies 28(4), 939–977.
Thakor, A. V. (2012). Incentives to innovate and financial crises. Journal of Financial Economics 103(1), 130–148.
Wan, S.-K., Y. Xie, and C. Hsiao (2018). Panel data approach vs synthetic control method.
Economics Letters 164, 121–123.
Weiss, T. G., D. Cortright, G. A. Lopez, and L. Minear (Eds.) (1997). Political Gain and
Civilian Pain: Humanitarian Impacts of Economic Sanctions. Rowman & Littlefield, Lanham: MD.
White, H. (1980). A Heteroskedasticity-Consistent Covariance Matrix Estimator and a
Direct Test for Heteroskedasticity. Econometrica 48, 817–838.
133

Appendix A
Appendix to Chapter 1
A.1 Introduction
This Appendix is composed of four main sections. Section A.2 gives details of data
sources and construction of some of the key variables used in our analysis. Sub-sections
A.2.1–A.2.4 provide details of how the proposed sanctions intensity variable and alternative sanction dummies are constructed. Sub-section A.2.5 gives information on conversion of data from the Iranian calender to the Gregorian calender. Sub-section A.2.6
provides details of data sources for the socioeconomic variables, and plots some of the
main macroeconomic variables discussed in Section 1.2 of the paper. In Section A.3 we
present details of the computation of impulse response functions (IRFs), forecast error
variance decompositions (FEVDs), and the bootstrapping procedure used to obtain error
bands for IRFs. Section A.4 reports other empirical results such as the AR specifications
for the sanctions indicator variable and the world output growth. Additional results for
the sanctions-augmented SVAR model, allowing for a number of controls are available
in Sub-section A.4.6. Sub-section A.4.7 provides the estimates of the IRFs for a shock to
the global output growth not presented in the main paper, and IRFs and FEVDs under
a different ordering of the variables in the SVAR model. Finally, a comprehensive list of
134

all major sanctions against Iran from November 1979 to January 2021 is provided in Table
A.74.
A.2 Data appendix
A.2.1 Sanctions intensity variable
Our sanctions intensity variable, st
, is based on newspapers coverage of sanction events
against Iran. Articles were retrieved from the platform ProQuest (www.proquest.com)
which covers the whole period of interest 1979q1–2020q3. ProQuest has detailed
newspapers archives with good search capabilities. The only exception to ProQuest
was the Financial Times Historical Archive accessed through Gale Historical Newspapers
(www.gale.com/intl/primary-sources/historical-newspapers), which helped to fill a gap
left by ProQuest for articles published in the Financial Times before 1996.
Criteria of inclusion
We focused on six major newspapers: the New York Times, the Washington Post, the
Los Angeles Times, the Wall Street Journal, the Guardian, and the Financial Times. We
only selected articles published in the newspapers print version thus disregarding blogs,
websites and other digital formats which are only available more recently; however, we
did allow for all types of articles to be included, e.g. we included both editorials and main
articles.
ProQuest has both a general ProQuest Central database, holding information for the
relatively more recent publications, and several historical newspaper-specific collections
for the most highly printed world outlets, ProQuest Historical Newspapers, which proved
useful in order to extend our series back to 1979. Accordingly, we used the ProQuest
Central data for the maximum period available for each newspaper, and complemented
each series with the ad-hoc historical data sets before such dates. See Table A.1 for details.
135

As mentioned already, the only exception was the articles published in the Financial Times
before 1996, for which there does not exist a historical archive on ProQuest, and instead
Gale Historical Newspapers were used.
To create the index of sanctions imposed on Iran ("sanctions on"), articles were required to include the following terms: "economic*", "sanction*", "against", "Iran*", with
the additional feature of excluding articles in which "lift*" was present. The star at the
end of the previous words allowed the search engine to pick words beginning with the
same initial letters thus including terms such as: "sanctioning", "Iranian", "lifting" etc.. Although the number of potential synonyms and keywords to describe the phenomenon is
virtually very high, this set of words seemed to capture rather well the extent to which
Iran was mentioned as target of international measures. We also found that further complicating the search did not produce sensible results, as the new commands often could
not be recognized by the search engine.
The search was carried out for each newspaper series separately by specifying the name
of the newspaper in the options list "Publication title – PUB". For some newspapers the
search engine produced a handful of duplicates of the same articles despite the option
"Exclude duplicate documents" under "Result page options" had being ticked. To address
this issue, all articles were manually checked before starting the download in order to
avoid double-counting of articles.1
For the period 1990q3–1991q2, the search commands for sanctions against Iran were
updated to exclude also the word "Iraq". This adjustment was necessary in order to avoid
confounding noise due to the events of the Iraq invasion of Kuwait in August 1990, and the
subsequent Gulf War period, from January to February 1991. These events received massive press coverage, which led Iran to be mentioned for geopolitical reasons, not because
of sanctions. Also, some newspapers reported two additional small spikes not strictly related to Iran: (i) For the terrorist attacks happened between December 1985 (in Rome and
1The extent of this technical hurdle varied considerably amongst outlets. It was particularly severe for journals such as
the Los Angeles Times, while virtually non-existent for other newspapers such as the New York Times.
136

Vienna airports) and April 1986 (in a West Berlin discotheque); (ii) For the "1998 Coimbatore bombings" attacks in southern India. In both cases, Iran was not the target of new
sanctions therefore a manual check deletion of these small number of occurrences had to
be carried out.
The intensity variable to capture the partial lifting of sanctions ("sanctions off") included the words beginning with "economic*", "sanction*", "against", "Iran*" but now allowing also for at least one of the following words: "lift*", "waive*" and "accord*". An
exception was made for the Historical Database of the Financial Times, which does not
support sophisticated search structures. Therefore, a simple research allowing for "sanctions against Iran" and "deal*" was conducted to capture the highest number of articles,
which were subsequently checked and skimmed manually to meet our criteria of inclusion.
A detailed chronological study of economic sanctions against Iran allowed us to restrict
our search of "sanctions off" on two time periods only. First, in 1981 when the Algiers Accords were signed and the "Tehran hostage crisis" ended; second, from 2016q1 to 2018q2,
when the Joint Comprehensive Plan of Action (JCPOA) was enacted by all world major
powers before U.S. President Trump withdrew the country from the agreement. Accordingly, for construction of the sanction-off index we focussed on the periods 1981q1–1981q4
and 2015q1–2018q2 in order to avoid unnecessary noise for the time in between and after
Trump’s announcement. The "sanctions off" period of our indicator was extended to one
year before the actual implementation of the JCPOA in order to allow for possible anticipatory effects.
137

Table A.1: Sources of newspaper articles over the period 1979m1–2020m9
Period
Historical dataset Modern dataset
New York Times 1979m1–1980m12 1981m1–2020m9
Los Angeles Times 1979m1–1984m12 1985m1–2020m9
Washington Post 1979m1–2002m12 2003m1–2020m9
Wall Street Journal 1979m1–1983m12 1984m1–2020m9
Guardian 1979m1–1996m12 1997m1–2020m9
Financial Times 1979m1–1995m12 1996m1–2020m9
Notes: "Historical data set" is the ProQuest Historical Newspapers data set for all newspapers except the Financial Times,
for which information have been retrieved from Gale Historical Newspapers. "Modern data set" is ProQuest Central database
for all newspapers considered.
A.2.2 Sanctions intensity variable construction
Having obtained a number of daily articles related to the sanctions imposed ("sanctions
on") and lifted ("sanctions off"), we proceeded with the following steps in order to build
our estimator, st(w) = st,on − w × st,of f . Here we focus on the construction of st,on. The
same procedure was used to construct st,of f .
First, we computed a monthly series for each of our J newspapers (J = 6) by averaging
our daily series over the number of articles per month. In turn, we carried out a simple
average across newspapers, which led us to have a single monthly series of "sanctions on"
articles; subsequently, we averaged the monthly observations over each quarter to obtain
the quarterly series. The "sanctions on" average was then divided by its maximum value
over the period 1989q1–2020q3 in order to obtain the indicator st,on; so that st,on index was
defined on the (0, 1) range. We obtained a second variable st,of f from our "sanctions off"
raw count by following the same steps just described. Finally, we estimated the weight,
w ∈ (0, 1), with a grid search in order to derive our final sanctions intensity variable st =
138

Table A.2: Quarterly estimates of the log-likelihood of Equation (A.1) estimated over the period
1989q1-2019q4 for values of w ∈ {0.1, 0.2, . . . 0.9}
Grid value, w Equation log-likelihood
0.1 258.095
0.2 258.213
0.3 258.289
0.4 258.320
0.5 258.305
0.6 258.248
0.7 258.153
0.8 258.028
0.9 257.880
Notes: The values on the grid of w have been used to construct different sanctions intensity indicators st(w) = st,on −
w × st,off . The maximum likelihood of Equation (A.1) across different grid values provided the specification for the
optimal weight wˆ. See Sections A.2.1 and A.2.2 for details on the construction of son,t and soff,t.
st,on − w × st,of f . The grid search was performed by running the regressions:
∆yt = β0 + β1∆yt−1 + β2st−1(w) + εt
, (A.1)
over the period 1989q1–2019q4, with ∆yt being Iran’s quarterly real output growth, and
with a step size of our grid equal to 0.1. The optimal weight was estimated as wˆ = 0.4,
although the shape of the likelihood was rather flat. Table A.2 provides the values of the
log-likelihood of Equation (A.1) estimated for different values of parameter w.
As a robustness check, we created a standardized version of our indicator by following the approach advanced by Baker, Bloom, and Davis (2016). We divided each of the J
newspapers monthly raw series by their respective standard deviations.2 The final standardized intensity variable was obtained as before by averaging across newspapers at
monthly frequency, taking the simple mean for each quarter (for both sanctions "on" and
"off"), dividing each series by their respective maxima over the period 1989q1–2020q3, and
2
For a measure of "sanctions on", we considered the standard deviations over the entire period 1979m1–2020m9. For
"sanctions off", the monthly raw counts during 1981m1–1981m12 and 2015m1–2018m6 were divided by the standard
deviations over their respective periods.
139

Figure A.1: Sanctions intensity variable and standardized sanctions intensity variable over the
period 1989q1–2020q3
Notes: See Section 1.3 of the paper for the sanctions intensity variable definition over the range (0,1). See Sections A.2.1
and A.2.2 in the data appendix for details on construction of both the sanctions intensity variables.
subtracting the "standardized sanctions off" series from the "standardized sanctions on".
We found these weighted "sanctions on" and "sanctions off" series to be very close to the
ones based on simple averages, and as a result the grid search applied to the weighted series also resulted in the estimate wˆ = 0.4. Even though this procedure was meant to avoid
newspapers with a larger number of articles per issue to carry unwarranted weight, the
two series co-move almost perfectly (ρ = 0.998). See Figure A.1. This finding is consistent
with Plante (2019), who adjusts for the number of total articles per month and finds that
his two measures correlate at 0.97.
A.2.3 U.S. Treasury sanctions variable construction
We also constructed a measure of sanctions intensity based on the U.S. Treasury "Specially Designated Nationals And Blocked Persons List (SDN)". The online database of the
Treasury keeps track only of the entities currently sanctioned. To compile a complete time
series list of all Iranian entities, individuals, and vessels being sanctioned by the U.S., we
140

used yearly pdf files available in the online archive of the U.S. Treasury. In this way, we
were able to follow over time each entity entering and exiting the database.3 The list of
sanctioned entities can be retrieved from 1994 onwards but the number of entries for Iran
up to 2005 is negligible. This is in line with the historical record of U.S. sanctions against
Iran. Therefore, we focussed on building our entry-exit matrix from 2006 onwards.
To construct the U.S. Treasury sanctions variable, we first summed the total number of
Iranian entities, individuals, and vessels being hit by U.S. sanctions.4
In the SDN lists, entities refer to companies (and institutions) of Iranian nationality, foreign companies having
offices in Iran, and – in light of secondary sanctions – all other foreign companies doing business with sanctioned Iranian companies. Iranian individuals, or foreigners doing
business with sanctioned Iranians, were tracked by First and Last Name, and Passport
number or National ID – when available. For vessels, we did not confine ourselves to vessels name or national flag given that these attributes were often changed. Instead, the
International Maritime Organization (IMO) unique identification number proved to be
important and completely reliable to follow vessels history.
The number of Iranian entries added to the SDN list allowed us to build an "SDN sanctions on" time series; similarly, the entries removed from the list provided the information
for an "SDN sanctions off" index. We obtained our final "U.S. Treasury sanctions variable"
by attaching a weight to the "SDN sanctions off" count equal to the newspaper-based indicator (w = 0.4) and subtracting it from the "SDN sanctions on" count. The final series
was then re-scaled by dividing it for its maximum value. See Figure A.2. The correlation
between the U.S. Treasury measure and st
is equal to 38 per cent over the period 2006q1–
2020q3. Notice that the series based on SDN has inevitably negative values over the JCPOA
period regardless of the weight one is willing to choose. This feature is due to the fact that
no new Iranian entities were added, while a large number of previously sanctioned entities
3The documents specify the exact day in which entities enter/exit the list during the year considered.
4Notice that SDN lists specify to which sanctions programs each entry belongs. In other words, according to whether
the aim is to hit entities related to Iran vis-à-vis other nations (say, North Korea) different codes are attached to them.
141

were removed.5
A.2.4 Sanctions dummy variables
In many applications sanctions are characterized by dummy variables that take discrete
values representing "sanctions on" and "sanctions off" periods. To evaluate the effectiveness of the sanctions intensity variable proposed in this paper (following a suggestion
from a referee), we also consider two alternative sanctions dummy variables.
The first dummy variable is constructed based on historical narratives on major sanction events, as summarized in Table A.74. Accordingly, we consider the sanctions dummy
variable, dt
, that takes the value of zero for t ≤ 2010q4, the value of unity over the period
2011q1 − 2015q2, the value of zero over the period 2015q3 − 2018q1, and unity thereafter.
The date 2015q3 coincides with new U.N. and U.S. sanctions in response to Iran’s increased
nuclear activity, 2015q2 is the date of the Joint Comprehensive Plan Of Action (JCPOA)
agreement, representing a "sanctions off" period. In 2018q2 new sanctions were imposed
on Iran unilaterally by the U.S. as a part of the "maximum pressure" strategy followed
by the U.S. President Trump. See also Table A.3 where "sanctions on" and "sanctions off"
periods are summarized.
Table A.3: Sanctions dummy variable description over the period 1989q1-2019q4
Historical period dt
1989q1-2010q4 0
2011q1-2015q2 1
2015q3-2018q1 0
2018q2-2019q4 1
As a second dummy variable we consider a discretized version of our sanctions inten5This could also be considered as a shortcoming of using such measure given that – in our framework – a negative value
of the sanctions intensity variable means an attempt to subsidize the Iranian economy through transfers, something far
from the actual process happening over the period 2016q1–2018q2.
142

Figure A.2: Sanctions intensity variable and the U.S. Treasury sanctions variable over the period 2006q1–2020q3
Notes: The U.S. Treasury sanctions variable is computed from the number of newly introduced and removed entries in the "Specially Designated Nationals And
Blocked Persons List" (SDN) of the U.S. Department of the Treasury. Major sanctions-related historical events are indicated by arrows and brackets. See Sections
A.2.1–A.2.3 for details of the construction of the sanctions variables.
143

sity variable, st
, which we denote by s
D
t
, taking the values of 0.5, 1.0, 1.5, and 2.0 in different sub-periods. The period 1989q1 − 1995q2 was characterized as a very mild sanctions
episode during the Presidency of Rafsanjani with s
D
t
set to 0.5. Following our newspapers sanctions intensity, and consistently with the historical narrative, we then set s
D
t = 1
over the period 1995q3 − 1998q2 to reflect the U.S. President Clinton decisions to adopt
slightly stricter measures under the Iran and Libya Sanctions Act of 1996, and the U.S. executive order 13059 of 1997. Tensions between U.S. and Iran abated somewhat under the
Khatami’s Presidency (1998q3 − 2005q4) with fewer sanctions, and to reflect this we set
s
D
t = 0.5 over this period. But, with the election of President Ahmadinejad, tensions between Iran and the West started to rise and the U.S. and its allies incrementally increased
their sanctions against Iran. Accordingly, we set s
D
t
equal to 1.0 and 2.0 over the subperiods 2006q1 − 2011q4 and 2012q1 − 2015q2, respectively. We then set s
D
t
to 0.5 during
2015q3−2018q1, which is the period marking the start of the JCPOA accord which, as noted
above, ended with the re-introduction of "maximum pressure" sanctions in 2018q2 by President Trump. To reflect this change, we increased s
D
t
to 1.5 over the period 2018q2−2019q4.
Table A.4 and Figure A.3 provide a summary and visual representation of s
D
t
in comparison to st
. By construction, we expect s
D
t and st to be highly correlated, and our main
purpose of considering s
D
t
is to see if much will be lost by discretization of st
.
Table A.4: Discretized sanctions intensity variable description
Historical period s
D
t
1989q1-1995q2 0.5
1995q3-1998q2 1.0
1998q3-2005q4 0.5
2006q1-2011q4 1.0
2012q1-2015q2 2.0
2015q3-2018q1 0.5
2018q2-2019q4 1.5
144

Figure A.3: Sanctions intensity variable, and discretized sanctions intensity variable, over the
period 1989q1–2020q3
A.2.5 Conversions from Iranian to Gregorian calendar
The data we use in our analysis are in Gregorian calendar. However, data retrieved
from Iranian sources, namely from the Central Bank of Iran and the Statistical Center of
Iran, follow the Iranian calendar format. The Iranian year starts on March 21st of the corresponding Gregorian year. Accordingly, we carried out three calendar conversions in order
for the Iranian data to be in line with the ones in the Gregorian format. In the following
expressions, Gy, Gq, and Gm stand for the variables transformed in the Gregorian calendar
at yearly, quarterly, and monthly frequencies, respectively, while Iy, Iq, and Im are the data
in the original Iranian format. For annual statistics, the following formula was applied:6
Gy =
80
365 Iy−1 +
285
365 Iy. For quarterly data, we converted the Iranian series according to:7
Gq =
8
9
Iq−1 +
1
9
Iq. Finally, for the monthly time series – we applied the following transformation: Gm =
1
3
Im−1 +
2
3
Im.
6Eighty days of the Gregorian year (from Jan. 1
st to Mar. 21st) were to be attributed to the previous Iranian year.
7
In the following expression, 8/9 represents the eighty days out of the approximately ninety days within a given quarter.
145

A.2.6 Economic and socio-demographic variables
In this section, we will refer to some of the Iranian data as being retrieved from the
"Quarterly Iran Data Set 2020". In this case, we extend and update the data for Iran in
the GVAR Data Set compiled by Mohaddes and Raissi (2020) until 2018q2 (and available
upon request); more recent observations for Iran were added by splicing forward the previously available series with new observations from Iranian sources. In this respect, the
conversions mentioned in Section A.2.5 were applied. All data from the Central Bank of
Iran (CBI) were obtained from the Economic Time Series Database. For global factors we
will refer to the "GVAR Data Set 2020". In this case, we use the latest version of the GVAR
Data Set provided by Mohaddes and Raissi (2020).
The quarterly real output of Iran was obtained by splicing forward the GVAR series in
the Quarterly Iran Data Set 2020 available until 2018q2 with the "Iran’s Quarterly National
Accounts" released by the Statistical Center of Iran until 2020q1.
Iran’s inflation was computed as first difference of the natural logarithm of Iran consumer price index (CPI). CPI data from the GVAR series in the Quarterly Iran Data Set
2020 available until 2018q2 were extended forward with data from the Statistical Center
of Iran, which provides Iranian monthly inflation bulletins. After having converted the
monthly series to the Gregorian calendar, it was possible to compute the quarterly inflation rate, and splice forward the Quarterly Iran Data Set until 2021q1. The CPI was then
re-based to have value equal to 100 in 1979q2.
The official foreign exchange statistics from 1979q2 to 2020q3 were retrieved in quarterly
format from Bank Markazi (Iran’s Central Bank), and converted to the Gregorian calendar.
The free market foreign exchange rate in quarterly format from 1979q2 to 2017q4 was also
retrieved from Bank Markazi. For 2018 onward the series were spliced forward with data
from bonbast.com – a highly cited website tracking the Iran’s rial free market rate against
all major currencies. In this regard, bonbast.com presents information for "buy" and "sell"
146

rates at daily frequency. We used the average of buy and sell rates. In this way we were
able to extend the historical series from Bank Markazi until 2021q1.
Data on oil exports revenues from 1999q1 to 2021q3 were retrieved from the CBI
through Haver analytics, and adjusted for the appropriate calendar conversion. Subsequently, the series was spliced backwards with the data from Esfahani, Mohaddes, and
Pesaran (2014). Esfahani, Mohaddes, and Pesaran (2014) provide data on international
oil price and Iran’s quantity of oil exported (th. barrels/day) since 1979q1 therefore the
two series were first multiplied to obtain oil revenues in millions of U.S. dollars.
Monetary statistics were also downloaded from the Bank Markazi website. The monetary aggregate M2 was computed as the sum of M1 and "quasi-money". Data were available at quarterly frequency, and – before converting them to the Gregorian calendar – the
observations from 2015q2 onwards had to be multiplied by 1,000 given a change of format
from billions to trillions of rials.
In order to account for global factors, we augmented our analyses with several variables. ∆p
0
t
is the rate of change of the oil price (first difference of the natural logarithm).
The oil price considered was the Brent crude (U.S. dollars/barrel). Data at quarterly frequency until 2020q1 were taken from the GVAR Data Set 2020. Observations for 2020q2
and 2020q3 were obtained by splicing the series with data from the U.S. Energy Information Administration (series name: "Europe Brent Spot Price FOB, Dollars per Barrel").
The E.I.A. provided information at monthly frequency therefore we first averaged the oil
prices over each quarter, and then spliced forward our GVAR time series.
The quarterly global realized volatility, grvt
, was taken directly from the GVAR Data
Set 2020 for the whole period 1979q2–2020q1; details about its construction can be found
in Chudik, Mohaddes, Pesaran, Raissi, and Rebucci (2020).
We used the GVAR Data Set 2020 and followed the procedure indicated by Chudik,
Mohaddes, Pesaran, Raissi, and Rebucci (2020) also for the construction of the other global
factors. The factors we considered are: the world real output growth, ∆ywt; the rate of
147

Figure A.4: Relevant Iran’s and World macroeconomic and financial time series over the period
1979–2020
Panel A Panel B
Iranian oil exports1 Shares of oil and gas vs. non-oil and
gas exports revenues 2
Panel C Panel D
Free market and official FX rates in logs3
Free market FX rate and CPI in logs4
Notes: 1. Annual data over the period 1979–2020. 2. Annual data over the period 1979–2019. 3. Quarterly data over
the period 1979q2–2020q3. Foreign exchange rates are expressed as number of Iran’s rials per U.S. dollars. 4. Quarterly
data over the period 1979q2–2020q3. CPI stands for Consumer Price Index, and it is equal to 100 in 1979q2. See Sections
A.2.5 and A.2.6 in the data appendix for details on calendar conversions, and sources of the data.
change of the world real exchange rate against the U.S. dollar, ∆ewt; the world real equity
returns, ∆reqwt; and the per cent change of the world nominal long-term interest rate,
∆rwt. These control variables were obtained by taking the first difference of the following
weighted cross-sectional averages: ywt =
Pn
i=0 wiyit, ewt =
Pn
i=0 wieit, reqwt =
Pn
i=0 wieqit,
rwt =
Pn
i=0 wirit, where yit, eit, eqit, rit are: the log of real output, the log of the real
148

Figure A.4: Relevant Iran’s and World macroeconomic and financial time series over the period
1979–2020
Panel E
Iran and World real output in logs5
Notes: 5. Quarterly data over the period 1979q2–2020q1. The world real output is a weighted average of the natural
logarithm of real output for 33 major economies. See Sections A.2.5 and A.2.6 in the data appendix for details on calendar
conversions, and sources of the data.
exchange rate against the U.S. dollar, the log of real equity prices, and the nominal long
term interest rates of country i in quarter t. The sample included 33 of the world major
economies, and the weights, wi
, were computed as the GDP-PPP average by country i out
of the overall world average output over the period 2014–2016:
wi =
P2016
t=2014 Y
P P P
P
it
n
i=0
P2016
t=2014 Y
P P P
it
. (A.2)
The GDP-PPP measure allows for international comparisons, and it was retrieved at yearly
frequency from the World Bank Open Data repository. The 33 countries are: Argentina, Australia, Austria, Belgium, Brazil, Canada, China, Chile, Finland, France, Germany, India,
Indonesia, Italy, Japan, South Korea, Malaysia, Mexico, the Netherlands, Norway, New
Zealand, Peru, the Philippines, South Africa, Saudi Arabia, Singapore, Spain, Sweden,
Switzerland, Thailand, Turkey, the U.K., and the U.S.A..
For some of the 33 countries, real equities returns, eqit, and nominal long term interest
149

rates, rit, were not available. As such, to compute reqwt and rwt we focussed on the countries for which we had information, and rescaled the weights accordingly. In particular,
the historical real equity prices, eqit, were available for 26 out of 33 countries (excluded
were Brazil, China, Indonesia, Mexico, Peru, Saudi Arabia, and Turkey). For the long run
interest rates, rit, data were available for 18 of the 33 countries (excluded were Argentina,
Brazil, China, Chile, Finland, India, Indonesia, Malaysia, Mexico, Peru, the Philippines,
Saudi Arabia, Singapore, Thailand, and Turkey).
Table A.5: Sources of quarterly data
Data series Source
Iranian variables1
Consumer price index Quarterly Iran Data Set 2020
Foreign exchange rate, Free Market Central Bank of Iran and bonbast.com
Foreign exchange rate, Official rate Central Bank of Iran
Money supply: M1 and Quasi-money Central Bank of Iran
Oil export revenues Quarterly Iran Data Set 2020
Real output Quarterly Iran Data Set 2020
Global control variables2 GVAR Data Set 2020 and World Bank
Notes: 1. The Quarterly Iran Data Set 2020 extends and updates the GVAR Data Set compiled by Mohaddes and Raissi
(2020), whose observations for Iran are available up to 2018q2. Such version of the data base including Iran is available
upon request. The most recent observations for the consumer price index taken from the Statistical Center of Iran can be
retrieved from the monthly inflation bulletins available at www.amar.org.ir. The data provided by the Central Bank of Iran
on foreign exchange rates are available from the Economic Time Series Database: tsd.cbi.ir, under "External Sector/Value
of Financial Assets (Exchange Rate and Coin Price)". Recent data on free market foreign exchange data can be retrieved
from www.bonbast.com. Money supply statistics are available under "Monetary and Credit Aggregates" at tsd.cbi.ir. The
updated data on oil export revenues were retrieved from the CBI through Haver analytics, and extended backwards by
using the data set of Esfahani, Mohaddes, and Pesaran (2014) available since 1979q1. The data used to extend the Iran’s
real output series are taken from the Statistical Center of Iran and can be retrieved under "Iran’s Quarterly National
Accounts (base year = 1390)" from www.amar.org.ir.
2. Raw data for each country composing the global averages were retrieved from the GVAR Data Set compiled by
Mohaddes and Raissi (2020) and available at www.mohaddes.org/gvar. The World Bank data (data.worldbank.org) have
been used to construct the GDP-PPP weights for each country (code indicator: "NY.GDP.MKTP.PP.CD"). The variables
included in this set of controls are: global nominal long term interest rate, global real equity price, global real exchange
rate, global real output, global realized volatility, and oil price (Brent crude). For oil price, the observations for 2020q2
and 2020q3 were obtained from the U.S. Energy Information Administration (series name: "Europe Brent Spot Price
FOB, Dollars per Barrel") available at www.eia.gov. Information on the U.S. consumer price index was retrieved from
the FRED data base fred.stlouisfed.org (series name: "Consumer Price Index: Total All Items for the United States, Index
2015=100, Quarterly, Seasonally Adjusted").
See Section A.2 for further details on variables construction.
150

A.3 Computation of IRFs, FEVDs and their error bands by
bootstrap
The SVAR model can also be used to compute the time profile of the responses of the
economy to shocks (sanction, domestic and foreign) using impulse response functions
(IRFs). For the purpose of computing IRFs, we drop money supply growth and foreign
variables except the world output growth, as none of these variables will prove to be statistically significant.
A.3.1 Impulse response analysis for SVAR model of the Iranian economy
IRFs for domestic shocks. Our starting point is Equation (1.6) where
qt = (∆x
0
t
, ∆ef t, ∆mt
, ∆pt
, ∆yt)
0
, and there are five domestic shocks εt =
(ε∆x0,t, ε∆e,t, ε∆m,t, ε∆p,t, ε∆y,t)
0
. The IRFs of one standard error shock to domestic
shocks are given by:
IRFq(h, √
σjj ) = E

qt+h

It−1, εt,j =
√
σjj
−E (qt+h |It−1 ), for j = ∆x
0
, ∆ef , ∆m, ∆p, ∆y,
where h = 0, 1, 2..., H, is the horizon of the IRFs, σjj = V ar (εjt), and It−1 is the information set at time t−1. The IRFs compare the expected outcome of the shock (intervention) to
an alternative counterfactual in the absence of the shock. Using the reduced form version
of (1.6), we have IRFq(h, √σjj ) = √σjj (GhA−1
0 ej ), where:
G` = Φ1G`−1 + Φ2G`−2, for ` = 1, 2, . . . , (A.3)
with G−1 = 0, and G0 = Im, Φj = A−1
0 Aj
, for j = 1, 2, and ej
is a m×1 (m = 5) selection
vector of zeros except for its j
th element which is unity. See Chapter 24 of Pesaran (2015).
15

More specifically, the impulse response effects of a positive one standard error shock to
the j
th domestic variable, √σjj , on the i
th variable at horizon h = 0, 1, ..., H, are given by
IRFij (h, √σjj ) = √σjj (e
0
iGhA−1
0 ej ), for i, j = ∆x
0
, ∆ef , ∆m, ∆p, ∆y.
IRFs for a shock to the sanctions intensity variable. Since global factors are assumed
to be strictly exogenous to the Iranian economy and unrelated to sanctions, then without
loss of generality the IRFs of sanction shocks can be obtained abstracting from the global
shocks. Accordingly, using (1.6) and (1.8), the moving average (MA) representation of
the domestic variables can be written as:
qt = G(1)A−1
0

aq +
as
1 − ρs
γs

+ b(L)ηt + G(L)A−1
0 εt
, (A.4)
where γs = γ0s + γ1s
, b(L) = G(L)A−1
0
(1 − ρsL)
−1
(γ0s + γ1sL), G(L) = P∞
`=0 G`L
`
,
and G`
is defined by the recursions in (A.3). Therefore, the responses of the i
th domestic
variable (the i
th element of qt) to a positive one standard error shock to the sanctions
intensity variable, ωs, are given by:
IRFi(h, ωs) = ωs(e
0
ibh), h = 0, 1, ..., H, i = ∆x
0
, ∆ef , ∆m, ∆p, ∆y. (A.5)
In the case where sanctions are imposed over h periods, the cumulative IRFs (CIRFs)
is given by:
CIRFq(h, ωs) = E (qt+h |It−1, st,t+h = ωsτh+1 ) − E (qt+h |It−1 )
where st,t+h = (st
, st+1, ..., st+h)
0 and τh+1 is an (h + 1)×1 vector of ones. The cumulative
responses of the i
th endogenous variable to sanctions shocks of size, ωs, that are sustained
152

over h periods are given by
CIRFi(h, ωs) = ωs
X
h
`=0
e
0
ib`
!
, for h = 0, 1, ..., H, i = ∆x
0
, ∆ef , ∆m, ∆p, ∆y.
IRFs for a global factor shock. As noted earlier, we only consider the shock to the world
output growth, ∆ywt, as the global factor in our analysis, and consider the following general linear process for ∆ywt:
∆ywt = g0 + c(L)v∆ywt
. (A.6)
Since the sanctions intensity variable and the world output growth are assumed to be
uncorrelated, abstracting from the sanctions intensity variable we can re-write (1.6),
A0qt = aq + A1qt−1 + A2qt−2 + δw∆ywt + εt
.
By combining (A.6) with the moving average representation of the above equation we
have:
qt = G(1)A−1
0
(aq + δwg0) + κ(L)v∆ywt + G(L)A−1
0 εt
, (A.7)
where κ(L) = P∞
`=0
κ`L
` = G(L)A−1
0 δwc(L), and G(L) is as defined above. Hence, the
impulse responses of the i
th element of qt to a single period shock to world output growth
is then given by:
IRFi(h, ω∆yw
) = ω∆yw
(e
0
iκh), h = 0, 1, ..., H, i = ∆x
0
, ∆ef , ∆m, ∆p, ∆y, (A.8)
where ω
2
∆yw
is the variance of v∆ywt
.
A.3.2 Forecast error variance decompositions
Another useful measure of dynamic propagation of shocks is forecast error variance
decompositions (FEVDs), which measure the proportion of forecast error variance of vari153

able qit (say, output growth) which is accounted for by a particular domestic shock, εjt, at
different horizons. We are particularly interested in estimating the relative importance of
domestic shocks vis-à-vis sanctions or world output shocks in explaining output growth
at different horizons. To obtain the FEVDs of both types of shocks, we first note that, by
building on (A.4) and (A.7), the n-step ahead forecast errors for the vector of domestic
variables, qt
, is given by:
ξt
(n) = Xn
`=0
b`ηt+n−` +
Xn
`=0
κ`v∆yw,t+n−` +
Xn
`=0
G`A−1
0 εt+n−`
,
where, as before, εt
is a m × 1 (with m = 5) vector of domestic shocks. Using standard results from the literature, the h-step ahead FEVD of the i
th variable in qt which is
accounted by the domestic shock εjt is given by
θij (h) = σjj Ph
`=0
e
0
iG`A−1
0 ej
2
Ph
`=0 e
0
iG`A−1
0 ΣA0−1
0 G
0
`
ei + ω2
s
Ph
`=0 e
0
ib`b
0
`
ei + ω
2
∆yw
Ph
`=0 e
0
iκ`κ
0
`
ei
, (A.9)
for i, j = ∆x
0
, ∆ef , ∆m, ∆p, ∆y, and Σ = Diag(σ∆x0∆x0
, σ∆e∆e, σ∆m∆m, σ∆p∆p, σ∆y∆y).
Similarly, the proportion of the forecast error variance of the i
th variable due to sanctions
intensity and world output growth shocks at horizon h are given by:
θis(h) = ω
2
s
Ph
`=0 e
0
ib`b
0
`
ei
Ph
`=0 e
0
iG`A−1
0 ΣA0−1
0 G0
`
ei + ω2
s
Ph
`=0 e
0
ib`b
0
`
ei + ω
2
∆yw
Ph
`=0 e
0
iκ`κ
0
`
ei
, (A.10)
and
θi∆yw
(h) =
ω
2
∆yw
Ph
`=0 e
0
iκ`κ
0
`
ei
Ph
`=0 e
0
iG`A−1
0 ΣA0−1
0 G0
`
ei + ω2
s
Ph
`=0 e
0
ib`b
0
`
ei + ω
2
∆yw
Ph
`=0 e
0
iκ`κ
0
`
ei
, (A.11)
respectively. Since all the shocks are assumed to be orthogonal, then it follows that
Pm
j=1 θij (h) + θis(h) + θi∆yw
(h) = 1.
In the case where the effects of sanctions shocks are cumulated keeping other shocks
15

fixed, we have:
θij (h) = σjj Ph
`=0
e
0
iG`A−1
0 ej
2
Ph
`=0 e
0
iG`A−1
0 ΣA0−1
0 G
0
`
ei + ω2
s
Ph
`=0 e
0
ib`
Ph
`=0 eib`
0
+ ω
2
∆yw
Ph
`=0 e
0
iκ`κ
0
`
ei
,
(A.12)
where θij (h) is the h-step ahead FEVD of the i
th variable in qt which is accounted by the domestic shock εjt, for i, j = ∆x
0
, ∆ef , ∆m, ∆p, ∆y, and Σ =
Diag(σ∆x0∆x0
, σ∆e∆e, σ∆m∆m, σ∆p∆p, σ∆y∆y). Similarly, the proportion of the forecast error
variance of the i
th variable due to a cumulated sanction intensity shock at horizon h is
given by:
θis(h) =
ω
2
s
Ph
`=0 e
0
ib`
Ph
`=0 eib`
0
Ph
`=0 e
0
iG`A−1
0 ΣA0−1
0 G0
`
ei + ω2
s
Ph
`=0 e
0
ib`
Ph
`=0 eib`
0
+ ω
2
∆yw
Ph
`=0 e
0
iκ`κ
0
`
ei
.
(A.13)
Finally, the contribution of world output growth shock to the forecast error variance of
the i
th variable in qt can be written as:
θi∆yw
(h) =
ω
2
∆yw
Ph
`=0 e
0
iκ`κ
0
`
ei
Ph
`=0 e
0
iG`A−1
0 ΣA0−1
0 G0
`
ei + ω2
s
Ph
`=0 e
0
ib`
Ph
`=0 eib`
0
+ ω
2
∆yw
Ph
`=0 e
0
iκ`κ
0
`
ei
.
(A.14)
A.3.3 IRFs and FEVDs alternative computation
To compute the IRFs and FEVDs, we provide an alternative computation approach
with respect to the one described above. We confirm that we obtain the same numerical
results using the formulae in Sub-sections A.3.1 and A.3.2, which we had included for
pedagogic reasons.
Re-write Equation (A.15) as:
ezt = Ψe
−1
0

ea+Ψe 1ezt−1 + Ψe 2ezt−2 + uet

,
15

with ezt = (∆x
0
t
, ∆ef t, ∆mt
, ∆pt
, ∆yt
, st
, ∆ywt)
0 and uet =
(ε∆x
0
t
, ε∆eft , ε∆mt
, ε∆pt
, ε∆yt
, εst
, ε∆ywt
)
0
. The IRF can be computed by following the
approach described above as:
IRFz(h) = √
σjj (FhΨe
−1
0 ej ),
where ej
is a (m+2)×1, with m = 5, selection vector of zeros except for its j
th element,
which is unity, and
F` = Φe1F`−1 + Φe2F`−2, for ` = 1, 2, . . .
where Φe1 = Ψe
−1
0 Ψe 1, Φe2 = Ψe
−1
0 Ψe 2, with F−1 = 0, and F0 = Im+2. Consequently, the
impulse response effects of a positive one standard error change in the j
th domestic shock,
εjt, on the i
th variable (the i
th element of ezt) are given by:
IRFij (h) = √
σjj (e
0
iFhΨe
−1
0 ej ), for h = 0, 1, ..., H, i, j = ∆x
0
t
, ∆ef t, ∆mt
, ∆pt
, ∆yt
, st
, ∆ywt .
The forecast errors can be now written more succinctly as:
eξt
(n) = Xn
`=0
F`Ψe
−1
0 uet+n−`
,
where, as before, uet
is a vector of (m + 2) × 1 shocks. Similarly, the proportion of the
forecast error variance of the i
th variable due to a shock to the j
th variable at horizon h is
given by:
θij (h) =
σjj Ph
`=0
e
0
iF`Ψe
−1
0 ej
2
Ph
`=0 e
0
iF`Ψe
−1
0 ΣΨe
0−1
0 F0
`
ei
, for i, j = ∆x
0
t
, ∆ef t, ∆mt
, ∆pt
, ∆yt
, st
, ∆ywt ,
with Σ = Diag(σ∆x0∆x0
, σ∆e∆e, ..., σ∆yw∆yw
). It can be proved that Pm
j=1 θij (h) +θis(h) +
θi∆yw
(h) = 1.
156

A.3.4 Bootstrapping procedure
In order to compute the impulse response functions (IRFs) and the associated confidence bands, we followed a bootstrap procedure by simulating the in-sample values of zt
in Equation (1.10), which we report here for convenience:
Ψ0zt = a+Ψ1zt−1 + Ψ2zt−2 + ut
. (A.15)
In Equation (A.15), zt = (qt
, st
,¯zwt)
0
is a vector of m domestic policy variables (qt), the
sanctions intensity variable (st), and the k global factors (¯zwt); a is a (m +k + 1)×1 vector
of constants, and ut are the residuals of the system. In order to generate our bootstrap
replications, we proceed as follows:
1. Generate the simulated residuals n
u
(r)
t
, r = 1, 2, ..., Ro
by re-sampling with replacement from the estimated residuals of each equation separately {ubt
, t = 3, 4, ..., T},
where R = 1, 000 is the number of random samples.
2. Let z
(r)
1989q1 = z1989q1, z
(r)
1989q2 = z1989q2 ∀r, and compute:
z
(r)
t = Ψb
−1
0

ba+Ψb 1z
(r)
t−1 + Ψb 2z
(r)
t−2 + u
(r)
t

t = 1989q3, ..., 2019q4
3. Use the data computed at point 2 to estimate the bootstrapped coefficients for each
replication:
z
(r)
t = Ψb
−1,(r)
0

ba
(r) + Ψb
(r)
1 z
(r)
t−1 + Ψb
(r)
2 z
(r)
t−2 + u
(r)
t

.
A.4 Additional empirical results
In this section we provide additional supplementary results in support of our empirical
results. Table A.6 provides estimates of AR(1) and AR(2) processes for the sanctions
157

intensity index, st
. As can be seen, an AR(1) model for st
is sufficient for modeling its
persistence and higher order lags are not required. Table A.7 gives the estimates of AR(1)
and AR(2) processes for the world output growth, and shows that the AR(1) specification
used in the paper provides a reasonable approximation.
A.4.1 Reduced form output growth equation including current and
lagged sanction variables
Table A.9 gives the estimates of the reduced form output growth equation given by
Equation (1.5), where we include both current and lagged values of the sanctions intensity
variable, st
. As can be seen, the estimates are very close to the ones presented in the paper,
which only includes st−1. Note that due to the persistence of the st process, when including
st and st−1 in the regressions, one should consider the sum of the coefficients of st and st−1
and its statistical significance.
A.4.2 Re-ordering the variables in the SVAR model
In the main paper, we presented estimates of the SVAR model under our preferred ordering, namely with oil export revenues (∆x
0
t
) included first, followed by the exchange
rate variable (∆ef t), money supply growth (∆mt), inflation (∆pt), and output growth
(∆yt), including the world output growth as control variable. Seasonal dummy variables
were included in all regressions, which proved to be highly significant in the money supply growth equation. Tables A.10 to A.14 display the regression results including a host
of additional control variables, and a number of their sub-sets. As can be seen, the estimates of the effects of sanctions on domestic variables are highly stable and consistent
across all specifications. It is also worth noting that none of the global factors seem to have
any significant impact on Iran’s output growth, partly due to Iran’s relative economic and
financial isolation from the rest of the global economy.
158

Table A.15 includes the estimates of the SVAR model following the same ordering as
our preferred specification in the paper but without money growth, which did not have
any statistically significant impact on the rest of the domestic variables. The results are
almost identical to the ones shown when money growth was included. Tables A.16 to
A.19 present further results to check the robustness of our results.
In Table A.20 we show that including seasonal dummies are not required once money
supply growth is dropped from the SVAR model. Results are in line with the ones in Table
A.15. Therefore, we proceeded with this specification when carrying out IRFs and FEVDs
analyses. Tables A.21 to A.24 provide additional results on the robustness of our main
findings.
Tables A.25 to A.30 present results of the SVAR model given by Equation (1.6) of the
paper with the domestic variables re-ordered by placing the foreign exchange variable
(∆ef t) first, followed by oil export revenues (∆x
0
t
), money supply growth (∆mt), inflation
(∆pt), and output growth (∆yt). The rationale for this ordering is that foreign exchange
is a fast-moving variable, which could move on announcement even before oil exports
are affected by sanctions. The results show that the differences with the ones presented
in the paper are minimal. The only difference appears to be a less precise estimate of
the effect of st−1 on oil export revenues. Also, once we drop the money supply growth
from the model, st−1 become statistically significant again. Table A.31 give the estimates
for the new ordering but without money supply growth, namely (∆ef t, ∆x
0
t
, ∆pt
, ∆yt),
and without seasonal dummies. Further results are provided in Tables A.32 to A.35 for
checking the robustness of the results when the ordering ∆ef t, ∆x
0
t
, ∆pt
, ∆yt
is used.
A.4.3 Using sanctions dummy variables
To investigate the value added of our continuous measure of sanctions intensity, st
, here
we present estimates of the reduced form output growth equation and the SVAR model
using the two alternative sanction dummy variables (namely dt and s
D
t
) considered in
159

Sub-section A.2.4. Table A.36 reports the results for the output growth equation when the
sanctions dummy variable, dt
, is used, and Table A.37 summarizes the results when the
discretized sanctions intensity variable, s
D
t
, is used. These results confirm the negative
effects of sanctions on output growth, but yield less precise estimates as compared to the
results reported in Table 1.3 when st
is used. This is also reflected in the better fit of the
output growth equations with st as compared with the two alternative sanctions dummy
variables.
Table A.38 presents the estimates of the sanctions-augmented SVAR model when the
proposed sanctions intensity estimator (st) is substituted with the sanctions dummy variable (dt). As can be seen, dt and its lagged value are statistically significant in the oil export
revenues equation, but do not show up significant in other equations. The fit of the exchange rate and output equations (as measured by R
2
) are around 0.083 and 0.101, when
dt
is used, as compared to 0.209 and 0.124, when st
is used. It is clear that overall st
is
much better at identifying the effects of sanctions on the Iranian economy as compared to
a (0, 1) dummy variable. See Table 1.4. The estimates based on the discretized version of st
,
namely s
D
t
, are summarized in Table A.39. When using s
D
t
the results are only marginally
better for inflation and output growth as compared to using dt
, but still less favorable as
compared to using st
.
A.4.4 Additional IRFs and FEVDs
Additional results for IRFs and FEVDs analyses are provided in Sub-section A.4.7. Figure A.5 displays the IRFs for one positive standard error shock to the global output growth,
and complements the IRFs results from our baseline SVAR model presented in Figure 1.2
of the paper.8 Following one quarter shock to global output growth, Iran’s oil exports increase by more than 3 per cent in the same quarter. The effects are still of the order of
1 per cent increase two quarters ahead before dissipating about one year after the shock.
8
See also Equation (A.8) in the main paper.
160

Iran’s rial appreciates against U.S. dollar by about 1.5 per cent in the same quarter. However, the results are not particularly persistent and become quantitatively less important
three quarters after the shock. The effects of global output growth on both inflation and
Iran’s output growth, on the other hand, are not statistically significant. These results are
in line with Iran’s relative economic isolation from the main advanced economies. Most
of the global shocks are reflected in the movements of oil export revenues and free market
foreign exchange rate, while domestic factors matter most for inflation and output growth.
As noted earlier, these results are not affected by using the alternative ordering (∆ef t,
∆x
0
t
, ∆pt
, ∆yt). IRFs and FEVDs based on this ordering are given in Figures A.6 and A.7,
and Table A.40, respectively. As can be seen using this ordering of the variables in the
SVAR has little impacts on IRFs and FEVDs. Figure A.8 also shows that the FEVDs results
from a sustained shock to sanctions, keeping all other shocks fixed, is in line with the
results presented in the paper for our base-line model. See Figure 1.3.
A.4.5 Additional results with robust standard errors
The estimates of the reduced form regressions and sanctions-augmented SVAR models
with White’s heteroskedastic consistent standard (see White, 1980) are reported in Subsection A.4.8. As to be expected the use of White’s standard errors results in reduced
level of statistical significance for most of the parameters, but the differences are largely
inconsequential.
161

Table A.6: Quarterly estimates of the sanctions intensity variable AR(1) and AR(2) models over
the period 1989q1–2020q3
st
(1) (2)
st−1 0.743∗∗∗ 0.639∗∗∗
(0.059) (0.089)
st−2 0.139
(0.089)
Constant 0.063∗∗∗ 0.055∗∗∗
(0.018) (0.019)
Adjusted R2 0.551 0.557
S.E. of regression (ωˆs) 0.125 0.125
Notes: Numbers in parentheses are least squares standard errors. ***p < 0.01, **p < 0.05, *p < 0.1. See Sections A.2.1
and A.2.2 in the data appendix for details on the construction of the sanctions intensity variable.
Table A.7: Quarterly estimates of the world real output growth AR(1) and AR(2) models over
the period 1989q1–2019q4
∆ywt
(1) (2)
∆yw,t−1 0.409∗∗∗ 0.393∗∗∗
(0.082) (0.091)
∆yw,t−2 0.038
(0.091)
Constant 0.006∗∗∗ 0.006∗∗∗
(0.001) (0.001)
Adjusted R2 0.161 0.155
Residual Std. Error 0.005 0.005
Notes: ∆ywt is the quarterly world output growth, computed as ywt =
Pn
i=1 wiyit, with {yit}
n
i=1 being the natural log
of real output for 33 major economies, and {wi}
n
i=1 are GDP-PPP weights. See Section A.2.6 in the data appendix for
details on the sources and construction of the data used.
162

Table A.8: Size of one standard error shock for the endogenous variables used in the IRFs analyses in Figure 1.2
Endogenous variable Size of one SE shock
st 0.120
∆x
0
t 0.197
∆ef t 0.083
∆pt 0.015
∆yt 0.029
∆ywt 0.005
163

Table A.9: Reduced form Iran’s output growth equation including contemporaneous sanctions
variable and estimated over the period 1989q1–2019q4
∆yt
(1) (2) (3) (4) (5) (6) (7)
st(βst
) −0.007 −0.008 −0.009 −0.009 −0.009 −0.009 −0.008
(0.023) (0.023) (0.023) (0.023) (0.023) (0.023) (0.024)
st−1(βst−1
) −0.027 −0.026 −0.025 −0.027 −0.027 −0.028 −0.029
(0.024) (0.024) (0.024) (0.024) (0.024) (0.024) (0.024)
∆yt−1(λ∆yt−1
) −0.206∗∗ −0.204∗∗ −0.206∗∗ −0.202∗∗ −0.217∗∗ −0.216∗∗ −0.220∗∗
(0.092) (0.092) (0.093) (0.093) (0.092) (0.092) (0.093)
∆x
0
t−1
0.016 0.015 0.016 0.017 0.014 0.014 0.015
(0.014) (0.014) (0.014) (0.014) (0.014) (0.014) (0.014)
∆ef,t−1 −0.005 −0.004 −0.005 −0.001 0.003 0.003 0.002
(0.033) (0.034) (0.034) (0.034) (0.034) (0.034) (0.034)
∆mt−1 −0.030 −0.039 −0.044 −0.035 −0.056 −0.059 −0.065
(0.101) (0.103) (0.104) (0.105) (0.104) (0.105) (0.106)
∆pt−1 −0.236∗ −0.230∗ −0.228∗ −0.242∗ −0.263∗∗ −0.269∗∗ −0.270∗∗
(0.122) (0.124) (0.124) (0.125) (0.124) (0.126) (0.126)
∆ywt 0.245 0.170 0.224 −0.120 −0.151 −0.111
(0.558) (0.605) (0.606) (0.627) (0.638) (0.646)
∆reqwt 0.015 0.023 0.015 0.004 0.002
(0.045) (0.046) (0.046) (0.058) (0.058)
∆rwt −4.500 −4.291 −4.447 −3.537
(4.157) (4.113) (4.160) (4.632)
∆ewt −0.279∗ −0.273∗ −0.307∗
(0.149) (0.151) (0.169)
grvt −0.036 −0.042
(0.115) (0.116)
∆p
0
t −0.011
(0.025)
βst + βst−1 −0.035∗∗ −0.034∗∗ −0.034∗∗ −0.036∗∗ −0.037∗∗ −0.037∗∗ −0.036∗∗
(0.017) (0.017) (0.017) (0.017) (0.017) (0.017) (0.017)
(βst + βst−1
)/(1 −
λ∆yt−1
)
−0.029∗∗ −0.028∗∗ −0.028∗∗ −0.030∗∗ −0.030∗∗ −0.030∗∗ −0.030∗∗
(0.014) (0.014) (0.014) (0.014) (0.014) (0.014) (0.014)
Adjusted R2 0.076 0.069 0.062 0.064 0.084 0.077 0.070
Notes: ∆yt = ln(Yt/Yt−1), Yt is Iran’s real output. st is the sanctions intensity variable. βst
, βst−1
and λ∆yt−1
are the
coefficients of st, st−1 and ∆yt−1, respectively; (βst + βst−1
)/(1− λ∆yt−1
) represents the long run effect of sanctions on
output growth. See Chapter 6 of Pesaran (2015). See the notes to Table 1.3 for further details on the sources and construction of data used. ∆x
0
t = (X
0
t −X
0
t−1)/X0
t−1, X0
t
is the oil exports revenues in U.S. dollars; ∆eft = ln(Eft/Ef,t−1),
Eft is the rial/U.S. dollar free market exchange rate; ∆mt = (M2t −M2,t−1)/M2,t−1, M2t is the monetary aggregate M2
obtained by summing the aggregates M1 and "quasi-money"; ∆pt = ln(Pt/Pt−1), Pt is Iran’s consumer price index.
∆ywt is the world output growth, computed as ywt =
Pn
i=1 wiyit, with {yit}
n
i=1 being the natural log of real output
for 33 major economies, and {wi}
n
i=1 are GDP-PPP weights. ∆reqwt is the rate of change of the global real equity price
index: reqwt =
Pn
i=1 wireqit, reqit is the natural log of the real equity price of country i in quarter t. ∆rwt is the
per cent change of the global nominal long term interest rate: rwt =
Pn
i=1 wirit, rit is the long term nominal interest rate of country i in quarter t. ∆ewt is the rate of change of the global real exchange rate vis-à-vis the U.S. dollar:
ewt =
Pn
i=1 wieit, eit is the natural log of the real exchange rate of country i in quarter t. grvt is the global realized
volatility. ∆p
0
t = ln(P
0
t /P0
t−1), P0
t
is the oil price (Brent crude). Numbers in parentheses are least squares standard
errors. ***p < 0.01, **p < 0.05, *p < 0.1. See Sections A.2.1, A.2.2, A.2.5, and A.2.6 in the data appendix for details on
the construction of the sanctions intensity variable, calendar conversions, and sources of the data used.
164

A.4.6 Additional sanctions-augmented SVAR analyses
Table A.10: Quarterly estimates of the equation for the oil export variable in the SVAR model
of Iran with domestic variables ordered as: oil exports, exchange rate returns, money supply
growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆x
0
t
(1) (2) (3) (4) (5) (6)
st 0.144 0.107 0.104 0.087 0.088 0.088
(0.152) (0.150) (0.151) (0.143) (0.140) (0.140)
st−1 −0.339∗∗ −0.288∗ −0.286∗ −0.279∗ −0.261∗ −0.261∗
(0.156) (0.155) (0.156) (0.147) (0.144) (0.145)
∆x
0
t−1 −0.035 −0.051 −0.050 −0.077 −0.088 −0.088
(0.092) (0.090) (0.091) (0.086) (0.084) (0.085)
∆ef,t−1 −0.442∗∗ −0.441∗∗ −0.443∗∗ −0.382∗ −0.432∗∗ −0.432∗∗
(0.221) (0.217) (0.219) (0.207) (0.203) (0.204)
∆mt−1 −0.128 −0.715 −0.744 −1.214 −1.141 −1.143
(0.911) (0.930) (0.947) (0.903) (0.882) (0.893)
∆pt−1 −0.156 0.052 0.060 −0.213 −0.041 −0.043
(0.804) (0.794) (0.799) (0.758) (0.743) (0.755)
∆yt−1 0.087 0.122 0.116 −0.062 −0.103 −0.103
(0.603) (0.592) (0.595) (0.564) (0.551) (0.554)
∆ywt 8.406∗∗ 8.132∗∗ 4.185 3.098 3.085
(3.649) (3.948) (3.871) (3.803) (3.862)
∆reqwt 0.056 −0.046 −0.154 −0.158
(0.297) (0.282) (0.279) (0.351)
∆ewt −3.507∗∗∗ −3.551∗∗∗ −3.548∗∗∗
(0.923) (0.901) (0.911)
∆rwt 63.486∗∗ 63.416∗∗
(25.058) (25.373)
grvt −0.015
(0.690)
Residual serial 1.202 2.406 2.446 2.382 5.176 5.166
correlation test [0.878] [0.662] [0.654] [0.666] [0.270] [0.271]
Adjusted R2 0.089 0.122 0.115 0.210 0.247 0.240
Notes: The variables are ordered as: ∆x
0
t
, ∆eft, ∆mt, ∆pt, and ∆yt, where: ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the
oil exports revenues in U.S. dollars; ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly rial/U.S. dollar free market exchange
rate; ∆mt = (M2t − M2,t−1)/M2,t−1, M2t is the monetary aggregate M2 obtained by summing the aggregates M1
and "quasi-money"; ∆pt = ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran; ∆yt = ln(Yt/Yt−1), Yt is the
quarterly real output of Iran. Seasonal dummies are included to allow for possible seasonality of the variables in the
regressions of the SVAR model in Equation (1.6) with qt =

∆x
0
t
, ∆eft, ∆mt, ∆pt, ∆yt
0
. Numbers in parentheses are
least squares standard errors, and those in square brackets are p-values. ***p < 0.01, **p < 0.05, *p < 0.1. "Residual
serial correlation test" is the Breusch–Godfrey LM test of serially uncorrelated errors with lag order of the test set to 4.
See the notes to Table A.9 for further details on the sources and construction of data used.
16

Table A.11: Quarterly estimates of the equation for exchange rate returns in the SVAR model
of Iran with domestic variables ordered as: oil exports, exchange rate returns, money supply
growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆ef t
(1) (2) (3) (4) (5) (6)
st 0.295∗∗∗ 0.305∗∗∗ 0.302∗∗∗ 0.303∗∗∗ 0.303∗∗∗ 0.307∗∗∗
(0.064) (0.064) (0.065) (0.064) (0.065) (0.064)
st−1 −0.221∗∗∗ −0.233∗∗∗ −0.230∗∗∗ −0.226∗∗∗ −0.226∗∗∗ −0.231∗∗∗
(0.067) (0.067) (0.067) (0.067) (0.067) (0.067)
∆x
0
t 0.014 0.029 0.028 0.047 0.048 0.048
(0.040) (0.040) (0.040) (0.043) (0.044) (0.044)
∆x
0
t−1
0.038 0.044 0.045 0.051 0.051 0.047
(0.039) (0.038) (0.039) (0.039) (0.039) (0.039)
∆ef,t−1 0.344∗∗∗ 0.350∗∗∗ 0.348∗∗∗ 0.346∗∗∗ 0.348∗∗∗ 0.346∗∗∗
(0.095) (0.094) (0.095) (0.095) (0.096) (0.095)
∆mt−1 −0.037 0.149 0.121 0.209 0.209 0.110
(0.384) (0.397) (0.404) (0.409) (0.411) (0.408)
∆pt−1 −0.278 −0.341 −0.333 −0.291 −0.295 −0.397
(0.339) (0.338) (0.340) (0.341) (0.344) (0.343)
∆yt−1 −0.133 −0.145 −0.151 −0.125 −0.124 −0.120
(0.254) (0.252) (0.253) (0.254) (0.255) (0.251)
∆ywt −2.639∗ −2.897∗ −2.423 −2.401 −2.907
(1.590) (1.712) (1.749) (1.764) (1.759)
∆reqwt 0.053 0.068 0.071 −0.121
(0.127) (0.127) (0.129) (0.160)
∆ewt 0.554 0.560 0.648
(0.441) (0.445) (0.442)
∆rwt −1.633 −4.492
(11.923) (11.849)
grvt −0.625∗∗
(0.313)
Residual serial 5.972 6.212 5.961 7.940 8.059 5.259
correlation test [0.201] [0.184] [0.202] [0.094] [0.089] [0.262]
Adjusted R2 0.196 0.209 0.203 0.207 0.200 0.221
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
166

Table A.12: Quarterly estimates of the equation for money supply growth in the SVAR model
of Iran with domestic variables ordered as: oil exports, exchange rate returns, money supply
growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆mt
(1) (2) (3) (4) (5) (6)
st −0.0004 −0.002 −0.003 −0.002 −0.002 0.002
(0.017) (0.017) (0.017) (0.017) (0.017) (0.017)
st−1 0.014 0.015 0.016 0.017 0.017 0.014
(0.017) (0.017) (0.017) (0.017) (0.017) (0.017)
∆x
0
t 0.008 0.006 0.006 0.011 0.014 0.015
(0.009) (0.010) (0.010) (0.010) (0.011) (0.010)
∆ef,t −0.009 −0.007 −0.008 −0.011 −0.012 −0.022
(0.023) (0.023) (0.023) (0.023) (0.023) (0.023)
∆x
0
t−1 −0.005 −0.005 −0.005 −0.003 −0.002 −0.003
(0.009) (0.009) (0.009) (0.009) (0.009) (0.009)
∆ef,t−1 −0.024 −0.025 −0.026 −0.027 −0.023 −0.021
(0.026) (0.027) (0.027) (0.027) (0.027) (0.026)
∆mt−1 0.235∗∗ 0.218∗∗ 0.211∗∗ 0.236∗∗ 0.236∗∗ 0.211∗∗
(0.092) (0.096) (0.098) (0.099) (0.099) (0.098)
∆pt−1 0.165 0.167 0.174 0.193 0.189 0.163
(0.115) (0.115) (0.117) (0.117) (0.117) (0.115)
∆yt−1 0.022 0.025 0.022 0.026 0.028 0.026
(0.063) (0.063) (0.063) (0.063) (0.063) (0.062)
∆pt−2 −0.076 −0.070 −0.077 −0.088 −0.094 −0.102
(0.103) (0.104) (0.105) (0.105) (0.105) (0.103)
∆ywt 0.233 0.152 0.260 0.304 0.134
(0.389) (0.421) (0.427) (0.427) (0.425)
∆reqwt 0.016 0.021 0.027 −0.026
(0.031) (0.031) (0.031) (0.038)
∆ewt 0.146 0.160 0.191∗
(0.107) (0.108) (0.106)
∆rwt −3.581 −4.412
(2.850) (2.815)
grvt −0.176∗∗
(0.076)
Residual serial 7.428 7.640 7.255 6.129 5.742 4.178
correlation test [0.115] [0.106] [0.123] [0.190] [0.219] [0.382]
Adjusted R2 0.469 0.466 0.462 0.467 0.470 0.491
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
167

Table A.13: Quarterly estimates of the equation for inflation in the SVAR model of Iran with domestic variables ordered as: oil exports, exchange rate returns, money supply growth, inflation,
and output growth, estimated over the period 1989q1-2019q4
∆pt
(1) (2) (3) (4) (5) (6)
st −0.028∗∗ −0.033∗∗ −0.032∗∗ −0.033∗∗ −0.033∗∗ −0.032∗∗
(0.013) (0.013) (0.013) (0.013) (0.013) (0.013)
st−1 0.032∗∗ 0.037∗∗∗ 0.036∗∗∗ 0.036∗∗∗ 0.036∗∗∗ 0.035∗∗
(0.014) (0.013) (0.013) (0.013) (0.013) (0.013)
∆x
0
t 0.001 −0.003 −0.003 −0.007 −0.007 −0.007
(0.008) (0.007) (0.007) (0.008) (0.008) (0.008)
∆ef,t 0.155∗∗∗ 0.163∗∗∗ 0.164∗∗∗ 0.166∗∗∗ 0.166∗∗∗ 0.163∗∗∗
(0.018) (0.017) (0.018) (0.018) (0.018) (0.018)
∆mt −0.061 −0.073 −0.070 −0.058 −0.057 −0.070
(0.076) (0.073) (0.073) (0.074) (0.075) (0.077)
∆x
0
t−1 −0.001 −0.003 −0.003 −0.004 −0.004 −0.005
(0.007) (0.007) (0.007) (0.007) (0.007) (0.007)
∆ef,t−1 −0.007 −0.009 −0.007 −0.007 −0.007 −0.007
(0.021) (0.020) (0.020) (0.020) (0.021) (0.021)
∆mt−1 0.035 −0.025 −0.016 −0.036 −0.036 −0.040
(0.075) (0.075) (0.076) (0.078) (0.078) (0.078)
∆pt−1 0.480∗∗∗ 0.488∗∗∗ 0.478∗∗∗ 0.462∗∗∗ 0.462∗∗∗ 0.458∗∗∗
(0.092) (0.089) (0.090) (0.090) (0.091) (0.091)
∆yt−1 0.033 0.042 0.045 0.042 0.042 0.042
(0.050) (0.048) (0.048) (0.048) (0.049) (0.049)
∆pt−2 0.162∗∗ 0.183∗∗ 0.192∗∗ 0.201∗∗ 0.202∗∗ 0.198∗∗
(0.082) (0.079) (0.080) (0.080) (0.081) (0.081)
∆ywt 0.865∗∗∗ 0.971∗∗∗ 0.893∗∗∗ 0.889∗∗∗ 0.847∗∗
(0.298) (0.322) (0.327) (0.330) (0.335)
∆reqwt −0.021 −0.024 −0.025 −0.039
(0.023) (0.024) (0.024) (0.030)
∆ewt −0.102 −0.104 −0.093
(0.083) (0.084) (0.085)
∆rwt 0.303 0.031
(2.210) (2.242)
grvt −0.048
(0.061)
Residual serial 9.241 8.061 5.714 6.473 6.510 6.759
correlation test [0.055] [0.089] [0.222] [0.166] [0.164] [0.149]
Adjusted R2 0.635 0.659 0.658 0.660 0.656 0.655
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
168

Table A.14: Quarterly estimates of the equation for output growth in the SVAR model of Iran
with domestic variables ordered as: oil exports, exchange rate returns, money supply growth,
inflation, and output growth, estimated over the period 1989q1-2019q4
∆yt
(1) (2) (3) (4) (5) (6)
st 0.024 0.029 0.028 0.026 0.026 0.027
(0.025) (0.026) (0.026) (0.026) (0.026) (0.026)
st−1 −0.051∗∗ −0.056∗∗ −0.055∗∗ −0.055∗∗ −0.055∗∗ −0.056∗∗
(0.026) (0.026) (0.026) (0.026) (0.026) (0.026)
∆x
0
t 0.023 0.025∗ 0.025∗ 0.020 0.025 0.026
(0.014) (0.014) (0.014) (0.015) (0.016) (0.016)
∆ef,t −0.130∗∗∗ −0.141∗∗∗ −0.142∗∗∗ −0.135∗∗∗ −0.136∗∗∗ −0.138∗∗∗
(0.043) (0.045) (0.045) (0.046) (0.046) (0.046)
∆mt 0.052 0.063 0.062 0.078 0.054 0.037
(0.141) (0.142) (0.143) (0.143) (0.144) (0.148)
∆pt 0.348∗ 0.387∗∗ 0.390∗∗ 0.373∗∗ 0.373∗∗ 0.364∗∗
(0.175) (0.181) (0.182) (0.183) (0.182) (0.183)
∆x
0
t−1
0.022 0.023∗ 0.024∗ 0.022 0.023∗ 0.023
(0.014) (0.014) (0.014) (0.014) (0.014) (0.014)
∆ef,t−1 0.037 0.041 0.041 0.040 0.046 0.047
(0.036) (0.036) (0.036) (0.036) (0.036) (0.036)
∆mt−1 0.013 0.046 0.039 0.009 0.014 0.009
(0.139) (0.144) (0.147) (0.150) (0.149) (0.150)
∆pt−1 −0.466∗∗∗ −0.505∗∗∗ −0.505∗∗∗ −0.506∗∗∗ −0.519∗∗∗ −0.523∗∗∗
(0.161) (0.167) (0.168) (0.168) (0.167) (0.168)
∆yt−1 −0.218∗∗ −0.221∗∗ −0.223∗∗ −0.230∗∗ −0.225∗∗ −0.224∗∗
(0.090) (0.090) (0.090) (0.091) (0.090) (0.091)
∆ywt −0.520 −0.595 −0.708 −0.619 −0.664
(0.592) (0.637) (0.647) (0.647) (0.655)
∆reqwt 0.015 0.010 0.021 0.003
(0.045) (0.045) (0.046) (0.058)
∆ewt −0.160 −0.133 −0.121
(0.160) (0.161) (0.163)
∆rwt −6.160 −6.499
(4.239) (4.305)
grvt −0.060
(0.118)
Residual serial 7.242 7.240 7.371 7.721 8.049 8.248
correlation test [0.124] [0.124] [0.118] [0.102] [0.090] [0.083]
Adjusted R2 0.126 0.124 0.117 0.117 0.126 0.120
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
169

Table A.15: Quarterly estimates of the SVAR model of Iran with domestic variables ordered
as: oil exports, exchange rate returns, inflation, and output growth, estimated over the period
1989q1-2019q4
∆x
0
t ∆ef,t ∆pt ∆yt
(1) (2) (3) (4)
st 0.119 0.302∗∗∗ −0.033∗∗ 0.027
(0.149) (0.063) (0.013) (0.026)
st−1 −0.308∗∗ −0.229∗∗∗ 0.035∗∗∗ −0.053∗∗
(0.152) (0.066) (0.013) (0.026)
∆x
0
t 0.028 −0.003 0.025∗
(0.040) (0.007) (0.014)
∆ef,t 0.163∗∗∗ −0.139∗∗∗
(0.017) (0.045)
∆pt 0.377∗∗
(0.179)
∆ywt 7.638∗∗ −2.471 0.800∗∗∗ −0.428
(3.503) (1.520) (0.284) (0.561)
∆x
0
t−1 −0.053 0.044 −0.003 0.023∗
(0.090) (0.038) (0.007) (0.014)
∆ef,t−1 −0.429∗∗ 0.347∗∗∗ −0.007 0.038
(0.216) (0.094) (0.020) (0.035)
∆pt−1 0.022 −0.335 0.478∗∗∗ −0.489∗∗∗
(0.792) (0.337) (0.088) (0.163)
∆yt−1 0.132 −0.147 0.039 −0.220∗∗
(0.591) (0.251) (0.048) (0.089)
∆pt−2 0.182∗∗
(0.078)
Residual serial 2.027 5.689 7.970 6.703
correlation test [0.731] [0.224] [0.093] [0.152]
Adjusted R2 0.126 0.215 0.661 0.137
Notes: The variables are ordered as: ∆x
0
t
, ∆eft, ∆pt, and ∆yt, where: ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the oil exports
revenues in U.S. dollars; ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly rial/U.S. dollar free market exchange rate; ∆pt =
ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran; ∆yt = ln(Yt/Yt−1), Yt is the quarterly real output of Iran.
st is the quarterly sanctions intensity variable. Seasonal dummies are included to allow for possible seasonality of the
variables in the regressions of the SVAR model in Equation (1.6) with qt =

∆x
0
t
, ∆eft, ∆pt, ∆yt
0
and zwt = (∆ywt)
0
.
Numbers in parentheses are least squares standard errors, and those in square brackets are p-values. ***p < 0.01,
**p < 0.05, *p < 0.1. "Residual serial correlation test" is the Breusch–Godfrey LM test of serially uncorrelated errors
with lag order of the test set to 4.
See the notes to Table A.9 for further details on the sources and construction of data used.
17

Table A.16: Quarterly estimates of the equation for the oil exports variable in the SVAR model of
Iran with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆x
0
t
(1) (2) (3) (4) (5) (6)
st 0.146 0.119 0.119 0.112 0.111 0.111
(0.151) (0.149) (0.150) (0.142) (0.139) (0.140)
st−1 −0.342∗∗ −0.308∗∗ −0.307∗∗ −0.313∗∗ −0.294∗∗ −0.293∗∗
(0.154) (0.152) (0.153) (0.145) (0.142) (0.143)
∆x
0
t−1 −0.035 −0.053 −0.053 −0.080 −0.091 −0.091
(0.091) (0.090) (0.091) (0.086) (0.084) (0.085)
∆ef,t−1 −0.440∗∗ −0.429∗∗ −0.429∗ −0.363∗ −0.414∗∗ −0.414∗∗
(0.220) (0.216) (0.218) (0.207) (0.203) (0.204)
∆pt−1 −0.158 0.022 0.024 −0.257 −0.079 −0.064
(0.801) (0.792) (0.796) (0.760) (0.745) (0.757)
∆yt−1 0.089 0.132 0.131 −0.030 −0.074 −0.075
(0.600) (0.591) (0.594) (0.566) (0.552) (0.555)
∆ywt 7.638∗∗ 7.545∗ 3.436 2.377 2.461
(3.503) (3.870) (3.845) (3.774) (3.842)
∆reqwt 0.017 −0.102 −0.209 −0.179
(0.293) (0.280) (0.276) (0.352)
∆ewt −3.337∗∗∗ −3.391∗∗∗ −3.407∗∗∗
(0.918) (0.896) (0.907)
∆rwt 64.555∗∗ 64.968∗∗
(25.121) (25.418)
grvt 0.093
(0.687)
Residual serial 1.157 2.027 2.052 1.565 3.775 3.937
correlation test [0.885] [0.731] [0.726] [0.815] [0.437] [0.415]
Adjusted R2 0.097 0.126 0.118 0.205 0.243 0.236
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
171

Table A.17: Quarterly estimates of the equation for exchange rate returns in the SVAR model of
Iran with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆ef t
(1) (2) (3) (4) (5) (6)
st 0.296∗∗∗ 0.302∗∗∗ 0.300∗∗∗ 0.299∗∗∗ 0.299∗∗∗ 0.305∗∗∗
(0.064) (0.063) (0.064) (0.064) (0.064) (0.063)
st−1 −0.222∗∗∗ −0.229∗∗∗ −0.227∗∗∗ −0.221∗∗∗ −0.221∗∗∗ −0.228∗∗∗
(0.066) (0.066) (0.066) (0.066) (0.066) (0.066)
∆x
0
t 0.014 0.028 0.027 0.044 0.045 0.046
(0.039) (0.040) (0.040) (0.042) (0.044) (0.043)
∆x
0
t−1
0.038 0.044 0.046 0.051 0.051 0.047
(0.038) (0.038) (0.039) (0.039) (0.039) (0.038)
∆ef,t−1 0.345∗∗∗ 0.347∗∗∗ 0.345∗∗∗ 0.342∗∗∗ 0.344∗∗∗ 0.344∗∗∗
(0.094) (0.094) (0.094) (0.094) (0.095) (0.094)
∆pt−1 −0.278 −0.335 −0.327 −0.284 −0.289 −0.395
(0.337) (0.337) (0.338) (0.339) (0.342) (0.341)
∆yt−1 −0.132 −0.147 −0.153 −0.130 −0.129 −0.123
(0.253) (0.251) (0.252) (0.253) (0.254) (0.250)
∆ywt −2.471 −2.795∗ −2.284 −2.262 −2.843
(1.520) (1.671) (1.722) (1.737) (1.735)
∆reqwt 0.059 0.077 0.080 −0.120
(0.124) (0.125) (0.127) (0.159)
∆ewt 0.515 0.522 0.629
(0.433) (0.438) (0.434)
∆rwt −1.651 −4.548
(11.882) (11.796)
grvt −0.636∗∗
(0.310)
Residual serial 6.132 5.689 5.541 7.009 7.125 4.742
correlation test [0.190] [0.224] [0.236] [0.135] [0.129] [0.315]
Adjusted R2 0.203 0.215 0.209 0.212 0.205 0.228
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
172

Table A.18: Quarterly estimates of the equation for inflation in the SVAR model of Iran with
domestic variables ordered as: oil exports, exchange rate returns, inflation, and output growth,
estimated over the period 1989q1-2019q4
∆pt
(1) (2) (3) (4) (5) (6)
st −0.029∗∗ −0.033∗∗ −0.032∗∗ −0.032∗∗ −0.032∗∗ −0.031∗∗
(0.013) (0.013) (0.013) (0.013) (0.013) (0.013)
st−1 0.032∗∗ 0.035∗∗∗ 0.034∗∗∗ 0.034∗∗ 0.034∗∗ 0.033∗∗
(0.013) (0.013) (0.013) (0.013) (0.013) (0.013)
∆x
0
t 0.001 −0.003 −0.003 −0.007 −0.007 −0.007
(0.007) (0.007) (0.007) (0.008) (0.008) (0.008)
∆ef,t 0.156∗∗∗ 0.163∗∗∗ 0.164∗∗∗ 0.166∗∗∗ 0.166∗∗∗ 0.165∗∗∗
(0.018) (0.017) (0.017) (0.018) (0.018) (0.018)
∆x
0
t−1 −0.0003 −0.003 −0.003 −0.004 −0.004 −0.004
(0.007) (0.007) (0.007) (0.007) (0.007) (0.007)
∆ef,t−1 −0.005 −0.007 −0.005 −0.005 −0.005 −0.005
(0.021) (0.020) (0.020) (0.020) (0.020) (0.021)
∆pt−1 0.468∗∗∗ 0.478∗∗∗ 0.467∗∗∗ 0.454∗∗∗ 0.455∗∗∗ 0.451∗∗∗
(0.090) (0.088) (0.089) (0.089) (0.089) (0.090)
∆yt−1 0.032 0.039 0.044 0.041 0.040 0.040
(0.049) (0.048) (0.048) (0.048) (0.048) (0.048)
∆pt−2 0.169∗∗ 0.182∗∗ 0.194∗∗ 0.200∗∗ 0.201∗∗ 0.199∗∗
(0.081) (0.078) (0.079) (0.079) (0.080) (0.080)
∆ywt 0.800∗∗∗ 0.932∗∗∗ 0.842∗∗∗ 0.836∗∗ 0.803∗∗
(0.284) (0.313) (0.320) (0.323) (0.330)
∆reqwt −0.023 −0.027 −0.028 −0.038
(0.023) (0.023) (0.024) (0.030)
∆ewt −0.101 −0.103 −0.097
(0.080) (0.081) (0.082)
∆rwt 0.501 0.356
(2.184) (2.209)
grvt −0.031
(0.059)
Residual serial 11.263 7.970 5.559 6.203 6.210 6.281
correlation test [0.024] [0.093] [0.235] [0.184] [0.184] [0.179]
Adjusted R2 0.640 0.661 0.661 0.663 0.660 0.657
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
173

Table A.19: Quarterly estimates of the equation for output growth in the SVAR model of Iran
with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆yt
(1) (2) (3) (4) (5) (6)
st 0.024 0.027 0.027 0.025 0.025 0.027
(0.025) (0.026) (0.026) (0.026) (0.026) (0.026)
st−1 −0.050∗ −0.053∗∗ −0.053∗∗ −0.053∗∗ −0.053∗∗ −0.055∗∗
(0.025) (0.026) (0.026) (0.026) (0.026) (0.026)
∆x
0
t 0.023 0.025∗ 0.025∗ 0.020 0.026∗ 0.026∗
(0.014) (0.014) (0.014) (0.015) (0.016) (0.016)
∆ef,t −0.130∗∗∗ −0.139∗∗∗ −0.140∗∗∗ −0.133∗∗∗ −0.135∗∗∗ −0.138∗∗∗
(0.043) (0.045) (0.045) (0.045) (0.045) (0.045)
∆pt 0.344∗∗ 0.377∗∗ 0.381∗∗ 0.362∗∗ 0.365∗∗ 0.359∗∗
(0.173) (0.179) (0.180) (0.181) (0.180) (0.181)
∆x
0
t−1
0.022 0.023∗ 0.023∗ 0.022 0.023∗ 0.023∗
(0.014) (0.014) (0.014) (0.014) (0.014) (0.014)
∆ef,t−1 0.036 0.038 0.038 0.037 0.045 0.046
(0.035) (0.035) (0.036) (0.036) (0.036) (0.036)
∆pt−1 −0.458∗∗∗ −0.489∗∗∗ −0.489∗∗∗ −0.489∗∗∗ −0.507∗∗∗ −0.516∗∗∗
(0.158) (0.163) (0.164) (0.164) (0.163) (0.165)
∆yt−1 −0.216∗∗ −0.220∗∗ −0.221∗∗ −0.227∗∗ −0.223∗∗ −0.223∗∗
(0.089) (0.089) (0.090) (0.090) (0.089) (0.090)
∆ywt −0.428 −0.531 −0.658 −0.577 −0.644
(0.561) (0.618) (0.631) (0.630) (0.641)
∆reqwt 0.018 0.012 0.023 0.002
(0.044) (0.045) (0.045) (0.057)
∆ewt −0.155 −0.131 −0.118
(0.156) (0.156) (0.157)
∆rwt −6.346 −6.663
(4.174) (4.219)
grvt −0.068
(0.113)
Residual serial 6.974 6.703 6.911 7.426 7.684 8.064
correlation test [0.137] [0.152] [0.141] [0.115] [0.104] [0.089]
Adjusted R2 0.141 0.137 0.131 0.131 0.141 0.136
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
174

Table A.20: Quarterly estimates of the SVAR model of Iran with domestic variables ordered
as: oil exports, exchange rate returns, inflation, and output growth, estimated over the period
1989q1-2019q4
∆x
0
t ∆ef,t ∆pt ∆yt
(1) (2) (3) (4)
st 0.111 0.309∗∗∗ −0.033∗∗ 0.028
(0.151) (0.063) (0.013) (0.025)
st−1 −0.305∗∗ −0.235∗∗∗ 0.036∗∗∗ −0.054∗∗
(0.154) (0.065) (0.013) (0.025)
∆x
0
t 0.018 −0.004 0.028∗∗
(0.039) (0.007) (0.014)
∆ef,t 0.162∗∗∗ −0.138∗∗∗
(0.017) (0.044)
∆pt 0.364∗∗
(0.177)
∆ywt 7.674∗∗ −2.452 0.808∗∗∗ −0.459
(3.556) (1.517) (0.280) (0.556)
∆x
0
t−1 −0.063 0.041 −0.002 0.021
(0.090) (0.038) (0.007) (0.013)
∆ef,t−1 −0.361 0.332∗∗∗ −0.006 0.041
(0.218) (0.092) (0.020) (0.034)
∆pt−1 −0.059 −0.338 0.477∗∗∗ −0.482∗∗∗
(0.803) (0.336) (0.086) (0.162)
∆yt−1 0.125 −0.135 0.040 −0.223∗∗
(0.600) (0.251) (0.047) (0.088)
∆pt−2 0.184∗∗
(0.077)
Residual serial 3.751 4.983 8.003 6.738
correlation test [0.441] [0.289] [0.091] [0.150]
Adjusted R2 0.097 0.214 0.668 0.152
Notes: The variables are ordered as: ∆x
0
t
, ∆eft, ∆pt, and ∆yt, where: ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the oil exports
revenues in U.S. dollars; ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly rial/U.S. dollar free market exchange rate; ∆pt =
ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran; ∆yt = ln(Yt/Yt−1), Yt is the quarterly real output of Iran.
st is the quarterly sanctions intensity variable. Seasonal dummies are not included to allow for possible seasonality
of the variables in the SVAR model. Numbers in parentheses are least squares standard errors, and those in square
brackets are p-values. ***p < 0.01, **p < 0.05, *p < 0.1. "Residual serial correlation test" is the Breusch–Godfrey LM
test of serially uncorrelated errors with lag order of the test set to 4.
See the notes to Table A.9 for further details on the sources and construction of data used.
175

Table A.21: Quarterly estimates of the equation for the oil exports variable in the SVAR model of
Iran with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆x
0
t
(1) (2) (3) (4) (5) (6)
st 0.136 0.111 0.109 0.106 0.104 0.104
(0.153) (0.151) (0.152) (0.144) (0.140) (0.141)
st−1 −0.338∗∗ −0.305∗∗ −0.304∗ −0.316∗∗ −0.295∗∗ −0.294∗∗
(0.155) (0.154) (0.155) (0.147) (0.143) (0.144)
∆x
0
t−1 −0.043 −0.063 −0.061 −0.094 −0.103 −0.103
(0.091) (0.090) (0.091) (0.087) (0.084) (0.085)
∆ef,t−1 −0.370∗ −0.361 −0.362 −0.305 −0.364∗ −0.363∗
(0.221) (0.218) (0.219) (0.208) (0.204) (0.205)
∆pt−1 −0.243 −0.059 −0.054 −0.327 −0.123 −0.116
(0.811) (0.803) (0.808) (0.768) (0.752) (0.764)
∆yt−1 0.083 0.125 0.121 −0.056 −0.107 −0.108
(0.609) (0.600) (0.603) (0.573) (0.558) (0.561)
∆ywt 7.674∗∗ 7.463∗ 3.031 2.068 2.109
(3.556) (3.904) (3.881) (3.796) (3.866)
∆reqwt 0.039 −0.069 −0.196 −0.182
(0.292) (0.278) (0.275) (0.350)
∆ewt −3.469∗∗∗ −3.514∗∗∗ −3.522∗∗∗
(0.922) (0.898) (0.910)
∆rwt 66.900∗∗∗ 67.105∗∗∗
(24.886) (25.204)
grvt 0.044
(0.692)
Residual serial 2.435 3.751 3.865 2.464 4.895 5.026
correlation test [0.656] [0.441] [0.425] [0.651] [0.298] [0.285]
Adjusted R2 0.068 0.097 0.089 0.183 0.225 0.218
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
176

Table A.22: Quarterly estimates of the equation for exchange rate returns in the SVAR model of
Iran with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆ef t
(1) (2) (3) (4) (5) (6)
st 0.302∗∗∗ 0.309∗∗∗ 0.305∗∗∗ 0.304∗∗∗ 0.304∗∗∗ 0.310∗∗∗
(0.063) (0.063) (0.064) (0.064) (0.064) (0.063)
st−1 −0.229∗∗∗ −0.235∗∗∗ −0.232∗∗∗ −0.226∗∗∗ −0.226∗∗∗ −0.233∗∗∗
(0.066) (0.065) (0.066) (0.066) (0.066) (0.065)
∆x
0
t 0.005 0.018 0.017 0.034 0.033 0.033
(0.038) (0.039) (0.039) (0.041) (0.043) (0.042)
∆x
0
t−1
0.035 0.041 0.044 0.050 0.050 0.045
(0.038) (0.038) (0.038) (0.038) (0.039) (0.038)
∆ef,t−1 0.331∗∗∗ 0.332∗∗∗ 0.330∗∗∗ 0.327∗∗∗ 0.326∗∗∗ 0.324∗∗∗
(0.093) (0.092) (0.093) (0.092) (0.094) (0.093)
∆pt−1 −0.282 −0.338 −0.329 −0.287 −0.284 −0.384
(0.337) (0.336) (0.337) (0.338) (0.341) (0.341)
∆yt−1 −0.121 −0.135 −0.142 −0.118 −0.119 −0.114
(0.252) (0.251) (0.252) (0.252) (0.253) (0.250)
∆ywt −2.452 −2.855∗ −2.319 −2.333 −2.908∗
(1.517) (1.656) (1.713) (1.726) (1.726)
∆reqwt 0.075 0.090 0.088 −0.104
(0.122) (0.123) (0.125) (0.156)
∆ewt 0.515 0.510 0.624
(0.430) (0.434) (0.432)
∆rwt 1.170 −1.757
(11.653) (11.590)
grvt −0.621∗∗
(0.309)
Residual serial 5.353 4.983 4.839 6.292 6.236 4.234
correlation test [0.253] [0.289] [0.304] [0.178] [0.182] [0.375]
Adjusted R2 0.203 0.214 0.209 0.212 0.205 0.226
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
177

Table A.23: Quarterly estimates of the equation for inflation in the SVAR model of Iran with
domestic variables ordered as: oil exports, exchange rate returns, inflation, and output growth,
estimated over the period 1989q1-2019q4
∆pt
(1) (2) (3) (4) (5) (6)
st −0.029∗∗ −0.033∗∗ −0.032∗∗ −0.032∗∗ −0.032∗∗ −0.031∗∗
(0.013) (0.013) (0.013) (0.013) (0.013) (0.013)
st−1 0.032∗∗ 0.036∗∗∗ 0.035∗∗∗ 0.034∗∗ 0.034∗∗ 0.033∗∗
(0.013) (0.013) (0.013) (0.013) (0.013) (0.013)
∆x
0
t 0.0005 −0.004 −0.003 −0.007 −0.007 −0.007
(0.007) (0.007) (0.007) (0.007) (0.008) (0.008)
∆ef,t 0.155∗∗∗ 0.162∗∗∗ 0.163∗∗∗ 0.166∗∗∗ 0.166∗∗∗ 0.164∗∗∗
(0.017) (0.017) (0.017) (0.017) (0.017) (0.017)
∆x
0
t−1
0.001 −0.002 −0.002 −0.004 −0.004 −0.004
(0.007) (0.007) (0.007) (0.007) (0.007) (0.007)
∆ef,t−1 −0.005 −0.006 −0.004 −0.004 −0.005 −0.004
(0.020) (0.020) (0.020) (0.020) (0.020) (0.020)
∆pt−1 0.466∗∗∗ 0.477∗∗∗ 0.465∗∗∗ 0.454∗∗∗ 0.454∗∗∗ 0.450∗∗∗
(0.089) (0.086) (0.087) (0.087) (0.087) (0.088)
∆yt−1 0.033 0.040 0.044 0.041 0.040 0.040
(0.049) (0.047) (0.047) (0.047) (0.048) (0.048)
∆pt−2 0.172∗∗ 0.184∗∗ 0.196∗∗ 0.201∗∗ 0.201∗∗ 0.199∗∗
(0.079) (0.077) (0.078) (0.078) (0.078) (0.078)
∆ywt 0.808∗∗∗ 0.952∗∗∗ 0.854∗∗∗ 0.850∗∗∗ 0.815∗∗
(0.280) (0.306) (0.314) (0.317) (0.324)
∆reqwt −0.026 −0.029 −0.030 −0.040
(0.022) (0.023) (0.023) (0.029)
∆ewt −0.103 −0.105 −0.098
(0.079) (0.079) (0.081)
∆rwt 0.317 0.167
(2.114) (2.138)
grvt −0.032
(0.058)
Residual serial 11.521 8.003 5.321 5.888 5.868 5.924
correlation test [0.021] [0.091] [0.256] [0.208] [0.209] [0.205]
Adjusted R2 0.647 0.668 0.669 0.671 0.668 0.666
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
178

Table A.24: Quarterly estimates of the equation for output growth in the SVAR model of Iran
with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆yt
(1) (2) (3) (4) (5) (6)
st 0.024 0.028 0.027 0.026 0.026 0.027
(0.025) (0.025) (0.025) (0.025) (0.025) (0.026)
st−1 −0.050∗∗ −0.054∗∗ −0.053∗∗ −0.054∗∗ −0.054∗∗ −0.055∗∗
(0.025) (0.025) (0.025) (0.025) (0.025) (0.026)
∆x
0
t 0.025∗ 0.028∗∗ 0.028∗∗ 0.023 0.028∗ 0.028∗
(0.013) (0.014) (0.014) (0.015) (0.015) (0.015)
∆ef,t −0.128∗∗∗ −0.138∗∗∗ −0.140∗∗∗ −0.134∗∗∗ −0.134∗∗∗ −0.137∗∗∗
(0.042) (0.044) (0.044) (0.045) (0.044) (0.045)
∆pt 0.328∗ 0.364∗∗ 0.371∗∗ 0.353∗ 0.355∗∗ 0.348∗
(0.171) (0.177) (0.178) (0.179) (0.178) (0.179)
∆x
0
t−1
0.019 0.021 0.022 0.020 0.021 0.021
(0.013) (0.013) (0.013) (0.014) (0.014) (0.014)
∆ef,t−1 0.038 0.041 0.041 0.040 0.047 0.047
(0.034) (0.034) (0.035) (0.035) (0.035) (0.035)
∆pt−1 −0.449∗∗∗ −0.482∗∗∗ −0.484∗∗∗ −0.483∗∗∗ −0.500∗∗∗ −0.508∗∗∗
(0.156) (0.162) (0.162) (0.162) (0.162) (0.163)
∆yt−1 −0.219∗∗ −0.223∗∗ −0.225∗∗ −0.231∗∗ −0.226∗∗ −0.226∗∗
(0.088) (0.088) (0.089) (0.089) (0.088) (0.089)
∆ywt −0.459 −0.595 −0.715 −0.645 −0.710
(0.556) (0.609) (0.624) (0.622) (0.634)
∆reqwt 0.024 0.019 0.031 0.011
(0.043) (0.043) (0.044) (0.056)
∆ewt −0.139 −0.117 −0.103
(0.153) (0.153) (0.155)
∆rwt −5.988 −6.293
(4.064) (4.109)
grvt −0.066
(0.112)
Residual serial 7.054 6.738 6.977 7.483 7.876 8.205
correlation test [0.133] [0.150] [0.137] [0.112] [0.096] [0.084]
Adjusted R2 0.154 0.152 0.146 0.145 0.154 0.149
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
179

Table A.25: Quarterly estimates of the SVAR model of Iran with domestic variables ordered as:
exchange rate returns, oil exports, money supply growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆ef,t ∆x
0
t ∆mt ∆pt ∆yt
(1) (2) (3) (4) (5)
st 0.308∗∗∗ 0.058 −0.002 −0.033∗∗ 0.029
(0.064) (0.165) (0.017) (0.013) (0.026)
st−1 −0.241∗∗∗ −0.250 0.015 0.037∗∗∗ −0.056∗∗
(0.066) (0.164) (0.017) (0.013) (0.026)
∆ef,t 0.158 −0.007 0.163∗∗∗ −0.141∗∗∗
(0.223) (0.023) (0.017) (0.045)
∆x
0
t 0.006 −0.003 0.025∗
(0.010) (0.007) (0.014)
∆mt −0.073 0.063
(0.073) (0.142)
∆pt 0.387∗∗
(0.181)
∆ywt −2.399 8.786∗∗ 0.233 0.865∗∗∗ −0.520
(1.550) (3.696) (0.389) (0.298) (0.592)
∆ef,t−1 0.337∗∗∗ −0.495∗∗ −0.025 −0.009 0.041
(0.092) (0.230) (0.027) (0.020) (0.036)
∆x
0
t−1
0.042 −0.058 −0.005 −0.003 0.023∗
(0.038) (0.091) (0.009) (0.007) (0.014)
∆mt−1 0.129 −0.735 0.218∗∗ −0.025 0.046
(0.395) (0.933) (0.096) (0.075) (0.144)
∆pt−1 −0.339 0.106 0.167 0.488∗∗∗ −0.505∗∗∗
(0.337) (0.800) (0.115) (0.089) (0.167)
∆yt−1 −0.142 0.144 0.025 0.042 −0.221∗∗
(0.252) (0.594) (0.063) (0.048) (0.090)
∆pt−2 −0.070 0.183∗∗
(0.104) (0.079)
Residual serial 5.987 2.379 7.640 8.061 7.240
correlation test [0.200] [0.666] [0.106] [0.089] [0.124]
Adjusted R2 0.212 0.119 0.466 0.659 0.124
Notes: The variables are ordered as: ∆eft, ∆x
0
t
, ∆mt, ∆pt, and ∆yt, where: ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly rial/U.S. dollar free market exchange rate; ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the oil exports revenues in U.S.
dollars; ∆mt = (M2t − M2,t−1)/M2,t−1, M2t is the monetary aggregate M2 obtained by summing the aggregates
M1 and "quasi-money"; ∆pt = ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran; ∆yt = ln(Yt/Yt−1),
Yt is the quarterly real output of Iran. st is the quarterly sanctions intensity variable. Seasonal dummies are included to allow for possible seasonality of the variables in the regressions of the SVAR model in Equation (1.6) with
qt =

∆eft, ∆x
0
t
, ∆mt, ∆pt, ∆yt
0
and zwt = (∆ywt)
0
. Numbers in parentheses are least squares standard errors, and
those in square brackets are p-values. ***p < 0.01, **p < 0.05, *p < 0.1. "Residual serial correlation test" is the Breusch–
Godfrey LM test of serially uncorrelated errors with lag order of the test set to 4.
See the notes to Table A.9 for further details on the sources and construction of data used.
18

Table A.26: Quarterly estimates of the equation for the exchange rate returns in the SVAR model
of Iran with domestic variables ordered as: exchange rate returns, oil exports, money supply
growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆ef t
(1) (2) (3) (4) (5) (6)
st 0.297∗∗∗ 0.308∗∗∗ 0.305∗∗∗ 0.307∗∗∗ 0.307∗∗∗ 0.311∗∗∗
(0.064) (0.064) (0.064) (0.064) (0.065) (0.064)
st−1 −0.226∗∗∗ −0.241∗∗∗ −0.238∗∗∗ −0.239∗∗∗ −0.239∗∗∗ −0.243∗∗∗
(0.065) (0.066) (0.066) (0.066) (0.067) (0.066)
∆ef,t−1 0.338∗∗∗ 0.337∗∗∗ 0.335∗∗∗ 0.328∗∗∗ 0.327∗∗∗ 0.326∗∗∗
(0.093) (0.092) (0.093) (0.093) (0.094) (0.093)
∆x
0
t−1
0.038 0.042 0.044 0.047 0.047 0.043
(0.038) (0.038) (0.039) (0.039) (0.039) (0.038)
∆mt−1 −0.039 0.129 0.100 0.153 0.154 0.055
(0.382) (0.395) (0.402) (0.406) (0.408) (0.406)
∆pt−1 −0.280 −0.339 −0.332 −0.301 −0.297 −0.399
(0.337) (0.337) (0.339) (0.341) (0.344) (0.343)
∆yt−1 −0.132 −0.142 −0.148 −0.128 −0.129 −0.125
(0.253) (0.252) (0.253) (0.254) (0.255) (0.252)
∆ywt −2.399 −2.667 −2.228 −2.252 −2.759
(1.550) (1.676) (1.741) (1.760) (1.755)
∆reqwt 0.054 0.066 0.063 −0.129
(0.126) (0.127) (0.129) (0.160)
∆ewt 0.390 0.390 0.478
(0.415) (0.417) (0.414)
∆rwt 1.417 −1.456
(11.596) (11.531)
grvt −0.626∗∗
(0.314)
Residual serial 5.781 5.987 5.734 7.126 7.058 4.946
correlation test [0.216] [0.200] [0.220] [0.129] [0.133] [0.293]
Adjusted R2 0.202 0.212 0.206 0.205 0.198 0.220
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
181

Table A.27: Quarterly estimates of the equation for the oil export variable in the SVAR model
of Iran with domestic variables ordered as: exchange rate returns, oil exports, money supply
growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆x
0
t
(1) (2) (3) (4) (5) (6)
st 0.119 0.058 0.056 0.017 0.019 0.016
(0.166) (0.165) (0.166) (0.157) (0.153) (0.155)
st−1 −0.320∗ −0.250 −0.248 −0.223 −0.208 −0.205
(0.165) (0.164) (0.165) (0.155) (0.152) (0.153)
∆ef,t 0.082 0.158 0.157 0.230 0.224 0.232
(0.225) (0.223) (0.224) (0.212) (0.207) (0.212)
∆ef,t−1 −0.470∗∗ −0.495∗∗ −0.496∗∗ −0.458∗∗ −0.505∗∗ −0.507∗∗
(0.235) (0.230) (0.232) (0.218) (0.214) (0.215)
∆x
0
t−1 −0.038 −0.058 −0.057 −0.087 −0.098 −0.098
(0.092) (0.091) (0.092) (0.087) (0.085) (0.085)
∆mt−1 −0.125 −0.735 −0.760 −1.250 −1.175 −1.156
(0.915) (0.933) (0.950) (0.903) (0.882) (0.892)
∆pt−1 −0.133 0.106 0.112 −0.144 0.026 0.049
(0.810) (0.800) (0.804) (0.760) (0.745) (0.759)
∆yt−1 0.097 0.144 0.139 −0.032 −0.074 −0.074
(0.606) (0.594) (0.598) (0.564) (0.551) (0.554)
∆ywt 8.786∗∗ 8.551∗∗ 4.698 3.603 3.725
(3.696) (4.002) (3.896) (3.829) (3.902)
∆reqwt 0.047 −0.061 −0.168 −0.128
(0.298) (0.282) (0.279) (0.352)
∆ewt −3.597∗∗∗ −3.638∗∗∗ −3.659∗∗∗
(0.926) (0.904) (0.916)
∆rwt 63.168∗∗ 63.754∗∗
(25.040) (25.351)
grvt 0.130
(0.702)
Residual serial 1.196 2.379 2.414 1.926 4.488 4.558
correlation test [0.879] [0.666] [0.660] [0.749] [0.344] [0.336]
Adjusted R2 0.082 0.119 0.111 0.212 0.249 0.242
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
182

Table A.28: Quarterly estimates of the equation for money supply growth in the SVAR model
of Iran with domestic variables ordered as: exchange rate returns, oil exports, money supply
growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆mt
(1) (2) (3) (4) (5) (6)
st −0.0004 −0.002 −0.003 −0.002 −0.002 0.002
(0.017) (0.017) (0.017) (0.017) (0.017) (0.017)
st−1 0.014 0.015 0.016 0.017 0.017 0.014
(0.017) (0.017) (0.017) (0.017) (0.017) (0.017)
∆ef,t −0.009 −0.007 −0.008 −0.011 −0.012 −0.022
(0.023) (0.023) (0.023) (0.023) (0.023) (0.023)
∆x
0
t 0.008 0.006 0.006 0.011 0.014 0.015
(0.009) (0.010) (0.010) (0.010) (0.011) (0.010)
∆ef,t−1 −0.024 −0.025 −0.026 −0.027 −0.023 −0.021
(0.026) (0.027) (0.027) (0.027) (0.027) (0.026)
∆x
0
t−1 −0.005 −0.005 −0.005 −0.003 −0.002 −0.003
(0.009) (0.009) (0.009) (0.009) (0.009) (0.009)
∆mt−1 0.235∗∗ 0.218∗∗ 0.211∗∗ 0.236∗∗ 0.236∗∗ 0.211∗∗
(0.092) (0.096) (0.098) (0.099) (0.099) (0.098)
∆pt−1 0.165 0.167 0.174 0.193 0.189 0.163
(0.115) (0.115) (0.117) (0.117) (0.117) (0.115)
∆yt−1 0.022 0.025 0.022 0.026 0.028 0.026
(0.063) (0.063) (0.063) (0.063) (0.063) (0.062)
∆pt−2 −0.076 −0.070 −0.077 −0.088 −0.094 −0.102
(0.103) (0.104) (0.105) (0.105) (0.105) (0.103)
∆ywt 0.233 0.152 0.260 0.304 0.134
(0.389) (0.421) (0.427) (0.427) (0.425)
∆reqwt 0.016 0.021 0.027 −0.026
(0.031) (0.031) (0.031) (0.038)
∆ewt 0.146 0.160 0.191∗
(0.107) (0.108) (0.106)
∆rwt −3.581 −4.412
(2.850) (2.815)
grvt −0.176∗∗
(0.076)
Residual serial 7.428 7.640 7.255 6.129 5.742 4.178
correlation test [0.115] [0.106] [0.123] [0.190] [0.219] [0.382]
Adjusted R2 0.469 0.466 0.462 0.467 0.470 0.491
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
183

Table A.29: Quarterly estimates of the equation for inflation in the SVAR model of Iran with domestic variables ordered as: exchange rate returns, oil exports, money supply growth, inflation,
and output growth, estimated over the period 1989q1-2019q4
∆pt
(1) (2) (3) (4) (5) (6)
st −0.028∗∗ −0.033∗∗ −0.032∗∗ −0.033∗∗ −0.033∗∗ −0.032∗∗
(0.013) (0.013) (0.013) (0.013) (0.013) (0.013)
st−1 0.032∗∗ 0.037∗∗∗ 0.036∗∗∗ 0.036∗∗∗ 0.036∗∗∗ 0.035∗∗
(0.014) (0.013) (0.013) (0.013) (0.013) (0.013)
∆ef,t 0.155∗∗∗ 0.163∗∗∗ 0.164∗∗∗ 0.166∗∗∗ 0.166∗∗∗ 0.163∗∗∗
(0.018) (0.017) (0.018) (0.018) (0.018) (0.018)
∆x
0
t 0.001 −0.003 −0.003 −0.007 −0.007 −0.007
(0.008) (0.007) (0.007) (0.008) (0.008) (0.008)
∆mt −0.061 −0.073 −0.070 −0.058 −0.057 −0.070
(0.076) (0.073) (0.073) (0.074) (0.075) (0.077)
∆ef,t−1 −0.007 −0.009 −0.007 −0.007 −0.007 −0.007
(0.021) (0.020) (0.020) (0.020) (0.021) (0.021)
∆x
0
t−1 −0.001 −0.003 −0.003 −0.004 −0.004 −0.005
(0.007) (0.007) (0.007) (0.007) (0.007) (0.007)
∆mt−1 0.035 −0.025 −0.016 −0.036 −0.036 −0.040
(0.075) (0.075) (0.076) (0.078) (0.078) (0.078)
∆pt−1 0.480∗∗∗ 0.488∗∗∗ 0.478∗∗∗ 0.462∗∗∗ 0.462∗∗∗ 0.458∗∗∗
(0.092) (0.089) (0.090) (0.090) (0.091) (0.091)
∆yt−1 0.033 0.042 0.045 0.042 0.042 0.042
(0.050) (0.048) (0.048) (0.048) (0.049) (0.049)
∆pt−2 0.162∗∗ 0.183∗∗ 0.192∗∗ 0.201∗∗ 0.202∗∗ 0.198∗∗
(0.082) (0.079) (0.080) (0.080) (0.081) (0.081)
∆ywt 0.865∗∗∗ 0.971∗∗∗ 0.893∗∗∗ 0.889∗∗∗ 0.847∗∗
(0.298) (0.322) (0.327) (0.330) (0.335)
∆reqwt −0.021 −0.024 −0.025 −0.039
(0.023) (0.024) (0.024) (0.030)
∆ewt −0.102 −0.104 −0.093
(0.083) (0.084) (0.085)
∆rwt 0.303 0.031
(2.210) (2.242)
grvt −0.048
(0.061)
Residual serial 9.241 8.061 5.714 6.473 6.510 6.759
correlation test [0.055] [0.089] [0.222] [0.166] [0.164] [0.149]
Adjusted R2 0.635 0.659 0.658 0.660 0.656 0.655
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
184

Table A.30: Quarterly estimates of the equation for output growth in the SVAR model of Iran
with domestic variables ordered as: exchange rate returns, oil exports, money supply growth,
inflation, and output growth, estimated over the period 1989q1-2019q4
∆yt
(1) (2) (3) (4) (5) (6)
st 0.024 0.029 0.028 0.026 0.026 0.027
(0.025) (0.026) (0.026) (0.026) (0.026) (0.026)
st−1 −0.051∗∗ −0.056∗∗ −0.055∗∗ −0.055∗∗ −0.055∗∗ −0.056∗∗
(0.026) (0.026) (0.026) (0.026) (0.026) (0.026)
∆ef,t −0.130∗∗∗ −0.141∗∗∗ −0.142∗∗∗ −0.135∗∗∗ −0.136∗∗∗ −0.138∗∗∗
(0.043) (0.045) (0.045) (0.046) (0.046) (0.046)
∆x
0
t 0.023 0.025∗ 0.025∗ 0.020 0.025 0.026
(0.014) (0.014) (0.014) (0.015) (0.016) (0.016)
∆mt 0.052 0.063 0.062 0.078 0.054 0.037
(0.141) (0.142) (0.143) (0.143) (0.144) (0.148)
∆pt 0.348∗ 0.387∗∗ 0.390∗∗ 0.373∗∗ 0.373∗∗ 0.364∗∗
(0.175) (0.181) (0.182) (0.183) (0.182) (0.183)
∆ef,t−1 0.037 0.041 0.041 0.040 0.046 0.047
(0.036) (0.036) (0.036) (0.036) (0.036) (0.036)
∆x
0
t−1
0.022 0.023∗ 0.024∗ 0.022 0.023∗ 0.023
(0.014) (0.014) (0.014) (0.014) (0.014) (0.014)
∆mt−1 0.013 0.046 0.039 0.009 0.014 0.009
(0.139) (0.144) (0.147) (0.150) (0.149) (0.150)
∆pt−1 −0.466∗∗∗ −0.505∗∗∗ −0.505∗∗∗ −0.506∗∗∗ −0.519∗∗∗ −0.523∗∗∗
(0.161) (0.167) (0.168) (0.168) (0.167) (0.168)
∆yt−1 −0.218∗∗ −0.221∗∗ −0.223∗∗ −0.230∗∗ −0.225∗∗ −0.224∗∗
(0.090) (0.090) (0.090) (0.091) (0.090) (0.091)
∆ywt −0.520 −0.595 −0.708 −0.619 −0.664
(0.592) (0.637) (0.647) (0.647) (0.655)
∆reqwt 0.015 0.010 0.021 0.003
(0.045) (0.045) (0.046) (0.058)
∆ewt −0.160 −0.133 −0.121
(0.160) (0.161) (0.163)
∆rwt −6.160 −6.499
(4.239) (4.305)
grvt −0.060
(0.118)
Residual serial 7.242 7.240 7.371 7.721 8.049 8.248
correlation test [0.124] [0.124] [0.118] [0.102] [0.090] [0.083]
Adjusted R2 0.126 0.124 0.117 0.117 0.126 0.120
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
185

Table A.31: Quarterly estimates of the SVAR model of Iran with domestic variables ordered
as: exchange rate returns, oil exports, inflation, and output growth, estimated over the period
1989q1-2019q4
∆ef,t ∆x
0
t ∆pt ∆yt
(1) (2) (3) (4)
st 0.311∗∗∗ 0.079 −0.033∗∗ 0.028
(0.063) (0.166) (0.013) (0.025)
st−1 −0.241∗∗∗ −0.280∗ 0.036∗∗∗ −0.054∗∗
(0.064) (0.163) (0.013) (0.025)
∆ef,t 0.102 0.162∗∗∗ −0.138∗∗∗
(0.223) (0.017) (0.044)
∆x
0
t −0.004 0.028∗∗
(0.007) (0.014)
∆pt 0.364∗∗
(0.177)
∆ywt −2.316 7.910∗∗ 0.808∗∗∗ −0.459
(1.482) (3.605) (0.280) (0.556)
∆ef,t−1 0.326∗∗∗ −0.394∗ −0.006 0.041
(0.091) (0.231) (0.020) (0.034)
∆x
0
t−1
0.040 −0.067 −0.002 0.021
(0.038) (0.091) (0.007) (0.013)
∆pt−1 −0.339 −0.025 0.477∗∗∗ −0.482∗∗∗
(0.335) (0.810) (0.086) (0.162)
∆yt−1 −0.133 0.138 0.040 −0.223∗∗
(0.250) (0.602) (0.047) (0.088)
∆pt−2 0.184∗∗
(0.077)
Residual serial 4.832 3.895 8.003 6.738
correlation test [0.305] [0.420] [0.091] [0.150]
Adjusted R2 0.219 0.091 0.668 0.152
Notes: The variables are ordered as: ∆eft, ∆x
0
t
, ∆pt, and ∆yt, where: ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly
rial/U.S. dollar free market exchange rate; ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the oil exports revenues in U.S. dollars;
∆pt = ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran; ∆yt = ln(Yt/Yt−1), Yt is the quarterly real output of
Iran. st is the quarterly sanctions intensity variable. Seasonal dummies are not included to allow for possible seasonality
of the variables in the SVAR model. Numbers in parentheses are least squares standard errors, and those in square
brackets are p-values. ***p < 0.01, **p < 0.05, *p < 0.1. "Residual serial correlation test" is the Breusch–Godfrey LM
test of serially uncorrelated errors with lag order of the test set to 4.
See the notes to Table A.9 for further details on the sources and construction of data used.
186

Table A.32: Quarterly estimates of the equation for exchange rate returns in the SVAR model of
Iran with domestic variables ordered as: exchange rate returns, oil exports, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆ef t
(1) (2) (3) (4) (5) (6)
st 0.303∗∗∗ 0.311∗∗∗ 0.307∗∗∗ 0.307∗∗∗ 0.307∗∗∗ 0.314∗∗∗
(0.063) (0.063) (0.063) (0.063) (0.064) (0.063)
st−1 −0.231∗∗∗ −0.241∗∗∗ −0.238∗∗∗ −0.236∗∗∗ −0.235∗∗∗ −0.243∗∗∗
(0.064) (0.064) (0.064) (0.065) (0.065) (0.064)
∆ef,t−1 0.329∗∗∗ 0.326∗∗∗ 0.323∗∗∗ 0.317∗∗∗ 0.314∗∗∗ 0.312∗∗∗
(0.091) (0.091) (0.091) (0.092) (0.092) (0.091)
∆x
0
t−1
0.035 0.040 0.043 0.047 0.046 0.041
(0.038) (0.038) (0.038) (0.038) (0.038) (0.038)
∆pt−1 −0.284 −0.339 −0.330 −0.298 −0.288 −0.388
(0.335) (0.335) (0.336) (0.338) (0.341) (0.340)
∆yt−1 −0.121 −0.133 −0.140 −0.120 −0.122 −0.118
(0.251) (0.250) (0.251) (0.252) (0.253) (0.250)
∆ywt −2.316 −2.724∗ −2.217 −2.265 −2.837
(1.482) (1.625) (1.706) (1.720) (1.721)
∆reqwt 0.076 0.088 0.082 −0.110
(0.122) (0.122) (0.125) (0.156)
∆ewt 0.397 0.395 0.507
(0.405) (0.407) (0.405)
∆rwt 3.363 0.477
(11.277) (11.221)
grvt −0.619∗∗
(0.308)
Residual serial 5.198 4.832 4.686 5.847 5.717 4.179
correlation test [0.268] [0.305] [0.321] [0.211] [0.221] [0.382]
Adjusted R2 0.209 0.219 0.215 0.215 0.208 0.229
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
187

Table A.33: Quarterly estimates of the equation for the oil exports variable in the SVAR model of
Iran with domestic variables ordered as: exchange rate returns, oil exports, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆x
0
t
(1) (2) (3) (4) (5) (6)
st 0.126 0.079 0.078 0.053 0.055 0.051
(0.168) (0.166) (0.167) (0.158) (0.154) (0.156)
st−1 −0.330∗∗ −0.280∗ −0.280∗ −0.274∗ −0.257∗ −0.253
(0.165) (0.163) (0.164) (0.155) (0.152) (0.153)
∆ef,t 0.032 0.102 0.101 0.175 0.160 0.168
(0.225) (0.223) (0.225) (0.213) (0.208) (0.213)
∆ef,t−1 −0.381 −0.394∗ −0.395∗ −0.360 −0.414∗ −0.416∗
(0.234) (0.231) (0.232) (0.219) (0.214) (0.216)
∆x
0
t−1 −0.044 −0.067 −0.066 −0.102 −0.111 −0.110
(0.092) (0.091) (0.092) (0.087) (0.085) (0.086)
∆pt−1 −0.234 −0.025 −0.021 −0.275 −0.077 −0.051
(0.817) (0.810) (0.814) (0.772) (0.756) (0.769)
∆yt−1 0.087 0.138 0.135 −0.035 −0.088 −0.088
(0.612) (0.602) (0.606) (0.574) (0.560) (0.562)
∆ywt 7.910∗∗ 7.738∗ 3.419 2.430 2.585
(3.605) (3.965) (3.915) (3.832) (3.919)
∆reqwt 0.031 −0.085 −0.209 −0.164
(0.294) (0.280) (0.276) (0.351)
∆ewt −3.538∗∗∗ −3.577∗∗∗ −3.607∗∗∗
(0.927) (0.903) (0.918)
∆rwt 66.363∗∗∗ 67.024∗∗∗
(24.942) (25.247)
grvt 0.148
(0.706)
Residual serial 2.461 3.895 4.005 2.396 4.791 5.085
correlation test [0.652] [0.420] [0.405] [0.663] [0.309] [0.279]
Adjusted R2 0.061 0.091 0.083 0.180 0.222 0.215
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
188

Table A.34: Quarterly estimates of the equation for inflation in the SVAR model of Iran with
domestic variables ordered as: exchange rate returns, oil exports, inflation, and output growth,
estimated over the period 1989q1-2019q4
∆pt
(1) (2) (3) (4) (5) (6)
st −0.029∗∗ −0.033∗∗ −0.032∗∗ −0.032∗∗ −0.032∗∗ −0.031∗∗
(0.013) (0.013) (0.013) (0.013) (0.013) (0.013)
st−1 0.032∗∗ 0.036∗∗∗ 0.035∗∗∗ 0.034∗∗ 0.034∗∗ 0.033∗∗
(0.013) (0.013) (0.013) (0.013) (0.013) (0.013)
∆ef,t 0.155∗∗∗ 0.162∗∗∗ 0.163∗∗∗ 0.166∗∗∗ 0.166∗∗∗ 0.164∗∗∗
(0.017) (0.017) (0.017) (0.017) (0.017) (0.017)
∆x
0
t 0.0005 −0.004 −0.003 −0.007 −0.007 −0.007
(0.007) (0.007) (0.007) (0.007) (0.008) (0.008)
∆ef,t−1 −0.005 −0.006 −0.004 −0.004 −0.005 −0.004
(0.020) (0.020) (0.020) (0.020) (0.020) (0.020)
∆x
0
t−1
0.001 −0.002 −0.002 −0.004 −0.004 −0.004
(0.007) (0.007) (0.007) (0.007) (0.007) (0.007)
∆pt−1 0.466∗∗∗ 0.477∗∗∗ 0.465∗∗∗ 0.454∗∗∗ 0.454∗∗∗ 0.450∗∗∗
(0.089) (0.086) (0.087) (0.087) (0.087) (0.088)
∆yt−1 0.033 0.040 0.044 0.041 0.040 0.040
(0.049) (0.047) (0.047) (0.047) (0.048) (0.048)
∆pt−2 0.172∗∗ 0.184∗∗ 0.196∗∗ 0.201∗∗ 0.201∗∗ 0.199∗∗
(0.079) (0.077) (0.078) (0.078) (0.078) (0.078)
∆ywt 0.808∗∗∗ 0.952∗∗∗ 0.854∗∗∗ 0.850∗∗∗ 0.815∗∗
(0.280) (0.306) (0.314) (0.317) (0.324)
∆reqwt −0.026 −0.029 −0.030 −0.040
(0.022) (0.023) (0.023) (0.029)
∆ewt −0.103 −0.105 −0.098
(0.079) (0.079) (0.081)
∆rwt 0.317 0.167
(2.114) (2.138)
grvt −0.032
(0.058)
Residual serial 11.521 8.003 5.321 5.888 5.868 5.924
correlation test [0.021] [0.091] [0.256] [0.208] [0.209] [0.205]
Adjusted R2 0.647 0.668 0.669 0.671 0.668 0.666
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
189

Table A.35: Quarterly estimates of the equation for output growth in the SVAR model of Iran
with domestic variables ordered as: exchange rate returns, oil exports, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆yt
(1) (2) (3) (4) (5) (6)
st 0.024 0.028 0.027 0.026 0.026 0.027
(0.025) (0.025) (0.025) (0.025) (0.025) (0.026)
st−1 −0.050∗∗ −0.054∗∗ −0.053∗∗ −0.054∗∗ −0.054∗∗ −0.055∗∗
(0.025) (0.025) (0.025) (0.025) (0.025) (0.026)
∆ef,t −0.128∗∗∗ −0.138∗∗∗ −0.140∗∗∗ −0.134∗∗∗ −0.134∗∗∗ −0.137∗∗∗
(0.042) (0.044) (0.044) (0.045) (0.044) (0.045)
∆x
0
t 0.025∗ 0.028∗∗ 0.028∗∗ 0.023 0.028∗ 0.028∗
(0.013) (0.014) (0.014) (0.015) (0.015) (0.015)
∆pt 0.328∗ 0.364∗∗ 0.371∗∗ 0.353∗ 0.355∗∗ 0.348∗
(0.171) (0.177) (0.178) (0.179) (0.178) (0.179)
∆ef,t−1 0.038 0.041 0.041 0.040 0.047 0.047
(0.034) (0.034) (0.035) (0.035) (0.035) (0.035)
∆x
0
t−1
0.019 0.021 0.022 0.020 0.021 0.021
(0.013) (0.013) (0.013) (0.014) (0.014) (0.014)
∆pt−1 −0.449∗∗∗ −0.482∗∗∗ −0.484∗∗∗ −0.483∗∗∗ −0.500∗∗∗ −0.508∗∗∗
(0.156) (0.162) (0.162) (0.162) (0.162) (0.163)
∆yt−1 −0.219∗∗ −0.223∗∗ −0.225∗∗ −0.231∗∗ −0.226∗∗ −0.226∗∗
(0.088) (0.088) (0.089) (0.089) (0.088) (0.089)
∆ywt −0.459 −0.595 −0.715 −0.645 −0.710
(0.556) (0.609) (0.624) (0.622) (0.634)
∆reqwt 0.024 0.019 0.031 0.011
(0.043) (0.043) (0.044) (0.056)
∆ewt −0.139 −0.117 −0.103
(0.153) (0.153) (0.155)
∆rwt −5.988 −6.293
(4.064) (4.109)
grvt −0.066
(0.112)
Residual serial 7.054 6.738 6.977 7.483 7.876 8.205
correlation test [0.133] [0.150] [0.137] [0.112] [0.096] [0.084]
Adjusted R2 0.154 0.152 0.146 0.145 0.154 0.149
Notes: See the notes to Table A.9 for further details on the sources and construction of data used.
190

Table A.36: Estimates of the reduced form Iran’s output growth equation using a sanctions
dummy variable estimated over the period 1989q1- 2019q4
∆yt
(1) (2) (3) (4) (5) (6) (7)
dt−1(βdt−1
) −0.014∗ −0.014∗ −0.014∗ −0.014∗ −0.013∗ −0.013∗ −0.013∗
(0.007) (0.008) (0.008) (0.008) (0.008) (0.008) (0.008)
∆yt−1(λ∆yt−1
) −0.207∗∗ −0.206∗∗ −0.208∗∗ −0.205∗∗ −0.214∗∗ −0.215∗∗ −0.218∗∗
(0.092) (0.092) (0.093) (0.093) (0.092) (0.093) (0.093)
∆x
0
t−1
0.017 0.016 0.017 0.018 0.016 0.015 0.016
(0.014) (0.014) (0.014) (0.014) (0.014) (0.014) (0.014)
∆ef,t−1 −0.013 −0.013 −0.013 −0.010 −0.007 −0.008 −0.009
(0.033) (0.033) (0.033) (0.033) (0.033) (0.033) (0.033)
∆mt−1 −0.042 −0.048 −0.053 −0.046 −0.066 −0.071 −0.078
(0.100) (0.102) (0.103) (0.104) (0.104) (0.105) (0.106)
∆pt−1 −0.214∗ −0.212∗ −0.209∗ −0.222∗ −0.245∗ −0.252∗∗ −0.253∗
(0.124) (0.125) (0.125) (0.126) (0.126) (0.127) (0.128)
∆ywt 0.156 0.058 0.107 −0.169 −0.230 −0.186
(0.561) (0.611) (0.613) (0.633) (0.645) (0.653)
∆reqwt 0.019 0.026 0.019 −0.0002 −0.002
(0.045) (0.046) (0.046) (0.058) (0.058)
∆rwt −4.060 −3.838 −4.120 −3.154
(4.145) (4.119) (4.165) (4.638)
∆ewt −0.240 −0.228 −0.265
(0.150) (0.152) (0.171)
grvt −0.063 −0.069
(0.116) (0.117)
∆p
0
t −0.012
(0.024)
βdt−1
/(1 − λ∆yt−1
) −0.012∗ −0.011∗ −0.011∗ −0.012∗ −0.010∗ −0.011∗ −0.011∗
(0.006) (0.006) (0.006) (0.006) (0.006) (0.006) (0.006)
Adjusted R2 0.078 0.070 0.064 0.063 0.076 0.070 0.064
Notes: ∆yt = ln(Yt/Yt−1), Yt is the quarterly real output of Iran. dt is the sanctions dummy variable. βdt−1
and λ∆yt−1
are the coefficients of dt−1 and ∆yt−1, respectively; βdt−1
/(1− λ∆yt−1
) represents the long run effect of sanctions on
output growth. See Chapter 6 of Pesaran (2015).
See the notes to Table A.9 for further details on the sources and construction of data used. Details on the construction
of dt are provided in Section A.2.4.
191

Table A.37: Reduced form Iran’s output growth equation using a discretized sanctions intensity
variable estimated over the period 1989q1–2019q4
∆yt
(1) (2) (3) (4) (5) (6) (7)
s
D
t−1
(βsD
t−1
) −0.010∗ −0.010∗ −0.010∗ −0.010∗ −0.009 −0.009 −0.009
(0.006) (0.006) (0.006) (0.006) (0.006) (0.006) (0.006)
∆yt−1(λ∆yt−1
) −0.202∗∗ −0.201∗∗ −0.202∗∗ −0.199∗∗ −0.209∗∗ −0.209∗∗ −0.214∗∗
(0.092) (0.092) (0.093) (0.093) (0.092) (0.093) (0.093)
∆x
0
t−1
0.015 0.015 0.016 0.016 0.014 0.014 0.015
(0.014) (0.014) (0.014) (0.014) (0.014) (0.014) (0.014)
∆ef,t−1 −0.016 −0.015 −0.016 −0.013 −0.010 −0.010 −0.012
(0.033) (0.033) (0.033) (0.033) (0.033) (0.033) (0.033)
∆mt−1 −0.032 −0.039 −0.044 −0.037 −0.058 −0.061 −0.069
(0.100) (0.103) (0.104) (0.105) (0.105) (0.106) (0.107)
∆pt−1 −0.226∗ −0.222∗ −0.220∗ −0.233∗ −0.255∗∗ −0.261∗∗ −0.261∗∗
(0.123) (0.124) (0.125) (0.126) (0.126) (0.127) (0.128)
∆ywt 0.182 0.095 0.144 −0.138 −0.169 −0.119
(0.560) (0.610) (0.613) (0.633) (0.643) (0.651)
∆reqwt 0.017 0.024 0.017 0.006 0.004
(0.045) (0.046) (0.046) (0.058) (0.058)
∆rwt −3.935 −3.722 −3.875 −2.730
(4.153) (4.126) (4.172) (4.645)
∆ewt −0.243 −0.237 −0.279
(0.151) (0.153) (0.171)
grvt −0.036 −0.044
(0.115) (0.116)
∆p
0
t −0.014
(0.025)
βsD
t−1
/(1 − λ∆yt−1
) −0.008∗ −0.008∗ −0.008∗ −0.008∗ −0.007 −0.007 −0.008
(0.005) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005)
Adjusted R2 0.074 0.067 0.060 0.059 0.072 0.064 0.059
Notes: ∆yt = ln(Yt/Yt−1), Yt is the quarterly real output of Iran. s
D
t
is the discretized sanctions intensity variable.
βsD
t−1
and λ∆yt−1
are the coefficients of s
D
t−1 and ∆yt−1, respectively; βsD
t−1
/(1− λ∆yt−1
) represents the long run effect
of sanctions on output growth. See Chapter 6 of Pesaran (2015).
See the notes to Table A.9 for further details on the sources and construction of data used. Details on the construction
of s
D
t are provided in Section A.2.4.
192

Table A.38: Quarterly estimates of the SVAR model of Iran using a sanctions dummy variable and
with domestic variables ordered as: oil exports, exchange rate returns, money supply growth,
inflation, and output growth, estimated over the period 1989q1-2019q4
∆x
0
t ∆ef,t ∆mt ∆pt ∆yt
(1) (2) (3) (4) (5)
dt 0.290∗∗ 0.079 0.004 0.001 −0.014
(0.116) (0.056) (0.013) (0.010) (0.019)
dt−1 −0.357∗∗∗ −0.040 −0.002 0.00002 0.003
(0.117) (0.057) (0.013) (0.010) (0.019)
∆x
0
t 0.037 0.004 −0.005 0.030∗∗
(0.044) (0.010) (0.008) (0.015)
∆ef,t −0.010 0.144∗∗∗ −0.104∗∗
(0.021) (0.017) (0.040)
∆mt −0.061 0.033
(0.076) (0.143)
∆pt 0.298∗
(0.176)
∆ywt 7.813∗∗ −1.600 0.229 0.758∗∗ −0.390
(3.592) (1.718) (0.390) (0.308) (0.590)
∆x
0
t−1 −0.043 0.043 −0.006 −0.002 0.022
(0.088) (0.041) (0.009) (0.007) (0.014)
∆ef,t−1 −0.556∗∗∗ 0.298∗∗∗ −0.019 0.007 0.023
(0.209) (0.101) (0.026) (0.020) (0.035)
∆mt−1 −0.542 −0.019 0.232∗∗ −0.002 0.006
(0.910) (0.427) (0.097) (0.079) (0.147)
∆pt−1 0.248 −0.302 0.159 0.441∗∗∗ −0.418∗∗
(0.786) (0.368) (0.114) (0.091) (0.164)
∆yt−1 0.193 −0.144 0.024 0.049 −0.227∗∗
(0.583) (0.273) (0.063) (0.050) (0.092)
∆pt−2 −0.060 0.221∗∗∗
(0.104) (0.082)
Residual serial 1.889 5.847 7.734 5.096 6.274
correlation test [0.756] [0.211] [0.102] [0.278] [0.180]
Adjusted R2 0.160 0.083 0.460 0.633 0.101
Notes: The variables are ordered as: ∆x
0
t
, ∆eft, ∆mt, ∆pt, and ∆yt, where: ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the oil
exports revenues in U.S. dollars; ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly rial/U.S. dollar free market exchange rate;
∆mt = (M2t−M2,t−1)/M2,t−1, M2t is the monetary aggregate M2 obtained by summing the aggregates M1 and "quasimoney"; ∆pt = ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran; ∆yt = ln(Yt/Yt−1), Yt is the quarterly real
output of Iran. dt is the sanctions dummy variable. Seasonal dummies are included to allow for possible seasonality
of the variables in the regressions of the SVAR model in Equation (1.6) with qt =

∆x
0
t
, ∆eft, ∆mt, ∆pt, ∆yt
0
and
zwt = (∆ywt)
0
. Numbers in parentheses are least squares standard errors, and those in square brackets are p-values.
***p < 0.01, **p < 0.05, *p < 0.1. "Residual serial correlation test" is the Breusch–Godfrey LM test of serially uncorrelated
errors with lag order of the test set to 4.
See the notes to Table A.9 for further details on the sources and construction of data used. Details on the construction
of dt are provided in Section A.2.4.
19

Table A.39: Quarterly estimates of the SVAR model of Iran using a discretized sanctions intensity variable and with domestic variables ordered as: oil exports, exchange rate returns, money
supply growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆x
0
t ∆ef,t ∆mt ∆pt ∆yt
(1) (2) (3) (4) (5)
s
D
t 0.124 0.073∗ 0.006 −0.011 0.013
(0.090) (0.042) (0.010) (0.007) (0.014)
s
D
t−1 −0.200∗∗ −0.060 −0.002 0.014∗ −0.021
(0.089) (0.043) (0.010) (0.007) (0.014)
∆x
0
t 0.034 0.005 −0.003 0.025∗
(0.044) (0.010) (0.008) (0.015)
∆ef,t −0.011 0.148∗∗∗ −0.121∗∗∗
(0.021) (0.017) (0.041)
∆mt −0.059 0.034
(0.075) (0.143)
∆pt 0.338∗
(0.178)
∆ywt 7.778∗∗ −1.903 0.238 0.793∗∗ −0.443
(3.623) (1.724) (0.389) (0.304) (0.590)
∆x
0
t−1 −0.075 0.035 −0.005 −0.0003 0.021
(0.090) (0.042) (0.009) (0.007) (0.014)
∆ef,t−1 −0.536∗∗ 0.311∗∗∗ −0.019 0.008 0.019
(0.210) (0.101) (0.025) (0.020) (0.035)
∆mt−1 −0.565 −0.015 0.225∗∗ −0.020 0.045
(0.921) (0.431) (0.097) (0.077) (0.146)
∆pt−1 0.109 −0.302 0.151 0.453∗∗∗ −0.461∗∗∗
(0.795) (0.371) (0.114) (0.090) (0.166)
∆yt−1 0.120 −0.187 0.027 0.046 −0.216∗∗
(0.586) (0.273) (0.063) (0.049) (0.091)
∆pt−2 −0.062 0.213∗∗∗
(0.103) (0.080)
Residual serial 2.613 2.834 7.270 6.829 5.194
correlation test [0.625] [0.586] [0.122] [0.145] [0.268]
Adjusted R2 0.143 0.072 0.463 0.644 0.107
Notes: The variables are ordered as: ∆x
0
t
, ∆eft, ∆mt, ∆pt, and ∆yt, where: ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the
oil exports revenues in U.S. dollars; ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly rial/U.S. dollar free market exchange
rate; ∆mt = (M2t − M2,t−1)/M2,t−1, M2t is the monetary aggregate M2 obtained by summing the aggregates M1
and "quasi-money"; ∆pt = ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran; ∆yt = ln(Yt/Yt−1), Yt is
the quarterly real output of Iran. s
D
t
is the discretized sanctions intensity variable. Seasonal dummies are included
to allow for possible seasonality of the variables in the regressions of the SVAR model in Equation (1.6) with qt =
∆x
0
t
, ∆eft, ∆mt, ∆pt, ∆yt
0
and zwt = (∆ywt)
0
. Numbers in parentheses are least squares standard errors, and those
in square brackets are p-values. ***p < 0.01, **p < 0.05, *p < 0.1. "Residual serial correlation test" is the Breusch–
Godfrey LM test of serially uncorrelated errors with lag order of the test set to 4.
See the notes to Table A.9 for further details on the sources and construction of data used. Details on the construction
of s
D
t are provided in Section A.2.4.
19

A.4.7 Additional IRFs and FEVDs results
Figure A.5: Impulse responses of the effects of a world output shock on oil exports, foreign
exchange, inflation, and output growth
One positive standard error shock to world output growth
195

Figure A.6: Impulse responses of the effects of sanctions and domestic shocks on foreign exchange, oil exports, inflation, and output growth
Panel A: One positive standard error shock to the sanctions intensity variable
Panel B: One positive standard error shock to the foreign exchange rate
Panel C: One positive standard error shock to the oil exports revenues
196

Figure A.7: Impulse responses of the effects of sanctions and domestic shocks on foreign exchange, oil exports, inflation, and output growth
Panel D: One positive standard error shock to inflation
Panel E: One positive standard error shock to Iran output growth
Panel F: One positive standard error shock to world output growth
197

Table A.40: Forecast error variance decomposition in the SVAR model with domestic variables
ordered as exchange rate returns, oil exports, inflation, and output growth
Panel A: FEVD for exchange rate Panel B: FEVD for oil exports
Quarter Proportion explained by a shock to: Quarter Proportion explained by a shock to:
ahead st ∆ef t ∆x
0
t ∆pt ∆yt ∆ywt ahead st ∆ef t ∆x
0
t ∆pt ∆yt ∆ywt
0 0.17 0.82 0.00 0.00 0.00 0.01 0 0.00 0.00 0.96 0.00 0.00 0.03
1 0.17 0.80 0.01 0.00 0.00 0.02 1 0.04 0.02 0.90 0.00 0.00 0.04
2 0.17 0.80 0.01 0.00 0.00 0.02 2 0.05 0.03 0.88 0.00 0.00 0.04
3 0.17 0.80 0.01 0.01 0.00 0.02 3 0.06 0.03 0.88 0.00 0.00 0.04
4 0.17 0.80 0.01 0.01 0.00 0.02 4 0.06 0.03 0.88 0.00 0.00 0.04
5 0.17 0.80 0.01 0.01 0.00 0.02 5 0.06 0.03 0.88 0.00 0.00 0.04
6 0.17 0.80 0.01 0.01 0.00 0.02 6 0.06 0.03 0.88 0.00 0.00 0.04
7 0.17 0.80 0.01 0.01 0.00 0.02 7 0.06 0.03 0.87 0.00 0.00 0.04
8 0.17 0.80 0.01 0.01 0.00 0.02 8 0.06 0.03 0.87 0.00 0.00 0.04
Panel C: FEVD for inflation Panel D: FEVD for output growth
Quarter Proportion explained by a shock to: Quarter Proportion explained by a shock to:
ahead st ∆ef t ∆x
0
t ∆pt ∆yt ∆ywt ahead st ∆ef t ∆x
0
t ∆pt ∆yt ∆ywt
0 0.01 0.43 0.00 0.55 0.00 0.01 0 0.00 0.04 0.03 0.03 0.90 0.00
1 0.04 0.47 0.00 0.48 0.00 0.01 1 0.02 0.04 0.03 0.06 0.85 0.00
2 0.05 0.49 0.00 0.45 0.00 0.01 2 0.03 0.05 0.03 0.06 0.83 0.00
3 0.06 0.50 0.00 0.44 0.00 0.01 3 0.04 0.05 0.03 0.06 0.83 0.00
4 0.06 0.50 0.00 0.43 0.00 0.01 4 0.04 0.05 0.03 0.06 0.82 0.00
5 0.06 0.50 0.00 0.43 0.00 0.01 5 0.04 0.05 0.03 0.06 0.82 0.00
6 0.07 0.50 0.00 0.43 0.00 0.01 6 0.05 0.05 0.03 0.06 0.82 0.00
7 0.07 0.50 0.00 0.42 0.00 0.01 7 0.05 0.05 0.03 0.06 0.82 0.00
8 0.07 0.50 0.00 0.42 0.00 0.01 8 0.05 0.05 0.03 0.06 0.82 0.00
Notes: st is the quarterly sanctions intensity variable. ∆eft = ln(Eft/Ef,t−1), Eft is the Iran rial/U.S. dollar quarterly
free market exchange rate. ∆x
0
t = (X
0
t −X
0
t−1)/X0
t−1, X0
t
is the oil exports revenues in U.S. dollars. ∆pt = ln(Pt/Pt−1),
Pt is the quarterly consumer price index of Iran. ∆yt = ln(Yt/Yt−1), Yt is the quarterly real output of Iran. ∆ywt is
the quarterly world output growth: ywt =
Pn
i=1 wiyit, with {yit}
n
i=1 being the natural log of real output for 33 major
economies, and wi the GDP-PPP weights.
See Sections A.2.1, A.2.2, A.2.5, and A.2.6 in the data appendix for details on the construction of the sanctions intensity
variable, calendar conversions, and sources of the data used.
198

Figure A.8: Forecast error variance decomposition for domestic variables in the SVAR model
with a cumulative shock to sanctions, and domestic variables ordered as FX returns, oil exports,
inflation, and output growth
199

A.4.8 Additional analyses using heteroskedastic-consistent standard
errors
Table A.41: Estimates of the reduced form Iran’s output growth equation estimated over the
period 1989q1–2019q4
∆yt
(1) (2) (3) (4) (5) (6) (7)
st−1(βst−1
) −0.033∗ −0.032∗ −0.032∗ −0.034∗ −0.034∗∗ −0.034∗ −0.035∗∗
(0.018) (0.018) (0.018) (0.018) (0.017) (0.018) (0.018)
∆yt−1(λ∆yt−1
) −0.204∗ −0.202∗ −0.203∗ −0.200∗ −0.214∗ −0.214∗ −0.218∗
(0.115) (0.116) (0.117) (0.119) (0.111) (0.112) (0.113)
∆x
0
t−1
0.016 0.016 0.016 0.017 0.014 0.014 0.015
(0.011) (0.011) (0.012) (0.012) (0.012) (0.012) (0.013)
∆ef,t−1 −0.004 −0.004 −0.004 0.0002 0.004 0.004 0.002
(0.039) (0.039) (0.039) (0.038) (0.038) (0.038) (0.038)
∆mt−1 −0.028 −0.037 −0.041 −0.032 −0.053 −0.056 −0.063
(0.098) (0.102) (0.103) (0.107) (0.103) (0.105) (0.111)
∆pt−1 −0.239∗ −0.234 −0.232 −0.246∗ −0.268∗ −0.273∗ −0.274∗
(0.145) (0.148) (0.148) (0.143) (0.147) (0.146) (0.146)
∆ywt 0.228 0.160 0.215 −0.129 −0.162 −0.117
(0.381) (0.426) (0.430) (0.476) (0.475) (0.493)
∆reqwt 0.013 0.021 0.013 0.002 −0.0001
(0.040) (0.042) (0.045) (0.064) (0.064)
∆rwt −4.518 −4.311 −4.474 −3.490
(4.471) (4.393) (4.525) (4.921)
∆ewt −0.278∗ −0.272∗ −0.309∗
(0.160) (0.160) (0.177)
grvt −0.038 −0.044
(0.116) (0.118)
∆p
0
t −0.012
(0.025)
βst−1
/(1 − λ∆yt−1
) −0.027∗ −0.027∗ −0.027∗ −0.028∗ −0.028∗∗ −0.028∗∗ −0.028∗∗
(0.015) (0.015) (0.015) (0.015) (0.014) (0.014) (0.014)
Adjusted R2 0.083 0.077 0.069 0.071 0.091 0.084 0.077
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
200

Table A.42: Estimates of the reduced form of the output growth equation for Iran including
contemporaneous sanctions variable and estimated over the period 1989q1–2019q4
∆yt
(1) (2) (3) (4) (5) (6) (7)
st(βst
) −0.007 −0.008 −0.009 −0.009 −0.009 −0.009 −0.008
(0.024) (0.024) (0.024) (0.024) (0.023) (0.024) (0.023)
st−1(βst−1
) −0.027 −0.026 −0.025 −0.027 −0.027 −0.028 −0.029
(0.027) (0.027) (0.027) (0.027) (0.025) (0.026) (0.026)
∆yt−1(λ∆yt−1
) −0.206∗ −0.204∗ −0.206∗ −0.202∗ −0.217∗ −0.216∗ −0.220∗
(0.117) (0.118) (0.119) (0.121) (0.112) (0.113) (0.114)
∆x
0
t−1
0.016 0.015 0.016 0.017 0.014 0.014 0.015
(0.012) (0.011) (0.012) (0.012) (0.012) (0.012) (0.013)
∆ef,t−1 −0.005 −0.004 −0.005 −0.001 0.003 0.003 0.002
(0.039) (0.039) (0.039) (0.038) (0.038) (0.038) (0.038)
∆mt−1 −0.030 −0.039 −0.044 −0.035 −0.056 −0.059 −0.065
(0.098) (0.103) (0.104) (0.108) (0.104) (0.106) (0.112)
∆pt−1 −0.236 −0.230 −0.228 −0.242∗ −0.263∗ −0.269∗ −0.270∗
(0.145) (0.148) (0.148) (0.143) (0.147) (0.146) (0.146)
∆ywt 0.245 0.170 0.224 −0.120 −0.151 −0.111
(0.391) (0.429) (0.434) (0.480) (0.480) (0.497)
∆reqwt 0.015 0.023 0.015 0.004 0.002
(0.041) (0.043) (0.046) (0.064) (0.065)
∆rwt −4.500 −4.291 −4.447 −3.537
(4.476) (4.394) (4.529) (4.946)
∆ewt −0.279∗ −0.273∗ −0.307∗
(0.161) (0.161) (0.179)
grvt −0.036 −0.042
(0.116) (0.118)
∆p
0
t −0.011
(0.025)
βst + βst−1 −0.035∗ −0.034∗ −0.034∗ −0.036∗ −0.037∗∗ −0.037∗ −0.036∗
(0.019) (0.019) (0.019) (0.019) (0.018) (0.019) (0.018)
(βst + βst−1
)/(1 − λ∆yt−1
) −0.029∗ −0.028∗ −0.028∗ −0.030∗∗ −0.030∗∗ −0.030∗∗ −0.030∗∗
(0.015) (0.015) (0.015) (0.015) (0.015) (0.015) (0.015)
Adjusted R2 0.076 0.069 0.062 0.064 0.084 0.077 0.070
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
201

Table A.43: Quarterly estimates of the SVAR model of Iran with domestic variables ordered as:
oil exports, exchange rate returns, money supply growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆x
0
t ∆ef,t ∆mt ∆pt ∆yt
(1) (2) (3) (4) (5)
st 0.107 0.305∗∗∗ −0.002 −0.033∗∗∗ 0.029
(0.106) (0.080) (0.015) (0.013) (0.024)
st−1 −0.288∗∗ −0.233∗∗∗ 0.015 0.037∗∗∗ −0.056∗∗
(0.136) (0.065) (0.018) (0.012) (0.028)
∆x
0
t 0.029 0.006 −0.003 0.025
(0.041) (0.007) (0.008) (0.016)
∆ef,t −0.007 0.163∗∗∗ −0.141∗∗∗
(0.018) (0.028) (0.042)
∆mt −0.073 0.063
(0.072) (0.118)
∆pt 0.387∗∗
(0.173)
∆ywt 8.406∗∗ −2.639 0.233 0.865∗∗∗ −0.520
(3.630) (1.641) (0.406) (0.285) (0.458)
∆x
0
t−1 −0.051 0.044 −0.005 −0.003 0.023∗
(0.107) (0.036) (0.007) (0.008) (0.012)
∆ef,t−1 −0.441∗ 0.350∗∗ −0.025 −0.009 0.041
(0.238) (0.175) (0.022) (0.027) (0.044)
∆mt−1 −0.715 0.149 0.218 −0.025 0.046
(0.765) (0.266) (0.146) (0.072) (0.129)
∆pt−1 0.052 −0.341 0.167∗ 0.488∗∗∗ −0.505∗∗∗
(0.707) (0.610) (0.088) (0.108) (0.195)
∆yt−1 0.122 −0.145 0.025 0.042 −0.221∗∗
(0.608) (0.230) (0.051) (0.039) (0.107)
∆pt−2 −0.070 0.183∗∗
(0.078) (0.077)
Residual serial 2.406 6.212 7.640 8.061 7.240
correlation test [0.662] [0.184] [0.106] [0.089] [0.124]
Adjusted R2 0.122 0.209 0.466 0.659 0.124
Notes: The variables are ordered as: ∆x
0
t
, ∆eft, ∆mt, ∆pt, and ∆yt, where: ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the
oil exports revenues in U.S. dollars; ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly rial/U.S. dollar free market exchange
rate; ∆mt = (M2t − M2,t−1)/M2,t−1, M2t is the monetary aggregate M2 obtained by summing the aggregates M1
and "quasi-money"; ∆pt = ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran; ∆yt = ln(Yt/Yt−1), Yt is
the quarterly real output of Iran. Seasonal dummies are included to allow for possible seasonality of the variables in
the regressions of the SVAR model in Equation (1.6) with qt =

∆x
0
t
, ∆eft, ∆mt, ∆pt, ∆yt
0
. Numbers in parentheses
are heteroskedastic-consistent standard errors obtained following the approach of White (1980), and those in square
brackets are p-values. ***p < 0.01, **p < 0.05, *p < 0.1. "Residual serial correlation test" is the Breusch–Godfrey LM
test of serially uncorrelated errors with lag order of the test set to 4.
See the notes to Table A.9 for further details on the sources and construction of data used.
20

Table A.44: Quarterly estimates of the equation for the oil export variable in the SVAR model
of Iran with domestic variables ordered as: oil exports, exchange rate returns, money supply
growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆x
0
t
(1) (2) (3) (4) (5) (6)
st 0.144 0.107 0.104 0.087 0.088 0.088
(0.118) (0.106) (0.107) (0.110) (0.110) (0.111)
st−1 −0.339∗∗ −0.288∗∗ −0.286∗∗ −0.279∗∗ −0.261∗∗ −0.261∗∗
(0.145) (0.136) (0.138) (0.130) (0.131) (0.131)
∆x
0
t−1 −0.035 −0.051 −0.050 −0.077 −0.088 −0.088
(0.102) (0.107) (0.109) (0.106) (0.105) (0.105)
∆ef,t−1 −0.442∗∗ −0.441∗ −0.443∗ −0.382 −0.432∗ −0.432∗
(0.224) (0.238) (0.240) (0.237) (0.221) (0.222)
∆mt−1 −0.128 −0.715 −0.744 −1.214∗ −1.141 −1.143
(0.842) (0.765) (0.792) (0.725) (0.713) (0.731)
∆pt−1 −0.156 0.052 0.060 −0.213 −0.041 −0.043
(0.690) (0.707) (0.721) (0.734) (0.756) (0.743)
∆yt−1 0.087 0.122 0.116 −0.062 −0.103 −0.103
(0.598) (0.608) (0.615) (0.550) (0.539) (0.541)
∆ywt 8.406∗∗ 8.132∗∗ 4.185 3.098 3.085
(3.630) (3.629) (2.946) (3.008) (3.008)
∆reqwt 0.056 −0.046 −0.154 −0.158
(0.270) (0.233) (0.224) (0.288)
∆ewt −3.507∗∗∗ −3.551∗∗∗ −3.548∗∗∗
(0.958) (0.936) (0.962)
∆rwt 63.486∗∗ 63.416∗∗
(26.040) (26.307)
grvt −0.015
(0.583)
Residual serial 1.202 2.406 2.446 2.382 5.176 5.166
correlation test [0.878] [0.662] [0.654] [0.666] [0.270] [0.271]
Adjusted R2 0.089 0.122 0.115 0.210 0.247 0.240
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
203

Table A.45: Quarterly estimates of the equation for exchange rate returns in the SVAR model
of Iran with domestic variables ordered as: oil exports, exchange rate returns, money supply
growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆ef t
(1) (2) (3) (4) (5) (6)
st 0.295∗∗∗ 0.305∗∗∗ 0.302∗∗∗ 0.303∗∗∗ 0.303∗∗∗ 0.307∗∗∗
(0.081) (0.080) (0.081) (0.080) (0.080) (0.077)
st−1 −0.221∗∗∗ −0.233∗∗∗ −0.230∗∗∗ −0.226∗∗∗ −0.226∗∗∗ −0.231∗∗∗
(0.066) (0.065) (0.066) (0.066) (0.066) (0.064)
∆x
0
t 0.014 0.029 0.028 0.047 0.048 0.048
(0.038) (0.041) (0.040) (0.045) (0.047) (0.045)
∆x
0
t−1
0.038 0.044 0.045 0.051 0.051 0.047
(0.035) (0.036) (0.037) (0.039) (0.039) (0.038)
∆ef,t−1 0.344∗ 0.350∗∗ 0.348∗∗ 0.346∗∗ 0.348∗∗ 0.346∗∗
(0.180) (0.175) (0.174) (0.170) (0.173) (0.165)
∆mt−1 −0.037 0.149 0.121 0.209 0.209 0.110
(0.272) (0.266) (0.260) (0.285) (0.286) (0.271)
∆pt−1 −0.278 −0.341 −0.333 −0.291 −0.295 −0.397
(0.615) (0.610) (0.616) (0.604) (0.595) (0.589)
∆yt−1 −0.133 −0.145 −0.151 −0.125 −0.124 −0.120
(0.232) (0.230) (0.234) (0.232) (0.230) (0.227)
∆ywt −2.639 −2.897 −2.423 −2.401 −2.907∗
(1.641) (1.808) (1.740) (1.772) (1.671)
∆reqwt 0.053 0.068 0.071 −0.121
(0.090) (0.093) (0.099) (0.126)
∆ewt 0.554 0.560 0.648
(0.553) (0.540) (0.559)
∆rwt −1.633 −4.492
(15.026) (14.853)
grvt −0.625∗∗
(0.305)
Residual serial 5.972 6.212 5.961 7.940 8.059 5.259
correlation test [0.201] [0.184] [0.202] [0.094] [0.089] [0.262]
Adjusted R2 0.196 0.209 0.203 0.207 0.200 0.221
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
204

Table A.46: Quarterly estimates of the equation for money supply growth in the SVAR model
of Iran with domestic variables ordered as: oil exports, exchange rate returns, money supply
growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆mt
(1) (2) (3) (4) (5) (6)
st −0.0004 −0.002 −0.003 −0.002 −0.002 0.002
(0.014) (0.015) (0.014) (0.014) (0.014) (0.013)
st−1 0.014 0.015 0.016 0.017 0.017 0.014
(0.017) (0.018) (0.018) (0.017) (0.017) (0.016)
∆x
0
t 0.008 0.006 0.006 0.011 0.014∗ 0.015∗
(0.007) (0.007) (0.007) (0.007) (0.008) (0.008)
∆ef,t −0.009 −0.007 −0.008 −0.011 −0.012 −0.022
(0.017) (0.018) (0.018) (0.018) (0.017) (0.018)
∆x
0
t−1 −0.005 −0.005 −0.005 −0.003 −0.002 −0.003
(0.007) (0.007) (0.008) (0.008) (0.008) (0.008)
∆ef,t−1 −0.024 −0.025 −0.026 −0.027 −0.023 −0.021
(0.022) (0.022) (0.022) (0.022) (0.021) (0.021)
∆mt−1 0.235∗ 0.218 0.211 0.236 0.236 0.211
(0.139) (0.146) (0.146) (0.156) (0.156) (0.149)
∆pt−1 0.165∗ 0.167∗ 0.174∗ 0.193∗∗ 0.189∗∗ 0.163
(0.089) (0.088) (0.090) (0.094) (0.093) (0.100)
∆yt−1 0.022 0.025 0.022 0.026 0.028 0.026
(0.051) (0.051) (0.051) (0.050) (0.051) (0.050)
∆pt−2 −0.076 −0.070 −0.077 −0.088 −0.094 −0.102
(0.079) (0.078) (0.079) (0.082) (0.086) (0.088)
∆ywt 0.233 0.152 0.260 0.304 0.134
(0.406) (0.441) (0.464) (0.457) (0.506)
∆reqwt 0.016 0.021 0.027 −0.026
(0.026) (0.027) (0.029) (0.039)
∆ewt 0.146 0.160 0.191∗
(0.108) (0.114) (0.114)
∆rwt −3.581 −4.412
(2.704) (2.695)
grvt −0.176∗∗
(0.078)
Residual serial 7.428 7.640 7.255 6.129 5.742 4.178
correlation test [0.115] [0.106] [0.123] [0.190] [0.219] [0.382]
Adjusted R2 0.469 0.466 0.462 0.467 0.470 0.491
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
205

Table A.47: Quarterly estimates of the equation for inflation in the SVAR model of Iran with domestic variables ordered as: oil exports, exchange rate returns, money supply growth, inflation,
and output growth, estimated over the period 1989q1-2019q4
∆pt
(1) (2) (3) (4) (5) (6)
st −0.028∗∗ −0.033∗∗∗ −0.032∗∗ −0.033∗∗∗ −0.033∗∗∗ −0.032∗∗
(0.012) (0.013) (0.013) (0.013) (0.013) (0.013)
st−1 0.032∗∗∗ 0.037∗∗∗ 0.036∗∗∗ 0.036∗∗∗ 0.036∗∗∗ 0.035∗∗∗
(0.012) (0.012) (0.012) (0.013) (0.013) (0.013)
∆x
0
t 0.001 −0.003 −0.003 −0.007 −0.007 −0.007
(0.008) (0.008) (0.008) (0.009) (0.009) (0.009)
∆ef,t 0.155∗∗∗ 0.163∗∗∗ 0.164∗∗∗ 0.166∗∗∗ 0.166∗∗∗ 0.163∗∗∗
(0.029) (0.028) (0.028) (0.028) (0.028) (0.028)
∆mt −0.061 −0.073 −0.070 −0.058 −0.057 −0.070
(0.076) (0.072) (0.071) (0.072) (0.074) (0.075)
∆x
0
t−1 −0.001 −0.003 −0.003 −0.004 −0.004 −0.005
(0.008) (0.008) (0.008) (0.009) (0.009) (0.009)
∆ef,t−1 −0.007 −0.009 −0.007 −0.007 −0.007 −0.007
(0.029) (0.027) (0.027) (0.027) (0.027) (0.027)
∆mt−1 0.035 −0.025 −0.016 −0.036 −0.036 −0.040
(0.074) (0.072) (0.074) (0.077) (0.078) (0.079)
∆pt−1 0.480∗∗∗ 0.488∗∗∗ 0.478∗∗∗ 0.462∗∗∗ 0.462∗∗∗ 0.458∗∗∗
(0.113) (0.108) (0.108) (0.107) (0.108) (0.110)
∆yt−1 0.033 0.042 0.045 0.042 0.042 0.042
(0.041) (0.039) (0.040) (0.039) (0.039) (0.039)
∆pt−2 0.162∗∗ 0.183∗∗ 0.192∗∗ 0.201∗∗ 0.202∗∗ 0.198∗∗
(0.080) (0.077) (0.078) (0.079) (0.080) (0.080)
∆ywt 0.865∗∗∗ 0.971∗∗∗ 0.893∗∗∗ 0.889∗∗∗ 0.847∗∗∗
(0.285) (0.315) (0.330) (0.335) (0.325)
∆reqwt −0.021 −0.024 −0.025 −0.039
(0.021) (0.022) (0.023) (0.029)
∆ewt −0.102 −0.104 −0.093
(0.088) (0.091) (0.091)
∆rwt 0.303 0.031
(2.528) (2.514)
grvt −0.048
(0.056)
Residual serial 9.241 8.061 5.714 6.473 6.510 6.759
correlation test [0.055] [0.089] [0.222] [0.166] [0.164] [0.149]
Adjusted R2 0.635 0.659 0.658 0.660 0.656 0.655
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
206

Table A.48: Quarterly estimates of the equation for output growth in the SVAR model of Iran
with domestic variables ordered as: oil exports, exchange rate returns, money supply growth,
inflation, and output growth, estimated over the period 1989q1-2019q4
∆yt
(1) (2) (3) (4) (5) (6)
st 0.024 0.029 0.028 0.026 0.026 0.027
(0.023) (0.024) (0.024) (0.023) (0.023) (0.023)
st−1 −0.051∗ −0.056∗∗ −0.055∗∗ −0.055∗∗ −0.055∗∗ −0.056∗∗
(0.027) (0.028) (0.028) (0.027) (0.027) (0.027)
∆x
0
t 0.023 0.025 0.025 0.020 0.025 0.026
(0.017) (0.016) (0.017) (0.019) (0.018) (0.018)
∆ef,t −0.130∗∗∗ −0.141∗∗∗ −0.142∗∗∗ −0.135∗∗∗ −0.136∗∗∗ −0.138∗∗∗
(0.041) (0.042) (0.043) (0.041) (0.041) (0.041)
∆mt 0.052 0.063 0.062 0.078 0.054 0.037
(0.118) (0.118) (0.118) (0.115) (0.120) (0.123)
∆pt 0.348∗∗ 0.387∗∗ 0.390∗∗ 0.373∗∗ 0.373∗∗ 0.364∗∗
(0.165) (0.173) (0.174) (0.173) (0.174) (0.175)
∆x
0
t−1
0.022∗ 0.023∗ 0.024∗ 0.022∗ 0.023∗ 0.023∗
(0.011) (0.012) (0.013) (0.012) (0.013) (0.013)
∆ef,t−1 0.037 0.041 0.041 0.040 0.046 0.047
(0.044) (0.044) (0.044) (0.045) (0.042) (0.043)
∆mt−1 0.013 0.046 0.039 0.009 0.014 0.009
(0.126) (0.129) (0.129) (0.132) (0.137) (0.140)
∆pt−1 −0.466∗∗ −0.505∗∗∗ −0.505∗∗∗ −0.506∗∗ −0.519∗∗∗ −0.523∗∗∗
(0.184) (0.195) (0.196) (0.198) (0.194) (0.192)
∆yt−1 −0.218∗∗ −0.221∗∗ −0.223∗∗ −0.230∗∗ −0.225∗∗ −0.224∗∗
(0.108) (0.107) (0.108) (0.105) (0.106) (0.108)
∆ywt −0.520 −0.595 −0.708 −0.619 −0.664
(0.458) (0.510) (0.526) (0.527) (0.522)
∆reqwt 0.015 0.010 0.021 0.003
(0.042) (0.044) (0.046) (0.066)
∆ewt −0.160 −0.133 −0.121
(0.175) (0.174) (0.174)
∆rwt −6.160 −6.499
(4.596) (4.779)
grvt −0.060
(0.129)
Residual serial 7.242 7.240 7.371 7.721 8.049 8.248
correlation test [0.124] [0.124] [0.118] [0.102] [0.090] [0.083]
Adjusted R2 0.126 0.124 0.117 0.117 0.126 0.120
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
207

Table A.49: Quarterly estimates of the SVAR model of Iran with domestic variables ordered
as: oil exports, exchange rate returns, inflation, and output growth, estimated over the period
1989q1-2019q4
∆x
0
t ∆ef,t ∆pt ∆yt
(1) (2) (3) (4)
st 0.119 0.302∗∗∗ −0.033∗∗∗ 0.027
(0.102) (0.079) (0.013) (0.024)
st−1 −0.308∗∗ −0.229∗∗∗ 0.035∗∗∗ −0.053∗
(0.132) (0.063) (0.012) (0.027)
∆x
0
t 0.028 −0.003 0.025
(0.040) (0.008) (0.016)
∆ef,t 0.163∗∗∗ −0.139∗∗∗
(0.028) (0.042)
∆pt 0.377∗∗
(0.172)
∆ywt 7.638∗∗ −2.471 0.800∗∗∗ −0.428
(3.472) (1.572) (0.289) (0.428)
∆x
0
t−1 −0.053 0.044 −0.003 0.023∗
(0.106) (0.036) (0.008) (0.012)
∆ef,t−1 −0.429∗ 0.347∗∗ −0.007 0.038
(0.236) (0.173) (0.026) (0.044)
∆pt−1 0.022 −0.335 0.478∗∗∗ −0.489∗∗
(0.698) (0.610) (0.103) (0.190)
∆yt−1 0.132 −0.147 0.039 −0.220∗∗
(0.605) (0.232) (0.039) (0.105)
∆pt−2 0.182∗∗
(0.071)
Residual serial 2.027 5.689 7.970 6.703
correlation test [0.731] [0.224] [0.093] [0.152]
Adjusted R2 0.126 0.215 0.661 0.137
Notes: The variables are ordered as: ∆x
0
t
, ∆eft, ∆pt, and ∆yt, where: ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the oil exports
revenues in U.S. dollars; ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly rial/U.S. dollar free market exchange rate; ∆pt =
ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran; ∆yt = ln(Yt/Yt−1), Yt is the quarterly real output of
Iran. st is the quarterly sanctions intensity variable. Seasonal dummies are included to allow for possible seasonality
of the variables in the regressions of the SVAR model in Equation (1.6) with qt =

∆x
0
t
, ∆eft, ∆pt, ∆yt
0
and zwt =
(∆ywt)
0
. Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980), and those in square brackets are p-values. ***p < 0.01, **p < 0.05, *p < 0.1. "Residual serial correlation
test" is the Breusch–Godfrey LM test of serially uncorrelated errors with lag order of the test set to 4.
See the notes to Table A.9 for further details on the sources and construction of data used.
20

Table A.50: Quarterly estimates of the equation for the oil exports variable in the SVAR model of
Iran with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆x
0
t
(1) (2) (3) (4) (5) (6)
st 0.146 0.119 0.119 0.112 0.111 0.111
(0.114) (0.102) (0.102) (0.107) (0.107) (0.109)
st−1 −0.342∗∗ −0.308∗∗ −0.307∗∗ −0.313∗∗ −0.294∗∗ −0.293∗∗
(0.139) (0.132) (0.134) (0.130) (0.130) (0.130)
∆x
0
t−1 −0.035 −0.053 −0.053 −0.080 −0.091 −0.091
(0.101) (0.106) (0.107) (0.105) (0.104) (0.104)
∆ef,t−1 −0.440∗∗ −0.429∗ −0.429∗ −0.363 −0.414∗ −0.414∗
(0.220) (0.236) (0.239) (0.239) (0.223) (0.225)
∆pt−1 −0.158 0.022 0.024 −0.257 −0.079 −0.064
(0.683) (0.698) (0.709) (0.734) (0.754) (0.742)
∆yt−1 0.089 0.132 0.131 −0.030 −0.074 −0.075
(0.594) (0.605) (0.611) (0.554) (0.546) (0.547)
∆ywt 7.638∗∗ 7.545∗∗ 3.436 2.377 2.461
(3.472) (3.530) (2.880) (2.956) (2.954)
∆reqwt 0.017 −0.102 −0.209 −0.179
(0.259) (0.226) (0.217) (0.295)
∆ewt −3.337∗∗∗ −3.391∗∗∗ −3.407∗∗∗
(0.937) (0.919) (0.951)
∆rwt 64.555∗∗ 64.968∗∗
(26.212) (26.557)
grvt 0.093
(0.587)
Residual serial 1.157 2.027 2.052 1.565 3.775 3.937
correlation test [0.885] [0.731] [0.726] [0.815] [0.437] [0.415]
Adjusted R2 0.097 0.126 0.118 0.205 0.243 0.236
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
209

Table A.51: Quarterly estimates of the equation for exchange rate returns in the SVAR model of
Iran with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆ef t
(1) (2) (3) (4) (5) (6)
st 0.296∗∗∗ 0.302∗∗∗ 0.300∗∗∗ 0.299∗∗∗ 0.299∗∗∗ 0.305∗∗∗
(0.080) (0.079) (0.079) (0.077) (0.078) (0.076)
st−1 −0.222∗∗∗ −0.229∗∗∗ −0.227∗∗∗ −0.221∗∗∗ −0.221∗∗∗ −0.228∗∗∗
(0.065) (0.063) (0.064) (0.064) (0.064) (0.062)
∆x
0
t 0.014 0.028 0.027 0.044 0.045 0.046
(0.038) (0.040) (0.040) (0.043) (0.045) (0.044)
∆x
0
t−1
0.038 0.044 0.046 0.051 0.051 0.047
(0.035) (0.036) (0.036) (0.039) (0.039) (0.037)
∆ef,t−1 0.345∗ 0.347∗∗ 0.345∗∗ 0.342∗∗ 0.344∗∗ 0.344∗∗
(0.179) (0.173) (0.172) (0.168) (0.171) (0.164)
∆pt−1 −0.278 −0.335 −0.327 −0.284 −0.289 −0.395
(0.613) (0.610) (0.616) (0.603) (0.594) (0.587)
∆yt−1 −0.132 −0.147 −0.153 −0.130 −0.129 −0.123
(0.232) (0.232) (0.235) (0.236) (0.234) (0.229)
∆ywt −2.471 −2.795 −2.284 −2.262 −2.843∗
(1.572) (1.757) (1.705) (1.742) (1.643)
∆reqwt 0.059 0.077 0.080 −0.120
(0.091) (0.096) (0.102) (0.125)
∆ewt 0.515 0.522 0.629
(0.541) (0.526) (0.544)
∆rwt −1.651 −4.548
(15.034) (14.849)
grvt −0.636∗∗
(0.310)
Residual serial 6.132 5.689 5.541 7.009 7.125 4.742
correlation test [0.190] [0.224] [0.236] [0.135] [0.129] [0.315]
Adjusted R2 0.203 0.215 0.209 0.212 0.205 0.228
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
210

Table A.52: Quarterly estimates of the equation for inflation in the SVAR model of Iran with
domestic variables ordered as: oil exports, exchange rate returns, inflation, and output growth,
estimated over the period 1989q1-2019q4
∆pt
(1) (2) (3) (4) (5) (6)
st −0.029∗∗ −0.033∗∗∗ −0.032∗∗ −0.032∗∗ −0.032∗∗ −0.031∗∗
(0.013) (0.013) (0.013) (0.012) (0.012) (0.013)
st−1 0.032∗∗∗ 0.035∗∗∗ 0.034∗∗∗ 0.034∗∗∗ 0.034∗∗∗ 0.033∗∗∗
(0.012) (0.012) (0.012) (0.012) (0.012) (0.013)
∆x
0
t 0.001 −0.003 −0.003 −0.007 −0.007 −0.007
(0.008) (0.008) (0.008) (0.008) (0.008) (0.008)
∆ef,t 0.156∗∗∗ 0.163∗∗∗ 0.164∗∗∗ 0.166∗∗∗ 0.166∗∗∗ 0.165∗∗∗
(0.029) (0.028) (0.028) (0.027) (0.028) (0.027)
∆x
0
t−1 −0.0003 −0.003 −0.003 −0.004 −0.004 −0.004
(0.008) (0.008) (0.008) (0.009) (0.009) (0.009)
∆ef,t−1 −0.005 −0.007 −0.005 −0.005 −0.005 −0.005
(0.028) (0.026) (0.026) (0.026) (0.026) (0.026)
∆pt−1 0.468∗∗∗ 0.478∗∗∗ 0.467∗∗∗ 0.454∗∗∗ 0.455∗∗∗ 0.451∗∗∗
(0.107) (0.103) (0.102) (0.102) (0.102) (0.105)
∆yt−1 0.032 0.039 0.044 0.041 0.040 0.040
(0.041) (0.039) (0.040) (0.039) (0.040) (0.040)
∆pt−2 0.169∗∗ 0.182∗∗ 0.194∗∗∗ 0.200∗∗∗ 0.201∗∗∗ 0.199∗∗∗
(0.074) (0.071) (0.073) (0.073) (0.074) (0.074)
∆ywt 0.800∗∗∗ 0.932∗∗∗ 0.842∗∗ 0.836∗∗ 0.803∗∗
(0.289) (0.321) (0.337) (0.340) (0.339)
∆reqwt −0.023 −0.027 −0.028 −0.038
(0.021) (0.022) (0.023) (0.029)
∆ewt −0.101 −0.103 −0.097
(0.082) (0.083) (0.084)
∆rwt 0.501 0.356
(2.435) (2.425)
grvt −0.031
(0.054)
Residual serial 11.263 7.970 5.559 6.203 6.210 6.281
correlation test [0.024] [0.093] [0.235] [0.184] [0.184] [0.179]
Adjusted R2 0.640 0.661 0.661 0.663 0.660 0.657
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
211

Table A.53: Quarterly estimates of the equation for output growth in the SVAR model of Iran
with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆yt
(1) (2) (3) (4) (5) (6)
st 0.024 0.027 0.027 0.025 0.025 0.027
(0.023) (0.024) (0.024) (0.023) (0.023) (0.023)
st−1 −0.050∗ −0.053∗ −0.053∗ −0.053∗∗ −0.053∗∗ −0.055∗∗
(0.027) (0.027) (0.027) (0.027) (0.027) (0.027)
∆x
0
t 0.023 0.025 0.025 0.020 0.026 0.026
(0.016) (0.016) (0.017) (0.019) (0.017) (0.017)
∆ef,t −0.130∗∗∗ −0.139∗∗∗ −0.140∗∗∗ −0.133∗∗∗ −0.135∗∗∗ −0.138∗∗∗
(0.041) (0.042) (0.042) (0.041) (0.041) (0.041)
∆pt 0.344∗∗ 0.377∗∗ 0.381∗∗ 0.362∗∗ 0.365∗∗ 0.359∗∗
(0.164) (0.172) (0.173) (0.173) (0.172) (0.171)
∆x
0
t−1
0.022∗ 0.023∗ 0.023∗ 0.022∗ 0.023∗ 0.023∗
(0.011) (0.012) (0.013) (0.012) (0.013) (0.013)
∆ef,t−1 0.036 0.038 0.038 0.037 0.045 0.046
(0.044) (0.044) (0.043) (0.044) (0.041) (0.042)
∆pt−1 −0.458∗∗ −0.489∗∗ −0.489∗∗ −0.489∗∗ −0.507∗∗∗ −0.516∗∗∗
(0.179) (0.190) (0.190) (0.193) (0.188) (0.186)
∆yt−1 −0.216∗∗ −0.220∗∗ −0.221∗∗ −0.227∗∗ −0.223∗∗ −0.223∗∗
(0.106) (0.105) (0.106) (0.103) (0.105) (0.107)
∆ywt −0.428 −0.531 −0.658 −0.577 −0.644
(0.428) (0.480) (0.497) (0.499) (0.501)
∆reqwt 0.018 0.012 0.023 0.002
(0.042) (0.045) (0.047) (0.065)
∆ewt −0.155 −0.131 −0.118
(0.165) (0.164) (0.164)
∆rwt −6.346 −6.663
(4.556) (4.697)
grvt −0.068
(0.121)
Residual serial 6.974 6.703 6.911 7.426 7.684 8.064
correlation test [0.137] [0.152] [0.141] [0.115] [0.104] [0.089]
Adjusted R2 0.141 0.137 0.131 0.131 0.141 0.136
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
212

Table A.54: Quarterly estimates of the SVAR model of Iran with domestic variables ordered
as: oil exports, exchange rate returns, inflation, and output growth, estimated over the period
1989q1-2019q4
∆x
0
t ∆ef,t ∆pt ∆yt
(1) (2) (3) (4)
st 0.111 0.309∗∗∗ −0.033∗∗∗ 0.028
(0.105) (0.077) (0.012) (0.024)
st−1 −0.305∗∗ −0.235∗∗∗ 0.036∗∗∗ −0.054∗
(0.129) (0.066) (0.012) (0.027)
∆x
0
t 0.018 −0.004 0.028∗
(0.035) (0.007) (0.015)
∆ef,t 0.162∗∗∗ −0.138∗∗∗
(0.027) (0.040)
∆pt 0.364∗∗
(0.168)
∆ywt 7.674∗∗ −2.452 0.808∗∗∗ −0.459
(3.771) (1.550) (0.293) (0.418)
∆x
0
t−1 −0.063 0.041 −0.002 0.021∗
(0.104) (0.035) (0.008) (0.012)
∆ef,t−1 −0.361 0.332∗∗ −0.006 0.041
(0.249) (0.169) (0.027) (0.043)
∆pt−1 −0.059 −0.338 0.477∗∗∗ −0.482∗∗
(0.757) (0.604) (0.102) (0.191)
∆yt−1 0.125 −0.135 0.040 −0.223∗∗
(0.609) (0.224) (0.038) (0.106)
∆pt−2 0.184∗∗∗
(0.070)
Residual serial 3.751 4.983 8.003 6.738
correlation test [0.441] [0.289] [0.091] [0.150]
Adjusted R2 0.097 0.214 0.668 0.152
Notes: The variables are ordered as: ∆x
0
t
, ∆eft, ∆pt, and ∆yt, where: ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the oil exports
revenues in U.S. dollars; ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly rial/U.S. dollar free market exchange rate; ∆pt =
ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran; ∆yt = ln(Yt/Yt−1), Yt is the quarterly real output of Iran.
st is the quarterly sanctions intensity variable. Seasonal dummies are not included to allow for possible seasonality
of the variables in the SVAR model. Numbers in parentheses are heteroskedastic-consistent standard errors obtained
following the approach of White (1980), and those in square brackets are p-values. ***p < 0.01, **p < 0.05, *p < 0.1.
"Residual serial correlation test" is the Breusch–Godfrey LM test of serially uncorrelated errors with lag order of the test
set to 4.
See the notes to Table A.9 for further details on the sources and construction of data used.
213

Table A.55: Quarterly estimates of the equation for the oil exports variable in the SVAR model of
Iran with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆x
0
t
(1) (2) (3) (4) (5) (6)
st 0.136 0.111 0.109 0.106 0.104 0.104
(0.117) (0.105) (0.105) (0.108) (0.106) (0.108)
st−1 −0.338∗∗ −0.305∗∗ −0.304∗∗ −0.316∗∗ −0.295∗∗ −0.294∗∗
(0.137) (0.129) (0.131) (0.127) (0.124) (0.125)
∆x
0
t−1 −0.043 −0.063 −0.061 −0.094 −0.103 −0.103
(0.098) (0.104) (0.106) (0.106) (0.108) (0.108)
∆ef,t−1 −0.370 −0.361 −0.362 −0.305 −0.364 −0.363
(0.232) (0.249) (0.250) (0.254) (0.238) (0.240)
∆pt−1 −0.243 −0.059 −0.054 −0.327 −0.123 −0.116
(0.748) (0.757) (0.767) (0.799) (0.815) (0.808)
∆yt−1 0.083 0.125 0.121 −0.056 −0.107 −0.108
(0.596) (0.609) (0.616) (0.546) (0.542) (0.543)
∆ywt 7.674∗∗ 7.463∗∗ 3.031 2.068 2.109
(3.771) (3.617) (2.877) (2.933) (2.925)
∆reqwt 0.039 −0.069 −0.196 −0.182
(0.283) (0.237) (0.232) (0.300)
∆ewt −3.469∗∗∗ −3.514∗∗∗ −3.522∗∗∗
(0.999) (0.970) (1.001)
∆rwt 66.900∗∗∗ 67.105∗∗∗
(24.601) (25.154)
grvt 0.044
(0.590)
Residual serial 2.435 3.751 3.865 2.464 4.895 5.026
correlation test [0.656] [0.441] [0.425] [0.651] [0.298] [0.285]
Adjusted R2 0.068 0.097 0.089 0.183 0.225 0.218
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
214

Table A.56: Quarterly estimates of the equation for exchange rate returns in the SVAR model of
Iran with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆ef t
(1) (2) (3) (4) (5) (6)
st 0.302∗∗∗ 0.309∗∗∗ 0.305∗∗∗ 0.304∗∗∗ 0.304∗∗∗ 0.310∗∗∗
(0.079) (0.077) (0.077) (0.076) (0.076) (0.074)
st−1 −0.229∗∗∗ −0.235∗∗∗ −0.232∗∗∗ −0.226∗∗∗ −0.226∗∗∗ −0.233∗∗∗
(0.067) (0.066) (0.066) (0.067) (0.067) (0.065)
∆x
0
t 0.005 0.018 0.017 0.034 0.033 0.033
(0.033) (0.035) (0.034) (0.039) (0.040) (0.039)
∆x
0
t−1
0.035 0.041 0.044 0.050 0.050 0.045
(0.034) (0.035) (0.036) (0.039) (0.039) (0.037)
∆ef,t−1 0.331∗ 0.332∗∗ 0.330∗∗ 0.327∗∗ 0.326∗ 0.324∗∗
(0.175) (0.169) (0.168) (0.163) (0.168) (0.160)
∆pt−1 −0.282 −0.338 −0.329 −0.287 −0.284 −0.384
(0.608) (0.604) (0.610) (0.599) (0.591) (0.582)
∆yt−1 −0.121 −0.135 −0.142 −0.118 −0.119 −0.114
(0.225) (0.224) (0.227) (0.226) (0.223) (0.218)
∆ywt −2.452 −2.855∗ −2.319 −2.333 −2.908∗
(1.550) (1.720) (1.670) (1.692) (1.586)
∆reqwt 0.075 0.090 0.088 −0.104
(0.086) (0.087) (0.095) (0.124)
∆ewt 0.515 0.510 0.624
(0.542) (0.529) (0.552)
∆rwt 1.170 −1.757
(14.412) (14.384)
grvt −0.621∗∗
(0.313)
Residual serial 5.353 4.983 4.839 6.292 6.236 4.234
correlation test [0.253] [0.289] [0.304] [0.178] [0.182] [0.375]
Adjusted R2 0.203 0.214 0.209 0.212 0.205 0.226
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
215

Table A.57: Quarterly estimates of the equation for inflation in the SVAR model of Iran with
domestic variables ordered as: oil exports, exchange rate returns, inflation, and output growth,
estimated over the period 1989q1-2019q4
∆pt
(1) (2) (3) (4) (5) (6)
st −0.029∗∗ −0.033∗∗∗ −0.032∗∗ −0.032∗∗∗ −0.032∗∗∗ −0.031∗∗
(0.013) (0.012) (0.013) (0.012) (0.012) (0.013)
st−1 0.032∗∗∗ 0.036∗∗∗ 0.035∗∗∗ 0.034∗∗∗ 0.034∗∗∗ 0.033∗∗∗
(0.012) (0.012) (0.012) (0.012) (0.012) (0.012)
∆x
0
t 0.0005 −0.004 −0.003 −0.007 −0.007 −0.007
(0.007) (0.007) (0.007) (0.008) (0.008) (0.008)
∆ef,t 0.155∗∗∗ 0.162∗∗∗ 0.163∗∗∗ 0.166∗∗∗ 0.166∗∗∗ 0.164∗∗∗
(0.028) (0.027) (0.027) (0.027) (0.027) (0.027)
∆x
0
t−1
0.001 −0.002 −0.002 −0.004 −0.004 −0.004
(0.009) (0.008) (0.008) (0.009) (0.009) (0.009)
∆ef,t−1 −0.005 −0.006 −0.004 −0.004 −0.005 −0.004
(0.029) (0.027) (0.026) (0.026) (0.026) (0.027)
∆pt−1 0.466∗∗∗ 0.477∗∗∗ 0.465∗∗∗ 0.454∗∗∗ 0.454∗∗∗ 0.450∗∗∗
(0.107) (0.102) (0.101) (0.100) (0.101) (0.103)
∆yt−1 0.033 0.040 0.044 0.041 0.040 0.040
(0.039) (0.038) (0.039) (0.038) (0.039) (0.039)
∆pt−2 0.172∗∗ 0.184∗∗∗ 0.196∗∗∗ 0.201∗∗∗ 0.201∗∗∗ 0.199∗∗∗
(0.073) (0.070) (0.072) (0.073) (0.073) (0.073)
∆ywt 0.808∗∗∗ 0.952∗∗∗ 0.854∗∗ 0.850∗∗ 0.815∗∗
(0.293) (0.322) (0.338) (0.340) (0.337)
∆reqwt −0.026 −0.029 −0.030 −0.040
(0.020) (0.021) (0.022) (0.027)
∆ewt −0.103 −0.105 −0.098
(0.080) (0.082) (0.083)
∆rwt 0.317 0.167
(2.317) (2.313)
grvt −0.032
(0.054)
Residual serial 11.521 8.003 5.321 5.888 5.868 5.924
correlation test [0.021] [0.091] [0.256] [0.208] [0.209] [0.205]
Adjusted R2 0.647 0.668 0.669 0.671 0.668 0.666
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
216

Table A.58: Quarterly estimates of the equation for output growth in the SVAR model of Iran
with domestic variables ordered as: oil exports, exchange rate returns, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆yt
(1) (2) (3) (4) (5) (6)
st 0.024 0.028 0.027 0.026 0.026 0.027
(0.023) (0.024) (0.024) (0.023) (0.023) (0.023)
st−1 −0.050∗ −0.054∗ −0.053∗∗ −0.054∗∗ −0.054∗∗ −0.055∗∗
(0.027) (0.027) (0.027) (0.027) (0.026) (0.027)
∆x
0
t 0.025∗ 0.028∗ 0.028∗ 0.023 0.028∗ 0.028∗
(0.015) (0.015) (0.016) (0.017) (0.016) (0.016)
∆ef,t −0.128∗∗∗ −0.138∗∗∗ −0.140∗∗∗ −0.134∗∗∗ −0.134∗∗∗ −0.137∗∗∗
(0.039) (0.040) (0.040) (0.039) (0.039) (0.039)
∆pt 0.328∗∗ 0.364∗∗ 0.371∗∗ 0.353∗∗ 0.355∗∗ 0.348∗∗
(0.161) (0.168) (0.170) (0.169) (0.167) (0.166)
∆x
0
t−1
0.019∗ 0.021∗ 0.022∗ 0.020 0.021∗ 0.021
(0.011) (0.012) (0.012) (0.012) (0.013) (0.013)
∆ef,t−1 0.038 0.041 0.041 0.040 0.047 0.047
(0.043) (0.043) (0.043) (0.043) (0.041) (0.041)
∆pt−1 −0.449∗∗ −0.482∗∗ −0.484∗∗ −0.483∗∗ −0.500∗∗∗ −0.508∗∗∗
(0.181) (0.191) (0.191) (0.193) (0.188) (0.185)
∆yt−1 −0.219∗∗ −0.223∗∗ −0.225∗∗ −0.231∗∗ −0.226∗∗ −0.226∗∗
(0.108) (0.106) (0.107) (0.104) (0.106) (0.108)
∆ywt −0.459 −0.595 −0.715 −0.645 −0.710
(0.418) (0.478) (0.498) (0.501) (0.504)
∆reqwt 0.024 0.019 0.031 0.011
(0.041) (0.044) (0.046) (0.064)
∆ewt −0.139 −0.117 −0.103
(0.159) (0.159) (0.159)
∆rwt −5.988 −6.293
(4.408) (4.539)
grvt −0.066
(0.117)
Residual serial 7.054 6.738 6.977 7.483 7.876 8.205
correlation test [0.133] [0.150] [0.137] [0.112] [0.096] [0.084]
Adjusted R2 0.154 0.152 0.146 0.145 0.154 0.149
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
217

Table A.59: Quarterly estimates of the SVAR model of Iran with domestic variables ordered as:
exchange rate returns, oil exports, money supply growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆ef,t ∆x
0
t ∆mt ∆pt ∆yt
(1) (2) (3) (4) (5)
st 0.308∗∗∗ 0.058 −0.002 −0.033∗∗∗ 0.029
(0.081) (0.126) (0.015) (0.013) (0.024)
st−1 −0.241∗∗∗ −0.250∗ 0.015 0.037∗∗∗ −0.056∗∗
(0.067) (0.151) (0.018) (0.012) (0.028)
∆ef,t 0.158 −0.007 0.163∗∗∗ −0.141∗∗∗
(0.205) (0.018) (0.028) (0.042)
∆x
0
t 0.006 −0.003 0.025
(0.007) (0.008) (0.016)
∆mt −0.073 0.063
(0.072) (0.118)
∆pt 0.387∗∗
(0.173)
∆ywt −2.399 8.786∗∗ 0.233 0.865∗∗∗ −0.520
(1.525) (3.792) (0.406) (0.285) (0.458)
∆ef,t−1 0.337∗ −0.495∗ −0.025 −0.009 0.041
(0.177) (0.261) (0.022) (0.027) (0.044)
∆x
0
t−1
0.042 −0.058 −0.005 −0.003 0.023∗
(0.036) (0.106) (0.007) (0.008) (0.012)
∆mt−1 0.129 −0.735 0.218 −0.025 0.046
(0.268) (0.772) (0.146) (0.072) (0.129)
∆pt−1 −0.339 0.106 0.167∗ 0.488∗∗∗ −0.505∗∗∗
(0.611) (0.734) (0.088) (0.108) (0.195)
∆yt−1 −0.142 0.144 0.025 0.042 −0.221∗∗
(0.226) (0.616) (0.051) (0.039) (0.107)
∆pt−2 −0.070 0.183∗∗
(0.078) (0.077)
Residual serial 5.987 2.379 7.640 8.061 7.240
correlation test [0.200] [0.666] [0.106] [0.089] [0.124]
Adjusted R2 0.212 0.119 0.466 0.659 0.124
Notes: The variables are ordered as: ∆eft, ∆x
0
t
, ∆mt, ∆pt, and ∆yt, where: ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly rial/U.S. dollar free market exchange rate; ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the oil exports revenues in U.S.
dollars; ∆mt = (M2t − M2,t−1)/M2,t−1, M2t is the monetary aggregate M2 obtained by summing the aggregates
M1 and "quasi-money"; ∆pt = ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran; ∆yt = ln(Yt/Yt−1),
Yt is the quarterly real output of Iran. st is the quarterly sanctions intensity variable. Seasonal dummies are included to allow for possible seasonality of the variables in the regressions of the SVAR model in Equation (1.6) with
qt =

∆eft, ∆x
0
t
, ∆mt, ∆pt, ∆yt
0
and zwt = (∆ywt)
0
. Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of White (1980), and those in square brackets are p-values. ***p < 0.01,
**p < 0.05, *p < 0.1. "Residual serial correlation test" is the Breusch–Godfrey LM test of serially uncorrelated errors
with lag order of the test set to 4.
See the notes to Table A.9 for further details on the sources and construction of data used.
21

Table A.60: Quarterly estimates of the equation for the exchange rate returns in the SVAR model
of Iran with domestic variables ordered as: exchange rate returns, oil exports, money supply
growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆ef t
(1) (2) (3) (4) (5) (6)
st 0.297∗∗∗ 0.308∗∗∗ 0.305∗∗∗ 0.307∗∗∗ 0.307∗∗∗ 0.311∗∗∗
(0.081) (0.081) (0.081) (0.080) (0.081) (0.078)
st−1 −0.226∗∗∗ −0.241∗∗∗ −0.238∗∗∗ −0.239∗∗∗ −0.239∗∗∗ −0.243∗∗∗
(0.068) (0.067) (0.068) (0.068) (0.069) (0.068)
∆ef,t−1 0.338∗ 0.337∗ 0.335∗ 0.328∗ 0.327∗ 0.326∗∗
(0.182) (0.177) (0.176) (0.171) (0.173) (0.166)
∆x
0
t−1
0.038 0.042 0.044 0.047 0.047 0.043
(0.035) (0.036) (0.036) (0.038) (0.038) (0.036)
∆mt−1 −0.039 0.129 0.100 0.153 0.154 0.055
(0.270) (0.268) (0.263) (0.283) (0.281) (0.266)
∆pt−1 −0.280 −0.339 −0.332 −0.301 −0.297 −0.399
(0.612) (0.611) (0.617) (0.607) (0.597) (0.591)
∆yt−1 −0.132 −0.142 −0.148 −0.128 −0.129 −0.125
(0.230) (0.226) (0.230) (0.225) (0.223) (0.220)
∆ywt −2.399 −2.667 −2.228 −2.252 −2.759∗
(1.525) (1.690) (1.670) (1.707) (1.615)
∆reqwt 0.054 0.066 0.063 −0.129
(0.090) (0.093) (0.096) (0.122)
∆ewt 0.390 0.390 0.478
(0.516) (0.515) (0.530)
∆rwt 1.417 −1.456
(14.685) (14.625)
grvt −0.626∗∗
(0.302)
Residual serial 5.781 5.987 5.734 7.126 7.058 4.946
correlation test [0.216] [0.200] [0.220] [0.129] [0.133] [0.293]
Adjusted R2 0.202 0.212 0.206 0.205 0.198 0.220
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
219

Table A.61: Quarterly estimates of the equation for the oil export variable in the SVAR model
of Iran with domestic variables ordered as: exchange rate returns, oil exports, money supply
growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆x
0
t
(1) (2) (3) (4) (5) (6)
st 0.119 0.058 0.056 0.017 0.019 0.016
(0.140) (0.126) (0.128) (0.128) (0.126) (0.129)
st−1 −0.320∗∗ −0.250∗ −0.248 −0.223 −0.208 −0.205
(0.161) (0.151) (0.153) (0.140) (0.142) (0.142)
∆ef,t 0.082 0.158 0.157 0.230 0.224 0.232
(0.207) (0.205) (0.204) (0.190) (0.177) (0.179)
∆ef,t−1 −0.470∗ −0.495∗ −0.496∗ −0.458∗ −0.505∗∗ −0.507∗∗
(0.251) (0.261) (0.262) (0.252) (0.234) (0.234)
∆x
0
t−1 −0.038 −0.058 −0.057 −0.087 −0.098 −0.098
(0.101) (0.106) (0.108) (0.106) (0.105) (0.105)
∆mt−1 −0.125 −0.735 −0.760 −1.250∗ −1.175 −1.156
(0.845) (0.772) (0.797) (0.731) (0.720) (0.734)
∆pt−1 −0.133 0.106 0.112 −0.144 0.026 0.049
(0.716) (0.734) (0.746) (0.755) (0.770) (0.763)
∆yt−1 0.097 0.144 0.139 −0.032 −0.074 −0.074
(0.602) (0.616) (0.623) (0.560) (0.547) (0.549)
∆ywt 8.786∗∗ 8.551∗∗ 4.698 3.603 3.725
(3.792) (3.764) (3.061) (3.142) (3.111)
∆reqwt 0.047 −0.061 −0.168 −0.128
(0.270) (0.231) (0.226) (0.296)
∆ewt −3.597∗∗∗ −3.638∗∗∗ −3.659∗∗∗
(0.951) (0.925) (0.956)
∆rwt 63.168∗∗ 63.754∗∗
(25.603) (25.886)
grvt 0.130
(0.597)
Residual serial 1.196 2.379 2.414 1.926 4.488 4.558
correlation test [0.879] [0.666] [0.660] [0.749] [0.344] [0.336]
Adjusted R2 0.082 0.119 0.111 0.212 0.249 0.242
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
220

Table A.62: Quarterly estimates of the equation for money supply growth in the SVAR model
of Iran with domestic variables ordered as: exchange rate returns, oil exports, money supply
growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆mt
(1) (2) (3) (4) (5) (6)
st −0.0004 −0.002 −0.003 −0.002 −0.002 0.002
(0.014) (0.015) (0.014) (0.014) (0.014) (0.013)
st−1 0.014 0.015 0.016 0.017 0.017 0.014
(0.017) (0.018) (0.018) (0.017) (0.017) (0.016)
∆ef,t −0.009 −0.007 −0.008 −0.011 −0.012 −0.022
(0.017) (0.018) (0.018) (0.018) (0.017) (0.018)
∆x
0
t 0.008 0.006 0.006 0.011 0.014∗ 0.015∗
(0.007) (0.007) (0.007) (0.007) (0.008) (0.008)
∆ef,t−1 −0.024 −0.025 −0.026 −0.027 −0.023 −0.021
(0.022) (0.022) (0.022) (0.022) (0.021) (0.021)
∆x
0
t−1 −0.005 −0.005 −0.005 −0.003 −0.002 −0.003
(0.007) (0.007) (0.008) (0.008) (0.008) (0.008)
∆mt−1 0.235∗ 0.218 0.211 0.236 0.236 0.211
(0.139) (0.146) (0.146) (0.156) (0.156) (0.149)
∆pt−1 0.165∗ 0.167∗ 0.174∗ 0.193∗∗ 0.189∗∗ 0.163
(0.089) (0.088) (0.090) (0.094) (0.093) (0.100)
∆yt−1 0.022 0.025 0.022 0.026 0.028 0.026
(0.051) (0.051) (0.051) (0.050) (0.051) (0.050)
∆pt−2 −0.076 −0.070 −0.077 −0.088 −0.094 −0.102
(0.079) (0.078) (0.079) (0.082) (0.086) (0.088)
∆ywt 0.233 0.152 0.260 0.304 0.134
(0.406) (0.441) (0.464) (0.457) (0.506)
∆reqwt 0.016 0.021 0.027 −0.026
(0.026) (0.027) (0.029) (0.039)
∆ewt 0.146 0.160 0.191∗
(0.108) (0.114) (0.114)
∆rwt −3.581 −4.412
(2.704) (2.695)
grvt −0.176∗∗
(0.078)
Residual serial 7.428 7.640 7.255 6.129 5.742 4.178
correlation test [0.115] [0.106] [0.123] [0.190] [0.219] [0.382]
Adjusted R2 0.469 0.466 0.462 0.467 0.470 0.491
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
221

Table A.63: Quarterly estimates of the equation for inflation in the SVAR model of Iran with domestic variables ordered as: exchange rate returns, oil exports, money supply growth, inflation,
and output growth, estimated over the period 1989q1-2019q4
∆pt
(1) (2) (3) (4) (5) (6)
st −0.028∗∗ −0.033∗∗∗ −0.032∗∗ −0.033∗∗∗ −0.033∗∗∗ −0.032∗∗
(0.012) (0.013) (0.013) (0.013) (0.013) (0.013)
st−1 0.032∗∗∗ 0.037∗∗∗ 0.036∗∗∗ 0.036∗∗∗ 0.036∗∗∗ 0.035∗∗∗
(0.012) (0.012) (0.012) (0.013) (0.013) (0.013)
∆ef,t 0.155∗∗∗ 0.163∗∗∗ 0.164∗∗∗ 0.166∗∗∗ 0.166∗∗∗ 0.163∗∗∗
(0.029) (0.028) (0.028) (0.028) (0.028) (0.028)
∆x
0
t 0.001 −0.003 −0.003 −0.007 −0.007 −0.007
(0.008) (0.008) (0.008) (0.009) (0.009) (0.009)
∆mt −0.061 −0.073 −0.070 −0.058 −0.057 −0.070
(0.076) (0.072) (0.071) (0.072) (0.074) (0.075)
∆ef,t−1 −0.007 −0.009 −0.007 −0.007 −0.007 −0.007
(0.029) (0.027) (0.027) (0.027) (0.027) (0.027)
∆x
0
t−1 −0.001 −0.003 −0.003 −0.004 −0.004 −0.005
(0.008) (0.008) (0.008) (0.009) (0.009) (0.009)
∆mt−1 0.035 −0.025 −0.016 −0.036 −0.036 −0.040
(0.074) (0.072) (0.074) (0.077) (0.078) (0.079)
∆pt−1 0.480∗∗∗ 0.488∗∗∗ 0.478∗∗∗ 0.462∗∗∗ 0.462∗∗∗ 0.458∗∗∗
(0.113) (0.108) (0.108) (0.107) (0.108) (0.110)
∆yt−1 0.033 0.042 0.045 0.042 0.042 0.042
(0.041) (0.039) (0.040) (0.039) (0.039) (0.039)
∆pt−2 0.162∗∗ 0.183∗∗ 0.192∗∗ 0.201∗∗ 0.202∗∗ 0.198∗∗
(0.080) (0.077) (0.078) (0.079) (0.080) (0.080)
∆ywt 0.865∗∗∗ 0.971∗∗∗ 0.893∗∗∗ 0.889∗∗∗ 0.847∗∗∗
(0.285) (0.315) (0.330) (0.335) (0.325)
∆reqwt −0.021 −0.024 −0.025 −0.039
(0.021) (0.022) (0.023) (0.029)
∆ewt −0.102 −0.104 −0.093
(0.088) (0.091) (0.091)
∆rwt 0.303 0.031
(2.528) (2.514)
grvt −0.048
(0.056)
Residual serial 9.241 8.061 5.714 6.473 6.510 6.759
correlation test [0.055] [0.089] [0.222] [0.166] [0.164] [0.149]
Adjusted R2 0.635 0.659 0.658 0.660 0.656 0.655
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
222

Table A.64: Quarterly estimates of the equation for output growth in the SVAR model of Iran
with domestic variables ordered as: exchange rate returns, oil exports, money supply growth,
inflation, and output growth, estimated over the period 1989q1-2019q4
∆yt
(1) (2) (3) (4) (5) (6)
st 0.024 0.029 0.028 0.026 0.026 0.027
(0.023) (0.024) (0.024) (0.023) (0.023) (0.023)
st−1 −0.051∗ −0.056∗∗ −0.055∗∗ −0.055∗∗ −0.055∗∗ −0.056∗∗
(0.027) (0.028) (0.028) (0.027) (0.027) (0.027)
∆ef,t −0.130∗∗∗ −0.141∗∗∗ −0.142∗∗∗ −0.135∗∗∗ −0.136∗∗∗ −0.138∗∗∗
(0.041) (0.042) (0.043) (0.041) (0.041) (0.041)
∆x
0
t 0.023 0.025 0.025 0.020 0.025 0.026
(0.017) (0.016) (0.017) (0.019) (0.018) (0.018)
∆mt 0.052 0.063 0.062 0.078 0.054 0.037
(0.118) (0.118) (0.118) (0.115) (0.120) (0.123)
∆pt 0.348∗∗ 0.387∗∗ 0.390∗∗ 0.373∗∗ 0.373∗∗ 0.364∗∗
(0.165) (0.173) (0.174) (0.173) (0.174) (0.175)
∆ef,t−1 0.037 0.041 0.041 0.040 0.046 0.047
(0.044) (0.044) (0.044) (0.045) (0.042) (0.043)
∆x
0
t−1
0.022∗ 0.023∗ 0.024∗ 0.022∗ 0.023∗ 0.023∗
(0.011) (0.012) (0.013) (0.012) (0.013) (0.013)
∆mt−1 0.013 0.046 0.039 0.009 0.014 0.009
(0.126) (0.129) (0.129) (0.132) (0.137) (0.140)
∆pt−1 −0.466∗∗ −0.505∗∗∗ −0.505∗∗∗ −0.506∗∗ −0.519∗∗∗ −0.523∗∗∗
(0.184) (0.195) (0.196) (0.198) (0.194) (0.192)
∆yt−1 −0.218∗∗ −0.221∗∗ −0.223∗∗ −0.230∗∗ −0.225∗∗ −0.224∗∗
(0.108) (0.107) (0.108) (0.105) (0.106) (0.108)
∆ywt −0.520 −0.595 −0.708 −0.619 −0.664
(0.458) (0.510) (0.526) (0.527) (0.522)
∆reqwt 0.015 0.010 0.021 0.003
(0.042) (0.044) (0.046) (0.066)
∆ewt −0.160 −0.133 −0.121
(0.175) (0.174) (0.174)
∆rwt −6.160 −6.499
(4.596) (4.779)
grvt −0.060
(0.129)
Residual serial 7.242 7.240 7.371 7.721 8.049 8.248
correlation test [0.124] [0.124] [0.118] [0.102] [0.090] [0.083]
Adjusted R2 0.126 0.124 0.117 0.117 0.126 0.120
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
223

Table A.65: Quarterly estimates of the SVAR model of Iran with domestic variables ordered
as: Exchange rate returns, oil exports, inflation, and output growth, estimated over the period
1989q1-2019q4
∆ef,t ∆x
0
t ∆pt ∆yt
(1) (2) (3) (4)
st 0.311∗∗∗ 0.079 −0.033∗∗∗ 0.028
(0.078) (0.120) (0.012) (0.024)
st−1 −0.241∗∗∗ −0.280∗∗ 0.036∗∗∗ −0.054∗
(0.069) (0.139) (0.012) (0.027)
∆ef,t 0.102 0.162∗∗∗ −0.138∗∗∗
(0.185) (0.027) (0.040)
∆x
0
t −0.004 0.028∗
(0.007) (0.015)
∆pt 0.364∗∗
(0.168)
∆ywt −2.316 7.910∗∗ 0.808∗∗∗ −0.459
(1.450) (3.858) (0.293) (0.418)
∆ef,t−1 0.326∗ −0.394 −0.006 0.041
(0.171) (0.278) (0.027) (0.043)
∆x
0
t−1
0.040 −0.067 −0.002 0.021∗
(0.035) (0.103) (0.008) (0.012)
∆pt−1 −0.339 −0.025 0.477∗∗∗ −0.482∗∗
(0.603) (0.790) (0.102) (0.191)
∆yt−1 −0.133 0.138 0.040 −0.223∗∗
(0.222) (0.613) (0.038) (0.106)
∆pt−2 0.184∗∗∗
(0.070)
Residual serial 4.832 3.895 8.003 6.738
correlation test [0.305] [0.420] [0.091] [0.150]
Adjusted R2 0.219 0.091 0.668 0.152
Notes: The variables are ordered as: ∆eft, ∆x
0
t
, ∆pt, and ∆yt, where: ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly
rial/U.S. dollar free market exchange rate; ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the oil exports revenues in U.S. dollars;
∆pt = ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran; ∆yt = ln(Yt/Yt−1), Yt is the quarterly real output of
Iran. st is the quarterly sanctions intensity variable. Seasonal dummies are not included to allow for possible seasonality
of the variables in the SVAR model. Numbers in parentheses are heteroskedastic-consistent standard errors obtained
following the approach of White (1980), and those in square brackets are p-values. ***p < 0.01, **p < 0.05, *p < 0.1.
"Residual serial correlation test" is the Breusch–Godfrey LM test of serially uncorrelated errors with lag order of the test
set to 4.
See the notes to Table A.9 for further details on the sources and construction of data used.
224

Table A.66: Quarterly estimates of the equation for exchange rate returns in the SVAR model of
Iran with domestic variables ordered as: exchange rate returns, oil exports, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆ef t
(1) (2) (3) (4) (5) (6)
st 0.303∗∗∗ 0.311∗∗∗ 0.307∗∗∗ 0.307∗∗∗ 0.307∗∗∗ 0.314∗∗∗
(0.079) (0.078) (0.078) (0.077) (0.077) (0.075)
st−1 −0.231∗∗∗ −0.241∗∗∗ −0.238∗∗∗ −0.236∗∗∗ −0.235∗∗∗ −0.243∗∗∗
(0.070) (0.069) (0.069) (0.069) (0.070) (0.069)
∆ef,t−1 0.329∗ 0.326∗ 0.323∗ 0.317∗ 0.314∗ 0.312∗
(0.176) (0.171) (0.170) (0.165) (0.168) (0.160)
∆x
0
t−1
0.035 0.040 0.043 0.047 0.046 0.041
(0.034) (0.035) (0.036) (0.038) (0.038) (0.035)
∆pt−1 −0.284 −0.339 −0.330 −0.298 −0.288 −0.388
(0.604) (0.603) (0.609) (0.601) (0.591) (0.582)
∆yt−1 −0.121 −0.133 −0.140 −0.120 −0.122 −0.118
(0.224) (0.222) (0.225) (0.221) (0.219) (0.214)
∆ywt −2.316 −2.724∗ −2.217 −2.265 −2.837∗
(1.450) (1.617) (1.626) (1.652) (1.551)
∆reqwt 0.076 0.088 0.082 −0.110
(0.085) (0.086) (0.091) (0.121)
∆ewt 0.397 0.395 0.507
(0.496) (0.495) (0.513)
∆rwt 3.363 0.477
(14.007) (14.068)
grvt −0.619∗∗
(0.308)
Residual serial 5.198 4.832 4.686 5.847 5.717 4.179
correlation test [0.268] [0.305] [0.321] [0.211] [0.221] [0.382]
Adjusted R2 0.209 0.219 0.215 0.215 0.208 0.229
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
225

Table A.67: Quarterly estimates of the equation for the oil exports variable in the SVAR model of
Iran with domestic variables ordered as: exchange rate returns, oil exports, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆x
0
t
(1) (2) (3) (4) (5) (6)
st 0.126 0.079 0.078 0.053 0.055 0.051
(0.136) (0.120) (0.120) (0.120) (0.118) (0.123)
st−1 −0.330∗∗ −0.280∗∗ −0.280∗∗ −0.274∗∗ −0.257∗ −0.253∗
(0.151) (0.139) (0.141) (0.131) (0.131) (0.133)
∆ef,t 0.032 0.102 0.101 0.175 0.160 0.168
(0.188) (0.185) (0.183) (0.179) (0.166) (0.172)
∆ef,t−1 −0.381 −0.394 −0.395 −0.360 −0.414 −0.416
(0.265) (0.278) (0.278) (0.273) (0.257) (0.257)
∆x
0
t−1 −0.044 −0.067 −0.066 −0.102 −0.111 −0.110
(0.096) (0.103) (0.105) (0.106) (0.108) (0.108)
∆pt−1 −0.234 −0.025 −0.021 −0.275 −0.077 −0.051
(0.780) (0.790) (0.798) (0.822) (0.835) (0.834)
∆yt−1 0.087 0.138 0.135 −0.035 −0.088 −0.088
(0.597) (0.613) (0.620) (0.553) (0.547) (0.549)
∆ywt 7.910∗∗ 7.738∗∗ 3.419 2.430 2.585
(3.858) (3.669) (2.931) (3.009) (2.982)
∆reqwt 0.031 −0.085 −0.209 −0.164
(0.283) (0.234) (0.233) (0.307)
∆ewt −3.538∗∗∗ −3.577∗∗∗ −3.607∗∗∗
(0.995) (0.964) (1.000)
∆rwt 66.363∗∗∗ 67.024∗∗∗
(24.414) (24.980)
grvt 0.148
(0.618)
Residual serial 2.461 3.895 4.005 2.396 4.791 5.085
correlation test [0.652] [0.420] [0.405] [0.663] [0.309] [0.279]
Adjusted R2 0.061 0.091 0.083 0.180 0.222 0.215
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
226

Table A.68: Quarterly estimates of the equation for inflation in the SVAR model of Iran with
domestic variables ordered as: exchange rate returns, oil exports, inflation, and output growth,
estimated over the period 1989q1-2019q4
∆pt
(1) (2) (3) (4) (5) (6)
st −0.029∗∗ −0.033∗∗∗ −0.032∗∗ −0.032∗∗∗ −0.032∗∗∗ −0.031∗∗
(0.013) (0.012) (0.013) (0.012) (0.012) (0.013)
st−1 0.032∗∗∗ 0.036∗∗∗ 0.035∗∗∗ 0.034∗∗∗ 0.034∗∗∗ 0.033∗∗∗
(0.012) (0.012) (0.012) (0.012) (0.012) (0.012)
∆ef,t 0.155∗∗∗ 0.162∗∗∗ 0.163∗∗∗ 0.166∗∗∗ 0.166∗∗∗ 0.164∗∗∗
(0.028) (0.027) (0.027) (0.027) (0.027) (0.027)
∆x
0
t 0.0005 −0.004 −0.003 −0.007 −0.007 −0.007
(0.007) (0.007) (0.007) (0.008) (0.008) (0.008)
∆ef,t−1 −0.005 −0.006 −0.004 −0.004 −0.005 −0.004
(0.029) (0.027) (0.026) (0.026) (0.026) (0.027)
∆x
0
t−1
0.001 −0.002 −0.002 −0.004 −0.004 −0.004
(0.009) (0.008) (0.008) (0.009) (0.009) (0.009)
∆pt−1 0.466∗∗∗ 0.477∗∗∗ 0.465∗∗∗ 0.454∗∗∗ 0.454∗∗∗ 0.450∗∗∗
(0.107) (0.102) (0.101) (0.100) (0.101) (0.103)
∆yt−1 0.033 0.040 0.044 0.041 0.040 0.040
(0.039) (0.038) (0.039) (0.038) (0.039) (0.039)
∆pt−2 0.172∗∗ 0.184∗∗∗ 0.196∗∗∗ 0.201∗∗∗ 0.201∗∗∗ 0.199∗∗∗
(0.073) (0.070) (0.072) (0.073) (0.073) (0.073)
∆ywt 0.808∗∗∗ 0.952∗∗∗ 0.854∗∗ 0.850∗∗ 0.815∗∗
(0.293) (0.322) (0.338) (0.340) (0.337)
∆reqwt −0.026 −0.029 −0.030 −0.040
(0.020) (0.021) (0.022) (0.027)
∆ewt −0.103 −0.105 −0.098
(0.080) (0.082) (0.083)
∆rwt 0.317 0.167
(2.317) (2.313)
grvt −0.032
(0.054)
Residual serial 11.521 8.003 5.321 5.888 5.868 5.924
correlation test [0.021] [0.091] [0.256] [0.208] [0.209] [0.205]
Adjusted R2 0.647 0.668 0.669 0.671 0.668 0.666
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
227

Table A.69: Quarterly estimates of the equation for output growth in the SVAR model of Iran
with domestic variables ordered as: exchange rate returns, oil exports, inflation, and output
growth, estimated over the period 1989q1-2019q4
∆yt
(1) (2) (3) (4) (5) (6)
st 0.024 0.028 0.027 0.026 0.026 0.027
(0.023) (0.024) (0.024) (0.023) (0.023) (0.023)
st−1 −0.050∗ −0.054∗ −0.053∗∗ −0.054∗∗ −0.054∗∗ −0.055∗∗
(0.027) (0.027) (0.027) (0.027) (0.026) (0.027)
∆ef,t −0.128∗∗∗ −0.138∗∗∗ −0.140∗∗∗ −0.134∗∗∗ −0.134∗∗∗ −0.137∗∗∗
(0.039) (0.040) (0.040) (0.039) (0.039) (0.039)
∆x
0
t 0.025∗ 0.028∗ 0.028∗ 0.023 0.028∗ 0.028∗
(0.015) (0.015) (0.016) (0.017) (0.016) (0.016)
∆pt 0.328∗∗ 0.364∗∗ 0.371∗∗ 0.353∗∗ 0.355∗∗ 0.348∗∗
(0.161) (0.168) (0.170) (0.169) (0.167) (0.166)
∆ef,t−1 0.038 0.041 0.041 0.040 0.047 0.047
(0.043) (0.043) (0.043) (0.043) (0.041) (0.041)
∆x
0
t−1
0.019∗ 0.021∗ 0.022∗ 0.020 0.021∗ 0.021
(0.011) (0.012) (0.012) (0.012) (0.013) (0.013)
∆pt−1 −0.449∗∗ −0.482∗∗ −0.484∗∗ −0.483∗∗ −0.500∗∗∗ −0.508∗∗∗
(0.181) (0.191) (0.191) (0.193) (0.188) (0.185)
∆yt−1 −0.219∗∗ −0.223∗∗ −0.225∗∗ −0.231∗∗ −0.226∗∗ −0.226∗∗
(0.108) (0.106) (0.107) (0.104) (0.106) (0.108)
∆ywt −0.459 −0.595 −0.715 −0.645 −0.710
(0.418) (0.478) (0.498) (0.501) (0.504)
∆reqwt 0.024 0.019 0.031 0.011
(0.041) (0.044) (0.046) (0.064)
∆ewt −0.139 −0.117 −0.103
(0.159) (0.159) (0.159)
∆rwt −5.988 −6.293
(4.408) (4.539)
grvt −0.066
(0.117)
Residual serial 7.054 6.738 6.977 7.483 7.876 8.205
correlation test [0.133] [0.150] [0.137] [0.112] [0.096] [0.084]
Adjusted R2 0.154 0.152 0.146 0.145 0.154 0.149
Notes: Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of
White (1980). See the notes to Table A.9 for further details on the sources and construction of data used.
228

Table A.70: Estimates of the reduced form Iran’s output growth equation using a sanctions
dummy variable estimated over the period 1989q1- 2019q4
∆yt
(1) (2) (3) (4) (5) (6) (7)
dt−1(βdt−1
) −0.014∗∗ −0.014∗∗ −0.014∗∗ −0.014∗∗ −0.013∗ −0.013∗ −0.013∗
(0.007) (0.007) (0.007) (0.007) (0.007) (0.007) (0.007)
∆yt−1(λ∆yt−1
) −0.207∗ −0.206∗ −0.208∗ −0.205∗ −0.214∗ −0.215∗ −0.218∗
(0.115) (0.116) (0.117) (0.118) (0.111) (0.113) (0.114)
∆x
0
t−1
0.017 0.016 0.017 0.018 0.016 0.015 0.016
(0.012) (0.012) (0.012) (0.012) (0.012) (0.012) (0.013)
∆ef,t−1 −0.013 −0.013 −0.013 −0.010 −0.007 −0.008 −0.009
(0.041) (0.042) (0.041) (0.040) (0.040) (0.040) (0.041)
∆mt−1 −0.042 −0.048 −0.053 −0.046 −0.066 −0.071 −0.078
(0.099) (0.104) (0.105) (0.108) (0.106) (0.107) (0.113)
∆pt−1 −0.214 −0.212 −0.209 −0.222 −0.245 −0.252∗ −0.253∗
(0.150) (0.152) (0.152) (0.148) (0.153) (0.151) (0.151)
∆ywt 0.156 0.058 0.107 −0.169 −0.230 −0.186
(0.361) (0.412) (0.416) (0.461) (0.463) (0.480)
∆reqwt 0.019 0.026 0.019 −0.0002 −0.002
(0.041) (0.042) (0.045) (0.063) (0.064)
∆rwt −4.060 −3.838 −4.120 −3.154
(4.476) (4.405) (4.525) (4.976)
∆ewt −0.240 −0.228 −0.265
(0.159) (0.161) (0.178)
grvt −0.063 −0.069
(0.114) (0.117)
∆p
0
t −0.012
(0.026)
βdt−1
/(1 − λ∆yt−1
) −0.012∗∗ −0.011∗∗ −0.011∗∗ −0.012∗∗ −0.010∗ −0.011∗ −0.011∗
(0.006) (0.006) (0.006) (0.006) (0.005) (0.006) (0.006)
Adjusted R2 0.078 0.070 0.064 0.063 0.076 0.070 0.064
Notes: ∆yt = ln(Yt/Yt−1), Yt is the quarterly real output of Iran. dt is the sanctions dummy variable. βdt−1
and λ∆yt−1
are the coefficients of dt−1 and ∆yt−1, respectively; βdt−1
/(1− λ∆yt−1
) represents the long run effect of sanctions on
output growth. See Chapter 6 of Pesaran (2015). Numbers in parentheses are heteroskedastic-consistent standard errors
obtained following the approach of White (1980).
See the notes to Table A.9 for further details on the sources and construction of data used. Details on the construction
of dt are provided in Section A.2.4.
229

Table A.71: Quarterly estimates of the reduced form Iran’s output growth equation using a discretized sanctions intensity variable estimated over the period 1989q1- 2019q4
∆yt
(1) (2) (3) (4) (5) (6) (7)
s
D
t−1
(βsD
t−1
) −0.010∗ −0.010∗ −0.010∗ −0.010∗ −0.009 −0.009 −0.009
(0.006) (0.006) (0.006) (0.006) (0.006) (0.006) (0.006)
∆yt−1(λ∆yt−1
) −0.202∗ −0.201∗ −0.202∗ −0.199∗ −0.209∗ −0.209∗ −0.214∗
(0.115) (0.115) (0.117) (0.118) (0.111) (0.112) (0.113)
∆x
0
t−1
0.015 0.015 0.016 0.016 0.014 0.014 0.015
(0.012) (0.012) (0.012) (0.012) (0.012) (0.012) (0.013)
∆ef,t−1 −0.016 −0.015 −0.016 −0.013 −0.010 −0.010 −0.012
(0.043) (0.043) (0.043) (0.041) (0.042) (0.042) (0.042)
∆mt−1 −0.032 −0.039 −0.044 −0.037 −0.058 −0.061 −0.069
(0.098) (0.103) (0.104) (0.108) (0.105) (0.107) (0.113)
∆pt−1 −0.226 −0.222 −0.220 −0.233 −0.255∗ −0.261∗ −0.261∗
(0.151) (0.153) (0.153) (0.149) (0.154) (0.152) (0.152)
∆ywt 0.182 0.095 0.144 −0.138 −0.169 −0.119
(0.377) (0.425) (0.427) (0.470) (0.475) (0.493)
∆reqwt 0.017 0.024 0.017 0.006 0.004
(0.041) (0.042) (0.045) (0.064) (0.064)
∆rwt −3.935 −3.722 −3.875 −2.730
(4.457) (4.381) (4.500) (4.934)
∆ewt −0.243 −0.237 −0.279
(0.159) (0.160) (0.178)
grvt −0.036 −0.044
(0.113) (0.115)
∆p
0
t −0.014
(0.026)
βsD
t−1
/(1 − λ∆yt−1
) −0.008∗ −0.008∗ −0.008∗ −0.008∗ −0.007 −0.007 −0.008
(0.005) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005)
Adjusted R2 0.074 0.067 0.060 0.059 0.072 0.064 0.059
Notes: ∆yt = ln(Yt/Yt−1), Yt is the quarterly real output of Iran. s
D
t
is the discretized sanctions intensity variable.
βsD
t−1
and λ∆yt−1
are the coefficients of s
D
t−1 and ∆yt−1, respectively; βsD
t−1
/(1− λ∆yt−1
) represents the long run effect
of sanctions on output growth. See Chapter 6 of Pesaran (2015). Numbers in parentheses are heteroskedastic-consistent
standard errors obtained following the approach of White (1980).
See the notes to Table A.9 for further details on the sources and construction of data used. Details on the construction
of s
D
t are provided in Section A.2.4.
230

Table A.72: Quarterly estimates of the SVAR model of Iran using a sanctions dummy variable and
with domestic variables ordered as: oil exports, exchange rate returns, money supply growth,
inflation, and output growth, estimated over the period 1989q1-2019q4
∆x
0
t ∆ef,t ∆mt ∆pt ∆yt
(1) (2) (3) (4) (5)
dt 0.290∗∗∗ 0.079 0.004 0.001 −0.014
(0.096) (0.078) (0.007) (0.018) (0.016)
dt−1 −0.357∗∗∗ −0.040 −0.002 0.00002 0.003
(0.095) (0.078) (0.007) (0.017) (0.017)
∆x
0
t 0.037 0.004 −0.005 0.030∗
(0.046) (0.008) (0.008) (0.017)
∆ef,t −0.010 0.144∗∗∗ −0.104∗∗∗
(0.016) (0.029) (0.040)
∆mt −0.061 0.033
(0.073) (0.126)
∆pt 0.298∗
(0.174)
∆ywt 7.813∗∗ −1.600 0.229 0.758∗∗ −0.390
(3.961) (1.296) (0.397) (0.310) (0.426)
∆x
0
t−1 −0.043 0.043 −0.006 −0.002 0.022∗
(0.104) (0.037) (0.007) (0.008) (0.012)
∆ef,t−1 −0.556∗∗ 0.298 −0.019 0.007 0.023
(0.220) (0.185) (0.022) (0.026) (0.046)
∆mt−1 −0.542 −0.019 0.232 −0.002 0.006
(0.732) (0.294) (0.148) (0.073) (0.124)
∆pt−1 0.248 −0.302 0.159∗ 0.441∗∗∗ −0.418∗∗
(0.691) (0.701) (0.087) (0.109) (0.206)
∆yt−1 0.193 −0.144 0.024 0.049 −0.227∗∗
(0.622) (0.249) (0.051) (0.040) (0.107)
∆pt−2 −0.060 0.221∗∗∗
(0.071) (0.073)
Residual serial 1.889 5.847 7.734 5.096 6.274
correlation test [0.756] [0.211] [0.102] [0.278] [0.180]
Adjusted R2 0.160 0.083 0.460 0.633 0.101
Notes: The variables are ordered as: ∆x
0
t
, ∆eft, ∆mt, ∆pt, and ∆yt, where: ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the oil
exports revenues in U.S. dollars; ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly rial/U.S. dollar free market exchange rate;
∆mt = (M2t−M2,t−1)/M2,t−1, M2t is the monetary aggregate M2 obtained by summing the aggregates M1 and "quasimoney"; ∆pt = ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran; ∆yt = ln(Yt/Yt−1), Yt is the quarterly real
output of Iran. dt is the sanctions dummy variable. Seasonal dummies are included to allow for possible seasonality
of the variables in the regressions of the SVAR model in Equation (1.6) with qt =

∆x
0
t
, ∆eft, ∆mt, ∆pt, ∆yt
0
and
zwt = (∆ywt)
0
. Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach
of White (1980), and those in square brackets are p-values. ***p < 0.01, **p < 0.05, *p < 0.1. "Residual serial correlation
test" is the Breusch–Godfrey LM test of serially uncorrelated errors with lag order of the test set to 4.
See the notes to Table A.9 for further details on the sources and construction of data used. Details on the construction
of dt are provided in Section A.2.4.
23

Table A.73: Quarterly estimates of the SVAR model of Iran using a discretized sanctions intensity variable and with domestic variables ordered as: oil exports, exchange rate returns, money
supply growth, inflation, and output growth, estimated over the period 1989q1-2019q4
∆x
0
t ∆ef,t ∆mt ∆pt ∆yt
(1) (2) (3) (4) (5)
s
D
t 0.124∗∗ 0.073 0.006 −0.011 0.013
(0.059) (0.066) (0.005) (0.011) (0.010)
s
D
t−1 −0.200∗∗∗ −0.060 −0.002 0.014 −0.021∗∗
(0.060) (0.067) (0.005) (0.012) (0.010)
∆x
0
t 0.034 0.005 −0.003 0.025
(0.044) (0.007) (0.008) (0.017)
∆ef,t −0.011 0.148∗∗∗ −0.121∗∗∗
(0.015) (0.027) (0.041)
∆mt −0.059 0.034
(0.074) (0.127)
∆pt 0.338∗
(0.175)
∆ywt 7.778∗∗ −1.903 0.238 0.793∗∗ −0.443
(3.887) (1.459) (0.389) (0.311) (0.440)
∆x
0
t−1 −0.075 0.035 −0.005 −0.0003 0.021∗
(0.108) (0.035) (0.008) (0.008) (0.012)
∆ef,t−1 −0.536∗∗∗ 0.311∗ −0.019 0.008 0.019
(0.190) (0.189) (0.022) (0.025) (0.046)
∆mt−1 −0.565 −0.015 0.225 −0.020 0.045
(0.751) (0.302) (0.146) (0.073) (0.127)
∆pt−1 0.109 −0.302 0.151∗ 0.453∗∗∗ −0.461∗∗
(0.636) (0.713) (0.088) (0.105) (0.200)
∆yt−1 0.120 −0.187 0.027 0.046 −0.216∗∗
(0.607) (0.245) (0.051) (0.040) (0.106)
∆pt−2 −0.062 0.213∗∗∗
(0.073) (0.075)
Residual serial 2.613 2.834 7.270 6.829 5.194
correlation test [0.625] [0.586] [0.122] [0.145] [0.268]
Adjusted R2 0.143 0.072 0.463 0.644 0.107
Notes: The variables are ordered as: ∆x
0
t
, ∆eft, ∆mt, ∆pt, and ∆yt, where: ∆x
0
t = (X
0
t − X
0
t−1)/X0
t−1, X0
t
is the
oil exports revenues in U.S. dollars; ∆eft = ln(Eft/Ef,t−1), Eft is the quarterly rial/U.S. dollar free market exchange
rate; ∆mt = (M2t − M2,t−1)/M2,t−1, M2t is the monetary aggregate M2 obtained by summing the aggregates M1
and "quasi-money"; ∆pt = ln(Pt/Pt−1), Pt is the quarterly consumer price index of Iran; ∆yt = ln(Yt/Yt−1), Yt
is the quarterly real output of Iran. s
D
t
is the discretized sanctions intensity variable. Seasonal dummies are included to allow for possible seasonality of the variables in the regressions of the SVAR model in Equation (1.6) with
qt =

∆x
0
t
, ∆eft, ∆mt, ∆pt, ∆yt
0
and zwt = (∆ywt)
0
. Numbers in parentheses are heteroskedastic-consistent standard errors obtained following the approach of White (1980), and those in square brackets are p-values. ***p < 0.01,
**p < 0.05, *p < 0.1. "Residual serial correlation test" is the Breusch–Godfrey LM test of serially uncorrelated errors
with lag order of the test set to 4. See the notes to Table A.9 for further details on the sources and construction of data
used. Details on the construction of s
D
t are provided in Section A.2.4.
23

A.5 Sanctions chronology
233

Table A.74: Chronology of major sanctions events against Iran over the period from November 1979 to January 2021
Date Event Diplomatic measures Direction Sanctioning
entity
Additional notes
Nov. 12, 1979 "Tehran hostage crisis" Oil embargo On U.S.A. Proclamation
4702(1979) by U.S.
President Carter
Nov. 14, 1979 "Tehran hostage crisis" Asset freeze of all Iranian government and Central
Bank of Iran properties within U.S. jurisdiction
On U.S.A. U.S. Executive Order
12170
Apr. 7, 1980 "Tehran hostage crisis" Sale and transportation of all goods to Iran
forbidden. No credit and loans to Iran allowed
On U.S.A. U.S. Executive Order
12205
Apr. 17, 1980 "Tehran hostage crisis" Ban on all imports from Iran. U.S. citizens prevented
from traveling to Iran or conducting financial
transactions in Iran
On U.S.A. U.S. Executive Order
12211
Jan. 19, 1981 Hostages release Revocation of prohibitions against transactions
involving Iran
Off U.S.A. U.S. Executive Order
12282
Jan. 19, 1984 1983 U.S. embassy
bombing in Beirut
Ineligibility for various forms of U.S. foreign
assistance. Arms embargo. Imposition of
miscellaneous financial restrictions
On U.S.A. "State Sponsor of
Terror" designation
Oct. 29, 1987 Support of
international terrorism
No goods or services of Iranian origin allowed to be
imported into the U.S.. Iranian oil refined in third
countries allowed
On U.S.A. U.S. Executive Order
12613
Oct. 23, 1992 Iran-Iraq Arms
Non-Proliferation Act
Measures to prevent transfer of goods or technology
to Iraq or Iran to avoid acquisition of Weapons of
Mass Destruction (WMD)
On U.S.A. U.S. Public Law 102-484
Nov. 14, 1994 Nuclear threat Controls and restrictions on goods related to WMD
technology
On U.S.A. U.S. Executive Order
12938
Mar. 15, 1995 Threat to national
security
Ban U.S. investment in Iran’s energy sector On U.S.A. U.S. Executive Order
12957
May 6, 1995 Threat to national
security
More comprehensive investment, trade, and
financial restrictions
On U.S.A. U.S. Executive Order
12959
Aug. 5, 1996 Iran and Libya
Sanctions Act (ILSA)
Two economic and/or financial sanctions (out of a
list of 6) on U.S. and non-U.S. companies providing
investments over $40 million in petroleum resources
in Iran
On U.S.A. U.S. Public Law 104 -
172
Nov. 22, 1996 "Blocking regulation"
following the U.S. ILSA
E.U. single member states encouraged to impose
sanctions in compliance with the ILSA of 1996
On E.U. Council Regulation
2271/96
Aug. 19, 1997 Nuclear threat Trade sanctions previously in place largely
expanded in scope
On U.S.A. U.S. Executive Order
13059
234

Date Event Diplomatic measures Direction Sanctioning
entity
Additional notes
Mar. 14, 2000 Iran Nonproliferation
Act
The President authorized to act against individuals
or organizations known to provide material aid to
WMD
On U.S.A. U.S. Public Law 106
-178
Aug. 3, 2001 ILSA Extension Act of
2001
ILSA Renewed for 5 years. Extends the scope of
previous sanctions. Max investments allowed from
40to20 millions
On U.S.A. U.S. Public Law 107-24
Sep. 23, 2001 Twin Towers attacks in
New York City
Blocking property and prohibiting transactions with
persons committing or supporting terrorism.
Iranians marginally involved
On U.S.A. U.S. Executive Order
13224
June 29, 2005 Nuclear threat Assets freeze of individual connected to Iran WMD
proliferation and their supporters
On U.S.A. U.S. Executive Order
13382
July 31, 2006 Nuclear threat Prohibits the transfer of any materials that could
contribute to Iran’s nuclear and ballistic missile
programmes
On U.N. United Nations
Security Council
Resolution 1696
Aug. 4, 2006 ILSA Extension The U.S. further extend ILSA until Sep. 29, 2006 On U.S.A. U.S. Public Law 109-267
Sep. 30, 2006 Iran Freedom and
Support Act (IFSA)
Secondary sanctions imposed. Oil-related
investments banned. Support to "pro-democracy"
groups opposed to Iran.
On U.S.A. U.S. Public Law 109-293
Dec. 23, 2006 Nuclear threat Trade embargo on nuclear-related goods and
technologies. Ban of financial support for nuclear
projects. Assets freeze
On U.S.A. United Nations
Security Council
Resolution 1737
Feb. 27, 2007 Nuclear threat Ban on export of nuclear technology, and financial
assistance related to nuclear activities. Assets freeze
and travel restrictions
On E.U. Council Common
Position
2007/140/CFSP
Mar. 24, 2007 Nuclear threat Sanctions strengthened on individuals and arms
related to the development of WMD
On U.N. United Nations
Security Council
Resolution 1747
July 17, 2007 Iraq War. Measures to
increase isolation of
Iraq
Assets freeze of people connected to Iraq War.
Iranians marginally hit
On U.S.A. U.S. Executive Order
13438
Mar. 3, 2008 Nuclear threat Ban on WMD technology transfers, financial
restrictions, call to monitor Iranian institutions and
individuals
On U.N. United Nations
Security Council
Resolution 1803
Sep. 27, 2008 Ongoing
uranium-enrichment
programs reported by
IAEA
Re-affirm previous sanctions On U.N. United Nations
Security Council
Resolution 1835
June 9, 2010 Nuclear threat Restrictions related to ballistic programs and WMD
technologies. Prohibits new banking relations
On U.N. United Nations
Security Council
Resolution 1929
235

Date Event Diplomatic measures Direction Sanctioning
entity
Additional notes
July 1, 2010 Comprehensive Iran
Sanctions,
Accountability, and
Divestment Act
(CISADA)
Scope of previous sanctions expanded. Curb on
import/export of petroleum. Extented FX, banking,
and property transactions
On U.S.A. U.S. Public Law 111-195
July 22, 2010 Nuclear threat Extra sanctions imposed by Canada (on top of the
U.N. ones) under the "Special Economic Measures
Act" – SEMA
On CAN SOR/2010 - 165
July 26, 2010 Nuclear threat U.N. 1929(2010) resolution embedded in the E.U.
framework. Additional economic, banking and
financial restrictions imposed
On E.U. Council Decision
2010/413/CFSP
Sep. 28, 2010 Human rights
violations
Assets freeze and limits to transfers and donations On U.S.A. U.S. Executive Order
13553
Apr. 12, 2011 Human rights violation Travel restrictions and assets freeze of people related
to human rights violations
On E.U. Council Decision
2011/235/CFSP
Apr. 29, 2011 Human rights violation Assets freeze of persons and entities involved with
abuses. Donations prohibited
On U.S.A. U.S. Executive Order
13572
May 23, 2011 Nuclear threat Enhanced sanctions from Iran Sanctions Act (ISA).
No credit, no FX, property block from U.S. financial
institutions, imports ban
On U.S.A. U.S. Executive Order
13574
June 9, 2011 Nuclear threat Extended mandate of the "panel of experts" that
supports the Iran Sanctions Committee for one year.
On U.N. U.N. Security Council
Resolution 1984
Nov. 20, 2011 Threat to national
security
Sanctions on entities and individuals helping the
Iran’s energy and petrochemical sectors maintenance
and expansion
On U.S.A. U.S. Executive Order
13590
Dec. 31, 2011 National Defence
Authorization Act
Sanctions against banks dealing with Iranian
financial institutions, Bank Markazi included.
Restricted export of Iranian oil.
On U.S.A. U.S. NDAA 2012 -
Sec. 1245
Jan. 23, 2012 Nuclear threat Oil embargo, assets freeze of Central Bank of Iran
(CBI). Embargo on gold, precious metals
On E.U. Council Decision
2012/35/CFSP
Feb. 5, 2012 Anti-money laundering
malpractice
Blocking property of the Government of Iran and
Iranian financial institutions, Bank Markazi included
On U.S.A. U.S. Executive Order
13599
Mar. 15, 2012 Nuclear threat Decision 2010/413/CFSP expanded with new but
marginal financial restrictions
On E.U. Council Decision
2012/152/CFSP
236

Date Event Diplomatic measures Direction Sanctioning
entity
Additional notes
Mar. 23, 2012 Nuclear threat Wide expansion of scope of import/export
restrictions and banking and financial sanctions
On E.U. Council Regulation No.
267/2012
Apr. 22, 2012 Human right violations Assets freeze of companies providing technology for
human rights abuses. Donations prohibited to
blocked entities and persons
On U.S.A. U.S. Executive Order
13606
May 1, 2012 Measures against
sanctions evaders
Extra sanctions for entities and persons found to
evade previously issued sanctions against Iran
On U.S.A. U.S. Executive Order
13608
June 7, 2012 Nuclear threat Renewed the mandate of the Committee’s Panel of
Experts to monitor Iran for 13 months
On U.N. U.N. Security Council
Resolution 2049
July 30, 2012 Threat to national
security
Sanctions on foreign institutions involved in deals
with Iran’s energy and petrochemical sectors
products.
On U.S.A. U.S. Executive Order
13622
Aug. 10, 2012 Iran Threat Reduction
and Syria Human
Rights Act
New multilateral sanctions on entities facilitating
Iranian transactions (oil sector mostly); amends the
ISA of 1996
On U.S.A. U.S. Public Law 112-158
Oct. 9, 2012 Threat to national
security
Expansion of assets freeze and financial restrictions On U.S.A. U.S. Executive Order
13628
Oct. 15, 2012 Nuclear threat Ban on trade and financial assistance to buy natural
gas, a range of manufacturing and software products
for ballistic missiles, and ship-building
On E.U. Council Decision
2012/635/CFSP
Jan. 2, 2013 U.S. National Defense
Authorization Act
Broad range of economic and financial sanctions
expanded
On U.S.A. U.S. Public Law 112-239
June 3, 2013 Nuclear threat Financial restrictions and assets freeze on foreign
institutions doing business in rials or in automotive
industry, among others
On U.S.A. U.S. Executive Order
13645
July 20, 2015 Joint Comprehensive
Plan Of Action
(JCPOA)
Agreement to schedule suspension and lift of U.N.
sanctions
Off U.N. U.N. Security Council
Resolution 2231
Oct. 18, 2015 JCPOA E.U. intermediate steps towards application of
JCPOA
Off E.U. Council Decision
2015/1863/CFSP
Jan. 16, 2016 JCPOA Implementiation day: The E.U., U.S., and U.N.
suspend or terminate nuclear-related sanctions. A
process of recovery of Iran’s assets for about $100
billions begins (never fully implemented)
Off U.N. U.N. Security Council
Resolution 2231
implementiation
Jan. 17, 2016 SWIFT re-activation Iranian banks access to the SWIFT system. U.S.
banks remain prohibited from doing business with
Iran directly or indirectly
Off World SWIFT press release:
"Update: Iran Sanctions
Agreement" – Jan. 17th,
2016
237

Date Event Diplomatic measures Direction Sanctioning
entity
Additional notes
Dec. 1, 2016 U.S. renews ISA for 10
years
U.S. renew the sanctions going on since 1996 on Iran On U.S.A. U.S. Congress Issue
H.R. 6297, Vote n. 155
May 8, 2018 JCPOA reduction of
scope
U.S. announcement of withdrawal from JCPOA On U.S.A. See Wikipedia page on
"United States
withdrawal from the
Joint Comprehensive
Plan of Action"
Aug. 6, 2018 U.S. "maximum
pressure"
Re-impose all sanctions lifted or waived by JCPOA On U.S.A U.S. Executive Order
13846
Nov. 5, 2018 U.S. "maximum
pressure"
Largest ever single-day action targeting the Iranian
regime. More than 700 individuals, entities, aircrafts
and vessels hit
On U.S.A. U.S. Treasury
Statement 541 of Nov. 5
2018
Nov. 9, 2018 Financial system
stability and integrity
protection
SWIFT restrictions On U.S.A. www.swift.com/aboutus/legal/compliance0/swift-and-sanctions
May 8, 2019 Threat to national
security
Sanctions on iron, steel, aluminum, and copper
sectors of Iran
On U.S.A. U.S. Executive Order
13871
June 24, 2019 Support of terrorist
militias in the Middle
East
Further assets freeze, secondary sanctions on
financial institutions
On U.S.A. U.S. Executive Order
13876
Jan. 10, 2020 Support of terrorist
militias in the Middle
East
Assets freeze related to entities and individuals
trading in the manufacturing sector, among others.
Restrictions on immigrants
On U.S.A. U.S. Executive Order
13902
Sep. 21, 2020 Threat to national
security
Sanctions related to the trade and financially support
arms trade
On U.S.A. U.S. Executive Order
13949
Oct. 8, 2020 JCPOA withdrawal 18 Iranian banks hit by further sanctions On U.S.A. U.S. Treasury
Statement 1147 of Oct.
8 2020
Dec. 16, 2020 -
Jan. 5, 2021
Support of
"destabilizing
activities" in the Middle
East
Sanctions on companies supporting: metal, steel,
petroleum and petrochemical sectors
On U.S.A. U.S. Treasury
Statements 1214, and
1226
Jan. 13, 2021 Entities designated
pursuant to Executive
Order 13876,
Sanctions on two organizations controlled by the
Supreme Leader
On U.S.A. U.S. Treasury
Statement 1234 of Jan.
13, 2021
238

Appendix B
Appendix to Chapter 2
The Appendix is structured as follows. Section B.1 describes additional stylized facts
to complete the picture provided in the main body of the paper. Sections B.2 and B.3
provide model’s proofs, a full-fledged characterization of the equilibrium conditions, and
the computation strategy description. Section B.4 describes additional results. Details on
data sources and construction can be found in Section B.5.
B.1 Additional facts
Figure B.1 shows that the share of U.S. Treasuries owned domestically vs. by the rest
of the world. The latter component has surged spectacularly from the mid-1990s – when
it was below 20 per cent – until the Great Financial Crisis (GFC) of 2010, when it reached
a peak of over 40 per cent. In more recent times, this has partially reverted to fall back in
a range of about 30 per cent.
Figure B.2 shows that the increase of market-based banking with respect to the traditional banking system is strictly correlated with the increase of private safe assets production with respect to traditional banking safe assets. The correlation between the two series
is 97 per cent.
Figure B.3 plots the increase in inequality vis-à-vis the portfolio share of U.S. house239

holds invested with to institutional investors. As inequality increases, the portfolio share
directed to asset managers has similarly increased.
Figure B.4 plots the relationship between the top 5 percent income share and the credit
to GDP for 18 economies over a period from 1970-2019. The plot is presented for each data
point, and presented in a quantile fashion to reduce the visual burden. Each dot represents
5 per cent of observations. The relationship seems to be quite strong across geographies
and over time.
Figure B.5 plots the income volatility process using the data by Guvenen, Pistaferri, and
Violante (2022). The level is particularly high for the very top quantile of the distribution.
Figure B.1: Share of U.S. Treasuries held domestically vis-à-vis abroad over the period 1970–2019
Notes: See Tables B.5 and B.6 in Section B.5 of the appendix for details on the variables sources and construction.
240

Figure B.2: Domestically-held private safe assets as a share of domestically-held public safe assets, and shadow banking sector as a share of traditional banking sector in the United States over
the period 1950-2019
Notes: The private safe assets are composed of: market mutual funds, commercial paper, and RePos. The public safe
assets are composed of: Treasuries, checking, and savings and time deposits. Both aggregates refer to domestically held
claims. See Tables B.5 and B.6 in Section B.5 of the appendix for details on the variables sources and construction.
Figure B.3: Top 1 percent share of the income distribution and households’ portfolio share invested in institutional investors in the United States over the period 1971-2019
Notes: The portfolio share in institutional funds mimics the one built by Jordà et al. (2019). Sources: Data on inequality
are retrieved from the online appendix of Auten and Splinter (2024). Data on households’ portfolio shares are from the
extended Survey of Consumer Finances (SCF+). See Tables B.5 and B.6 in Section B.5 of the Appendix for details on the
variables sources and construction.
241

Figure B.4: Quantile plot showing the correlation between top 5 per cent share of the income
distribution and credit to GDP (in logs) for 18 economies over the period 1970-2019
Notes: Each dot represents a bin of one per cent across all the inequality-credit pairs. The line of best fit is constructed as
the correlation between all the credit and inequality pairs in the panel. Sources: Credit data is obtained from the Bank
of International Settlements. See Table B.5 in Section B.5 of the appendix for details on the variables sources.
242

Figure B.5: Income volatility by income level in the United States across quantiles over the period
1998-2019
Panel A: Total population across genders
Panel B: Male population
Panel C: Female population
Source: Guvenen, Pistaferri, and Violante (2022). Notes: See the cited paper for variables description and construction.
243

B.2 Model derivations
Equilibrium definition. A sequential market equilibrium for the economy presented
in Section 2.3 of the paper is a set of prices {pKt, qBt, qM t, qLt, dt
, wt}
∞
t=0 and quantities
{c
(I)
t
, k(I)
t
, b(I)
t
, m
(I)
t
, lt
, c
(P)
t
, φ1t
, φ2t}
∞
t=0 such that:
1. Investors maximize their utility and the returns on their portfolio according to Problem (PI ) at the optimal prices.
2. Workers maximize their utility according to Problem (PW ) at the optimal prices.
3. Firms are price-takers and statically maximize their profits from Problem (PF ) given
optimal factor prices d
∗
t
, w∗
t
.
4. Price-taker financial intermediaries transform debt into safe assets according to a
linear technology lt = mt
, at optimal prices q
∗
Lt = q
∗
M t.
5. The Government budget constraint holds with equality in each period at the optimal
price q
∗
Bt, subject to a constraint ¯b.
6. Markets for goods, risky capital, safe assets, quasi-safe assets, and labor clear.
Optimality conditions. The model features aggregation (Angeletos, 2007), therefore the
overall amount of assets for each agent, Ait = (pKt(1 + it) + dt) kit + bit + χ
M
it mit, is a
sufficient state variable. Log preferences ensure that consumption is a constant fraction
(1 − β) of income. The overall policy function for savings is therefore βAit, which one
can segment according to three different shares — one for each of the financial assets. It
follows that investors’ policy functions are:
c
(I)
it = (1 − β)Ait (B.1)
qBtb
(I)
i,t+1 = βφ1tAit (B.2)
244

qM tm
(I)
i,t+1 = βφ2tAit (B.3)
pKtk
(I)
i,t+1 = β(1 − φ1t − φ2t)Ait (B.4)
By taking the FOCs of the Problem (PI ) with respect to c
(I)
it and b
(I)
i,t+1, the Euler Equation of
Investors pinning down the marginal decisions between intertemporal consumption and
safe bonds can be written as:
qBt = βEt
"
c
(I)
it
c
(I)
i,t+1#
(B.5)
Substitute Equation (B.1) twice in (B.5) and obtain:
qBt = βEt
"
✘✘✘✘ (1 − β)A
(I)
it
✘✘✘✘ (1 − β)A
(I)
i,t+1#
Use the definition of A
(I)
is = (pKs(1 + is) + ds)kis + bis + χ
M
is mis for s = t + 1, and plug
back in the previous equation.
qBt = βEt
"
A
(I)
it
(pK,t+1(1 + i,t+1) + dt+1)ki,t+1 + bi,t+1 + χ
M
i,t+1mi,t+1#
(B.6)
Substitute the policy functions analytic forms in (B.2), (B.3), (B.4) and Equation (B.6) to
obtain:
qBt = ✁
β✁Et

✟A✟it
(pK,t+1(1+i,t+1)+dt+1)✟βA✟it
(1−φ1t−φ2t
)
pKt
+✟βA✟it
φ1t
qBt
+✟βA✟it
φ2t
qMt
χM
i,t+1
qBt = Et
h
1
φ1tRBt+φ2tRMt+(1−φ1t−φ2t)Rt+1 i
1 = Et
h
RBt
φ1tRBt+φ2tRMt+(1−φ1t−φ2t)Rt+1 i
(B.7)
where RBt = 1/qBt, RM t = χ
M
it /qBt, and Rt+1 = [pK,t+1(1 + i,t+1) + dt+1]/pKt.
Similarly, by taking the FOCs with respect to consumption and quasi-safe assets
245

(mi,t+1), it is possible to write the implicit optimal condition for φ2t
:
1 = Et

RM t
φ1tRBt + φ2tRM t + (1 − φ1t − φ2t)Rt+1
(B.8)
The optimal condition for the workers is:
qLt = β
u
0
(c
(W)
t+1 )
u
0
(c
(W)
t
)
(1 − L + λlt+1) (B.9)
Market Clearing Conditions.
bt+1 = ¯b ∀t (B.10)
lt+1 = mt+1 ⇐⇒ qM t = qLt ∀t (B.11)
kt+1 = 1 ∀t (B.12)
Lt = 1 ∀t (B.13)
The price of consumption is taken as numéraire.
Proof of Lemma 2.
The law of motion for the assets distribution follows from the definition of assets for the
economy:
Ait , (pt(1 + it) + dt)kit + bit + +χ
M
it mit ∀t
⇒ Ai,t+1 = (pt+1(1 + i,t+1) + dt+1) ki,t+1 + bi,t+1 + χ
M
i,t+1mi,t+1
By re-arranging and plugging the policy functions into the previous equation, we can re246

write it as:
Ai,t+1 = (pK,t+1(1 + i,t+1) + dt+1)
βAit(1 − φ1t − φ2t)
pKt
+
1
qBt
φ1tβAit +
χ
M
i,t+1
qM t
φ2tβAit
= βAit
(1 − φ1t − φ2t)Ri,t+1 + φ1tR
B
t+1 + φ2tR
M
i,t+1
(B.14)
B.3 Numerical solution
The model features “aggregation”, therefore the income distribution is not a relevant
variable to pin down the equilibrium prices and quantities. I can proceed then to solve
numerically in two parts. First, I compute the steady state abstracting from the income distribution of agents. In this case, I solve a system of non-linear equations around the steady
state. The system is composed by the policy functions (B.1)-(B.4), the optimal conditions
for portfolio shares (B.7) and (B.8), the Euler equation (B.9) and the budget constraint
for the workers problem, and the market clearing conditions (B.10)-(B.12), which can be
solved for {c
(P)
t
, φ1t
, φ2t
, pKt, kt+1, qM t, mt+1, qBt, bt+1, qLt, lt+1}.
With that in hand, I use the steady state values for {p
∗
Kt, q∗
M t, q∗
Bt, φ∗
1t
, φ∗
2t}, and plug
them in the law of motion in Equation (B.14) to solve for the ergodic income distribution.1
The algorithm to find the distribution proceeds according to the following steps:
• Guess an initial asset distribution, Mft
, over a grid, A˜
t
, with an arbitrarily small bin
size, µ˜.
Let the grid lower bound be a scalar A > 0 arbitrarily close to zero for all t. Choose
an upper bound for the grid A¯
t
large enough to include at least the true total income
of the economy, A∗
, computed before.
Let me t be the initial distribution mass for a bin located on the grid point a˜mt ∈ A˜
t
1
Such procedure is isomorphic, yet computationally faster, than the contemporaneous solution for the distribution and
the steady state variables.
247

such that PMt
m˜ =1 m˜ t = 1 ∀t.
• Let every bin, m˜ t
, be hit by idiosyncratic shocks, i,t+1, i ∈ {1, 2, . . . I}. Compute the
new asset values a˜i,m,t+1 for each m˜ t (originally located in position a˜i,m,t) using the
assets law of motion in (B.14).
• Allocate each shock realization on the new grid A˜
t+1. To do so, assume ∼ U[,¯],
then each shock realization will carry a weight 1/I to be multiplied by the original
probability mass, m˜ t
, associated with each grid point, a˜mt. In other words, each
shock realization moves a mass mt/I.
– If a˜i,m0
,t+1 < A, then allocate all of the distribution weight carried by the realization, mt/I, to the first point on the grid, A.
– If A < a˜i,m0
,t+1 < A¯
t
, then allocate a part, ω, of the weight, mt/I, to the grid
point a˜m0
,t+1 and (1 − ω) to a˜m0+1,t+1 according the their distance from the grid
points ω = 1 − (˜ai,m0
,t+1 − a˜m0
,t+1)/µ. In this way, each original weight is split
according to the linear distance between the two most adjacent grid points.2
– If a˜i,m0
,t+1 > A¯
t
, then add new grid points to the previous grid, A˜
t to form a new
grid A˜
t+1. The number of new points to add depends on how far away the top
realizations fall with respect to A¯
t
: (˜ai,m0
,t+1 − A¯
t)/µ gives the number of grid
points to add. Compute the specific weights ω for each bin between the new
adjacent grid points according to the procedure illustrated above.
• Sum all probability masses, mi,t+1, for each new grid point, am,t+1 on the new grid
A˜
t+1 to achieve a new distribution Mft+1.
• Remove a fraction (1 − δ) from each bin to account for the survival rate δ.
2To be sure, the “point of departure” on the initial grid a˜m,t is not necessarily the same as the "point of arrival" in the
new grid a˜m0
,t+1.
248

• Re-allocate the fraction of population (1 − δ) to the individuals with average value
on the grid to ensure that no income destruction occurs.
– If the average income value falls between two grid points, allocate them with a
weight that is proportional to the distance to their closest point — as explained
above.
• Check if the new and the old distribution coincide up to an arbitrarily small scalar:
Mft ≈ Mft+1.
– If not, impose Mft = Mft+1, and start the loop over.
– Else, convergence has been reached. The sought-after stationary distribution
has been found.
• Make sure that the total level of income for the economy corresponds to the true
value: Mf∗ A˜ = A∗
. If not, change the initial distribution guess A˜
t
, and start over.
• Repeat by refining the grid to make sure the result is robust.
In other words, to find the stationary distribution I guess an initial distribution for the
equilibrium values, and I subsequently operate an asymmetric grid expansion (for the
right tail) until equilibrium is found. The grid expansion is asymmetric because – on the
one hand – Inada conditions prevent agents from consuming negative amounts; therefore,
even agents with a complete streak of negative shock realizations will be able to obtain a
strictly positive asset level albeit arbitrarily close to zero. On the other hand, lucky agents
with a complete streak of positive income shocks may become arbitrarily rich ex-post. To
avoid the distribution from depending excessively on the last bin of rich lucky agents, the
expanding distribution spreads the lucky agents over new bins according to their income
level. In this way, agents are appropriately allocated to their actual income rather than be
approximated by an arbitrary last bin of an otherwise fixed grid. The initial guess of the
distribution can be slightly changed to begin with in order to make sure that the overall
249

income of the economy corresponds to the steady state levels. However, robustness checks
have proved that the sensitivity to such changes is very low.
B.4 Empirical robustness checks
In order to provide further evidence to the empirical results in a panel setting showed
in Tables 2.5a and 2.5b allowing for the Arellano-Bond correction method to be taken into
account.
∆yit = δ∆yi,t−1 +
X
5
s=2
βs∆xi,t−s + β0,i + κt + γ
0Xit + it, (B.15)
∆yit = δ∆yi,t−1 +
X
5
s=2
βs∆xi,t−s +
X
5
s=2
βs∆xi,t−s1(i ∈ A) + β0,i + κt + γ
0Xit + it (B.16)
Given that the panel is dynamic, it is important to carry out an additional robustness check
to check that nuisance parameters do not affect the estimates. The results are provided in
Tables B.1 and B.2. The tables show that the long-run coefficients are the just as large as
the ones found in the body of the paper for the direct channel (around 0.29). Furthermore, the inclusion of the market based dummy variable is not important. The short-run
effects are about half the magnitude of the long-run effects (0.15) – consistent with the
idea that it may take time to fully manifest – and sometimes less precise than the long-run
counterparts.
About the feedback effects, the results are again not true in general. More credit activity
does not seem to lead to more inequality, unless a dummy accounting for the market-based
system and the pricing effects of asset valuation is included.
In light of the potential bias arising from the auto-regressive component in a panel setting, I repeat the analysis with the Arellano-Bond estimator in Tables B.3 and B.4. Results
are consistent therefore the bias tends to zero rapidly enough.
250

Table B.1: Regression results for the short- and long-run effects of inequality on total loans for a
host of 18 economies over the period 1970-2019
Total loans
(1) (2) (3) (4) (5) (6) (7) (8)
Top 5 income share 0.593*** 0.344*** 0.283** 0.293** 0.638*** 0.389*** 0.319** 0.296**
(Long run effect) (0.167) (0.131) (0.123) (0.134) (0.185) (0.140) (0.135) (0.138)
Top 5 income share × -0.295 -0.292 -0.237 -0.018
Mkt-based dummy (Long run) (0.401) (0.375) (0.392) (0.329)
Top 5 income share 0.242*** 0.157** 0.133* 0.139* 0.258*** 0.176** 0.150** 0.140*
(Short run effect) (0.076) (0.068) (0.065) (0.070) (0.078) (0.071) (0.068) (0.071)
Top 5 income share × -0.119 -0.132 -0.111 -0.008
Mkt-based dummy (Short run) (0.154) (0.166) (0.180) (0.156)
Time fixed effect
Domestic controls
Globalization controls
USA excluded
R
2
0.588 0.637 0.641 0.642 0.591 0.638 0.643 0.643
Countries/Obs. 18/674 18/673 18/669 17/620 18/674 18/673 18/669 17/620
Table B.2: Regression results for the short- and long-run effects of total loans on inequality for a
host of 18 economies over the period 1970-2019
Top 5 income share
(1) (2) (3) (4) (5) (6) (7) (8)
Total loans -0.071* -0.039 -0.042 -0.049 -0.112*** -0.075** -0.076** -0.076**
(Long run effect) (0.040) (0.036) (0.037) (0.039) (0.038) (0.051) (0.037) (0.038)
Total loans × 0.228*** 0.204*** 0.183*** 0.187***
Mkt-based dummy (Long run) (0.051) (0.042) (0.044) (0.043)
Total loans -0.074* -0.041 -0.045 -0.052 -0.117*** -0.081** -0.083** -0.082*
(Short run effect) (0.040) (0.037) (0.039) (0.041) (0.039) (0.038) (0.039) (0.040)
Total loans × 0.239*** 0.221*** 0.199*** 0.202***
Mkt-based dummy (Short run) (0.050) (0.041) (0.044) (0.043)
Time fixed effect
Domestic controls
Globalization controls
USA excluded
R
2
0.185 0.213 0.237 0.238 0.214 0.243 0.267 0.271
Countries/Obs. 18/732 18/731 18/727 17/678 18/732 18/731 18/727 17/678
251

Table B.3: Regression results for the short- and long-run effects of inequality on total loans using
the Arellano-Bond estimator for a host of 18 economies over the period 1970-2019
Total loans
(1) (2) (3) (4) (5) (6) (7) (8)
Top 5 income share 0.593*** 0.344*** 0.283** 0.293** 0.638*** 0.389*** 0.319** 0.296**
(Long run effect) (0.156) (0.122) (0.114) (0.124) (0.172) (0.129) (0.125) (0.127)
Top 5 income share × -0.295 -0.292 -0.237 -0.018
Mkt-based dummy (Long run) (0.372) (0.347) (0.362) (0.302)
Top 5 income share 0.242*** 0.157** 0.133** 0.139** 0.258*** 0.176*** 0.150** 0.140**
(Short run effect) (0.071) (0.063) (0.060) (0.065) (0.072) (0.066) (0.063) (0.065)
Top 5 income share × -0.119 -0.132 -0.111 -0.008
Mkt-based dummy (Short run) (0.143) (0.153) (0.166) (0.143)
Time fixed effect
Domestic controls
Globalization controls
USA excluded
R
2
0.588 0.637 0.641 0.642 0.591 0.638 0.643 0.643
Countries/Obs. 18/656 18/655 18/651 17/603 18/656 18/655 18/651 17/603
Table B.4: Regression results for the short- and long-run effects of total loans on inequality using
the Arellano-Bond estimator for a host of 18 economies over the period 1970-2019
Top 5 income share
(1) (2) (3) (4) (5) (6) (7) (8)
Total loans -0.077** -0.042 -0.044 -0.051 -0.114*** -0.078** -0.077** -0.077*
(Long run effect) (0.039 (0.036 (0.037 (0.039 (0.039 (0.060 (0.038 (0.039
Total loans × 0.232*** 0.213*** 0.192*** 0.202***
Mkt-based dummy (Long run) (0.060 (0.047 (0.048 (0.047
Total loans -0.080** -0.045 -0.047 -0.055 -0.120*** -0.084** -0.084** -0.083**
(Short run effect) (0.040 (0.038 (0.039 (0.041 (0.040 (0.040 (0.040 (0.041
Total loans × 0.243*** 0.230*** 0.208*** 0.218***
Mkt-based dummy (Short run) (0.059 (0.046 (0.048 (0.046
Time fixed effect
Domestic controls
Globalization controls
USA excluded
R
2
0.185 0.213 0.237 0.238 0.214 0.243 0.267 0.271
Countries/Obs. 18/732 18/731 18/727 17/678 18/732 18/731 18/727 17/678
252

B.5 Data sources and construction
Idiosyncratic volatility. To compute the idiosyncratic risk on capital used in the calibration exercise, I use CRSP data on returns of stocks listed on the New York Stock Exchange (NYSE), the Nasdaq, and the American Stock Exchange (AMEX) over the period
1970-2019. Furthermore, by following the approach by Fu (2009), I retrieve data on FamaFrench 3 market factors from Professor French website.
The steps of the procedure can be described as follows:
1. Retrieve stock returns from CRSP data for the stock exchanges listed above at daily
frequency.
2. Retrieve daily data on T-bills and Fama-French 3 market factors — Equity premium,
High-minus-Low (HML), Small-minus-Large (SML) —, and merge the data sets.
3. Run the following cross-section regressions at monthly frequency for all the trading
days available:
Ridm − R
f
dm = β0i + β1i(R
M
dm − R
f
dm) + β2iHMLdm + β3iSMLdm + idm ∀i, m
where i = 1, . . . , N identifies the firm, and d = 1, . . . , D identifies the day of the
month m = 1, . . . , M. Compute and store the residuals ˆidm.
4. Compute the daily standard deviation, σˆidm, of ˆidm ∀i, m, and transform it into
monthly volatility (σˆim) by multiplying it for the square root of the firm-specific
number of trading days in the month, Dim.
5. Average the monthly volatility across firms for each month: σˆm =
PN
i=1 σˆim.
6. To find the idiosyncratic volatility over a period of time such as the steady states 1970-
1979 and 2010-2019, annualize the monthly volatility (ˆσy =
√
12ˆσm), and average the
volatility over time for the horizon of interest.
253

Table B.5: Variables description and sources
Variable Variable details Source
Bottom 90 percent income share 1- US Top 10 post-tax national income share, equal split (“sdiinc992jUS”) World Inequality Database
Capital income share 1-labor income share NIPA, Table 2
Consumer price index Consumer Price Index: Total All Items for the United States, Index
2015=100, Annual, Seasonally Adjusted
Organization for Economic
Co-operation and Development
Credit (developed economies) Credit to Private non-financial sector from all sectors at market value,
Domestic currency - Adjusted for breaks
Bank of International
Settlements
Debt liabilities (developed
economies)
Sum of the stocks of portfolio debt liabilities and other investment
liabilities, nonresident
Milesi-Ferretti (2022)
Financial sector, total financial
assets
Domestic Financial Sectors; Total Financial Assets, Level, Millions of
Dollars, Annual, Not Seasonally Adjusted (FBTFASA027N)
Federal Reserve Board, Financial
Accounts of the US - Z.1 Tables
Financial sector, total financial
assets domestically-held
See variable construction Federal Reserve Board, Financial
Accounts of the US - Z.1 Tables
Financial sector, total financial
assets held abroad
See variable construction Federal Reserve Board, Financial
Accounts of the US - Z.1 Tables
Government spending
(developed economies)
Government expenditure (nominal, local currency) Macrohistory database by Jordà
et al. (2019)
GDP, nominal (developed
economies)
Gross Domestic Product in current LCU (NY.GDP.MKTP.CN) World Bank
GDP, nominal (U.S.) Gross Domestic Product, Billions of Dollars, Annual, Not Seasonally
Adjusted
IMF through FRED
Idiosyncratic volatility Stock prices idiosyncratic volatility CRSP
Income volatility by quantile Earnings income volatility with respect to permanent income component GRID dataset by Guvenen,
Pistaferri, and Violante (2022)
254

Interest rates, nominal (AAA
corporate bond yields)
Moody’s Seasoned Aaa Corporate Bond Yield, Percent, Monthly, Not
Seasonally Adjusted
Moody’s through FRED
Interest rates, nominal
(Treasuries)
Market Yield on U.S. Treasury Securities at 10-Year Constant Maturity,
Quoted on an Investment Basis, Percent, Annual, Not Seasonally Adjusted
Federal Reserve Board, Financial
Accounts of the US - H.15 Tables
Labor income share Compensation employees / (Personal Income + Subsidies - Taxes). See
variable construction
NIPA, Table 2
Loans (developed economies) Total loans in local currency unit Macrohistory database by Jordà
et al. (2019)
Money supply (developed
economies)
Broad money (nominal, local currency) Macrohistory database by Jordà
et al. (2019)
Old dependency ratio
(developed economies)
Age dependency ratio, old (% of working-age population). Variable code:
SP.POP.DPND.OL
World Bank
Population (developed
economies)
Total national population Macrohistory database by Jordà
et al. (2019)
Portfolio share in institutional
funds
Household balance sheet composition in the US. Agricultural land,
pension, insurance and investment fund claims.
Survey of Consumer Finances
(SCF+) through Jordà et al.
(2019)
Quality-adjusted finance output
in levels
Stock of outstanding intermediated assets adjusted for quality (fin all ck) Philippon (2015), online
appendix
Safe assets, domestically-held Safe assets, total financial assets - safe assets held abroad Federal Reserve Board, Financial
Accounts of the US - Z.1 Tables
Safe assets, held abroad See variable construction table Federal Reserve Board, Financial
Accounts of the US - Z.1 Tables
Safe assets, total financial assets See variable construction table Federal Reserve Board, Financial
Accounts of the US - Z.1 Tables
Safe assets, public and
traditional banking
domestically-held
See variable construction table Federal Reserve Board, Financial
Accounts of the US - Z.1 Tables
255

Safe assets, shadow banking
domestically-held
See variable construction table Federal Reserve Board, Financial
Accounts of the US - Z.1 Tables
Shadow banking, total financial
assets
See variable construction table Federal Reserve Board, Financial
Accounts of the US - Z.1 Tables
Top 1 per cent fiscal income, AS Top 1 per cent fiscal income share, set income groups by size-adjusted
income and number of individuals
Auten and Splinter (2024),
online appendix
Top 1 per cent fiscal income, PSZ Top 1 per cent fiscal income share World Inequality Database
Top 5 per cent income share
(developed economies)
Pre-tax national income share, equal split (“sptinc992j”) World Inequality Database
Top 5 per cent wealth share Top 5 per cent net wealth World Inequality Database
Trade balance Exports (nominal, local currency) - Imports (nominal, local currency) Macrohistory database by Jordà
et al. (2019)
Traditional banking, total
financial assets
Private Depository Institutions; Total Financial Assets, Level
(BOGZ1FL704090005A)
Federal Reserve Board, Financial
Accounts of the US - Z.1 Tables
Treasuries, domestic holdings Federal Government; Treasury Securities; Liability, Level (FL893161705A) Federal Reserve Board, Financial
Accounts of the US - Z.1 Tables
Treasuries, foreign holdings Rest of the World; Treasury Securities; Asset, Market Value Levels
(LM263061105A)
Federal Reserve Board, Financial
Accounts of the US - Z.1 Tables
Notes: All values for which no specific geographic definition is provided refer to the United States. Codes in parentheses refer to the Financial Accounts series code
number. Countries included among developed economies: Australia, Belgium, Canada, Denmark, Finland, France, Germany, Ireland, Italy, Japan, the Netherlands,
Norway, Portugal, Spain, Sweden, Switzerland, the UK, USA.
*Accessed through FRED – Federal Reserve Economic Data, St. Louis Fed. **Accessed through Jordà, Knoll, Kuvshinov, Schularick, and Taylor (2019).
256

Table B.6: Construction of macro-financial variables with series references
Variable Variable construction details
Financial sector, total financial assets domestically-held Sum of the following components:
Domestic Financial Sectors; Net Interbank Transactions; Liability, Level (BOGZ1FL794110005A)
Domestic Financial Sectors; Checkable Deposits and Currency; Liability, Level
(BOGZ1FL793120005A)
Private Depository Institutions; Total Time and Savings Deposits; Liability, Level
(BOGZ1FL703130005A)
Money Market Funds; Total Financial Assets, Level (MMMFFAA027N)
Domestic Financial Sectors; Federal Funds and Security Repurchase Agreements; Liability, Level
(BOGZ1FL792150005A)
Domestic Financial Sectors; Open Market Paper; Liability, Level (FBMPLIA027N)
GSEs and Agency- and GSE-Backed Mortgage Pools; U.S. Government Agency Securities; Liability,
Level (GSEMPUA027N)
Domestic Financial Sectors; Corporate and Foreign Bonds; Liability, Level (FBCFLIA027N)
Domestic Financial Sectors; Loans; Liability, Level (BOGZ1FL794123005A)
Mutual Funds; Mutual Fund Shares; Liability, Market Value Levels (BOGZ1LM653164205A)
Domestic Financial Sectors; Trade Payables; Liability, Level (BOGZ1FL793170005A)
Life Insurance Companies; Life Insurance Reserves; Liability, Level (BOGZ1FL543140005A)
Insurance Companies and Pension Funds; Pension Entitlements; Liability (BOGZ1FL583150005A)
Domestic Financial Sectors; Total Miscellaneous Liabilities, Level (BOGZ1FL793190005A)
Financial sector, total financial assets held abroad Sum of the following components:
Rest of the World; Net Interbank Transactions with Banks in Foreign Countries; Asset, Level
(ROWNIBA027N)
Rest of the World; U.S. Checkable Deposits and Currency; Asset, Level (BOGZ1FL263020005A)
Rest of the World; U.S. Total Time and Savings Deposits; Asset, Level (ROWTDAA027N)
Rest of the World; U.S. Money Market Fund Shares; Asset, Level (ROWMMMA027N)
257

Rest of the World; Security Repurchase Agreements; Asset, Level (BOGZ1FL262051003A)
Rest of the World; Commercial Paper; Asset, Market Value Levels (BOGZ1LM263069103A)
Rest of the World; Agency- and GSE-Backed Securities; Asset, Market Value Levels
(BOGZ1LM263061705A)
Rest of the World; Corporate Bonds; Asset, Market Value Levels (BOGZ1LM263063005A)
Rest of the World; U.S. Mutual Fund Shares; Asset, Market Value Levels (BOGZ1LM263064203A)
Rest of the World; Trade Receivables; Asset, Market Value Levels (BOGZ1LM263070005A)
Life Insurance Companies; Assumed Life Insurance Reserve Credit from Non-U.S. Insurers;
Liability, Level (BOGZ1FL543141905A)
Life Insurance Companies; Assumed Pension Entitlement Reserve Credit from Non-U.S. Insurers;
Liability, Level (BOGZ1FL543151905A)
Rest of the World; Assumed Policy Payables by U.S. Reinsurers from Non-U.S. Insurers; Liability,
Level (BOGZ1FL263076005A)
Inflation Rate of change of Consumer Price Index
Labor income share National Income and Products Account (NIPA) – Table 2. Computed as: Compensation of
employees (Line 2) / [Personal Income (Line 1) - Personal current taxes (Line 26) + Government
social benefits to persons (Line 17)]
Real interest rate Interest rates, nominal (AAA corporate bond yields) - inflation
Safe assets, held abroad Following Gorton, Lewellen, and Metrick (2012), sum of the following components (with bonds
and GSEs accounted for 85 per cent of the value):
Rest of the World; Agency- and GSE-Backed Securities; Asset, Market Value Levels
(BOGZ1LM263061705A)
Rest of the World; Commercial Paper; Asset, Market Value Levels (BOGZ1LM263069103A)
Rest of the World; Corporate Bonds; Asset, Market Value Levels (BOGZ1LM263063005A)
Rest of the World; Municipal Securities; Asset, Level (ROWMLAA027N)
Rest of the World; Security Repurchase Agreements; Asset, Level (BOGZ1FL262051003A)
Rest of the World; Treasury Securities; Asset, Market Value Levels (BOGZ1LM263061105A)
258

Rest of the World; U.S. Checkable Deposits and Currency; Asset, Level (BOGZ1FL263020005A)
Rest of the World; U.S. Money Market Fund Shares; Asset, Level (ROWMMMA027N)
Rest of the World; U.S. Total Time and Savings Deposits; Asset, Level (ROWTDAA027N)
Safe assets, total financial assets Following Gorton, Lewellen, and Metrick (2012), sum of the following components (with bonds
and GSEs accounted for 85 per cent of the value):
All Sectors; Agency- and GSE-Backed Securities; Asset, Level (BOGZ1FL893061705A)
All Sectors; Corporate and Foreign Bonds; Asset, Level (BOGZ1FL893063005A)
All Sectors; Federal Funds and Security Repurchase Agreements; Asset, Level
(BOGZ1FL892050005A)
All Sectors; Municipal Securities; Asset, Level (BOGZ1FL893062005A)
All Sectors; Open Market Paper; Liability, Level (BOGZ1FL893169175A)
All Sectors; Treasury Securities; Asset, Level (BOGZ1FL893061105A)
Domestic Financial Sectors; Checkable Deposits and Currency; Liability, Level
(BOGZ1FL793120005A)
Money Market Funds; Total Financial Assets, Level (MMMFFAA027N)
Private Depository Institutions; Total Time and Savings Deposits; Liability, Level
(BOGZ1FL703130005A)
Safe assets, public and traditional banking domestically-held Sum of the total - foreign components of the following elements:
All Sectors; Treasury Securities; Asset, Level (BOGZ1FL893061105A) minus Rest of the World;
Treasury Securities; Asset, Market Value Levels (BOGZ1LM263061105A)
All Sectors; Municipal Securities; Asset, Level (BOGZ1FL893062005A) minus Rest of the World;
Municipal Securities; Asset, Level (ROWMLAA027N)
Domestic Financial Sectors; Checkable Deposits and Currency; Liability, Level
(BOGZ1FL793120005A) minus Rest of the World; U.S. Checkable Deposits and Currency; Asset,
Level (BOGZ1FL263020005A)
Private Depository Institutions; Total Time and Savings Deposits; Liability, Level
(BOGZ1FL703130005A) minus Rest of the World; U.S. Total Time and Savings Deposits; Asset,
Level (ROWTDAA027N)
Safe assets, shadow banking domestically-held Sum of the total - foreign components of the following elements:
Money Market Funds; Total Financial Assets, Level (MMMFFAA027N) minus Rest of the World;
U.S. Money Market Fund Shares; Asset, Level (ROWMMMA027N)
259

Domestic Financial Sectors; Federal Funds and Security Repurchase Agreements; Liability, Level
(BOGZ1FL792150005A) minus Rest of the World; Security Repurchase Agreements; Asset, Level
(BOGZ1FL262051003A)
Domestic Financial Sectors; Open Market Paper; Liability, Level (FBMPLIA027N) minus Rest of
the World; Commercial Paper; Asset, Market Value Levels (BOGZ1LM263069103A)
0.85*(GSEs and Agency- and GSE-Backed Mortgage Pools; U.S. Government Agency Securities;
Liability, Level (GSEMPUA027N) minus Rest of the World; Agency- and GSE-Backed Securities;
Asset, Market Value Levels (BOGZ1LM263061705A))
Shadow banking, total financial assets Sum of the following components:
Agency-and GSE-Backed Mortgage Pools; Total Mortgages; Asset, Level (BOGZ1FL413065005A)
Exchange-Traded Funds; Total Financial Assets, Market Value Levels (BOGZ1LM564090005A)
Finance Companies; Total Financial Assets, Level (BOGZ1FL614090005A)
Funding Corporations: Other Financial Business; Total Financial Assets, Level
(BOGZ1FL504090005A)
Government-Sponsored Enterprises; Total Financial Assets, Level (BOGZ1FL404090005A)
Issuers of Asset-Backed Securities; Total Financial Assets, Level (BOGZ1FL674090005A)
Money Market Funds; Total Financial Assets, Level (MMMFFAA027N)
Private Pension Funds; Total Financial Assets, Level (BOGZ1FL574090005A)
Real Estate Investment Trusts; Total Financial Assets, Level (BOGZ1FL644090005A)
Security Brokers and Dealers; Total Financial Assets, Level (BOGZ1FL664090005A)
260

Appendix C
Appendix to Chapter 3
Section C.1 of this appendix refer to the anecdotal evidence that justifies the narrative
described in the main body of the paper. Such pieces of information may inform a fullfledged textual analysis of the deregulation index to account for expectations in the future.
Section Section C.2 illustrates the theoretical proofs of the propositions present in the main
paper. Section C.3 provides further empirical results. Section C.4 lists the details of the
sources utilized for the production of the stylized facts and the statistical estimates.
261

C.1 Anecdotal evidence
Figure C.1: Anecdotal evidence of regulators and traditional banks assessment of the changing
banking landscape
BNP Paribas Annual report 2016.
SF Fed - Economic Letters (July 2001-22).
Jamie Dimon letter to
JPM investors. April 4
th, 2019.
C.2 Additional theoretical proofs
From the FOCs of the household, we can find the the Euler equations and the noarbitrage conditions:
ct+1
ct
= β
RK
t+1 + p
K
t+1(1 − δ)
p
K
t
(EEK)
262

= β
RBK
t+1 + p
BK
t+1
p
BK
t
(EEBK)
= β
RSB
t+1 + p
SB
t+1
p
SB
t
(EESB)
Equation (EEK) is the Euler equation with respect to capital, while (EEBK) and
(EESB) pin down the conditions with respect to the banking and shadow banking sectors, respectively. By combing the three conditions, it is possible to derive the no arbitrage
conditions.
The consumption goods maximization problem leads to two optimal conditions with respect to RK
t and wt
:
α
Ct
φ
K
t kt
=R
K
t
(C.1)
(1 − α)p
C
t Ct =wt (C.2)
While the problem (3.8 for real investments lead to the optimal condition: p
K
t AK = RK
t
.
Proof of Proposition 1 Recall the Euler Equation:
ct+1
ct
= β
RK
t+1 + p
K
t+1(1 − δ)
p
K
t
= β
RK
t+1
RK
t
(AK + 1 − δ) (EEK.bis)
(C.3)
where the second line follows from: p
K
t AK = RK
t
; but from (C.1), we have:
R
K
t = α
Ct
φ
K
t kt
⇒
RK
t+1
RK
t
=
gC
gφK
gK
(C.4)
263

therefore:
gC = β
gC
gφK gK
(AK + 1 − δ) (C.5)
gφK
gK = β(AK + 1 − δ) (EEK.ter)
But: gK = AK(1 − φ
K
t
) + 1 − δ, therefore:
gφK =
β(AK + 1 − δ)
AK(1 − φ
K
t
) + 1 − δ
(C.6)
If φ
K
t
constant, then gφK = 1, and:
φ
K = 1 + (1 − β)
1 − δ
AK
− β
Assume gφK > 1 then (1 − φ
K
t
) decreases in time. We have two cases, of which only one
is economically meaningful. If φ
K
t
is not bounded, then given that gφK
gK is constant, and
from the expression in Equation (C.6) we conclude that gφK
is negative, but given that
gφK > 1 by assumption, we reach a contradiction. If φ
K
t
is bounded in the interval (0,1),
then after having increased to 1, gφK =
β(AK+1−δ)
1−δ which is again a contradiction. The same
steps can be applied if we assume gφK < −1. We conclude φK must be constant for all
t.
Growth rates: From the Euler equations and the optimal pricing conditions, we have:
Ct+1
Ct
= β
RK
t+1 + p
K
t+1(1 − δ)
p
K
t
= β
RK
t+1
RK
t
(AK + 1 − δ)
but RK
t+1
RK
t
=
Ct+1/Ct
kt+1/kt
264

Kt+1
Kt
| {z }
gK
= β(AK + 1 − δ)
Given the proof of φ
K
t = φ
K∀t, and using the law of motion of real capital:
Kt+1
Kt
| {z }
gK
= AK(1 − φK) + 1 − δ
but gK = β(AK + 1 − δ)
⇒ φ
∗
K = (1 − β)
AK + (1 − δ)
AK
⇒
Ct+1
Ct
| {z }
gC
=
AC(φKkt+1)
αL
1−α
AC(φKkt)
αL1−α
=

kt+1
kt
α
= (β(1 + AK − δ))α
For the financial sector:
gIBK =
ABKF
BK
t+1
ABKF
BK
t
=
F
BK
t+1
F
BK
t
=
I
BK
t
F
BK
t
+ 1
= 1 + ABK
= gFBK
gFSB =
I
BK
t
F
BK
t
+ 1
= 1 + ASB
265

For the Nominal shares, we have:
Ct+1
Ct
= β
RBK
t+1 + p
BK
t+1
p
BK
t
but p
BK
t ABK =
σ
σ − 1
R
BK
t
Ct+1
Ct
= β
RBK
t+1
RBK
t

(σ − 1)ABK + σ
σ

but Ct+1
Ct
= (β(1 + AK − δ))α
⇒
RBK
t+1
RBK
t
=
(β(AK + 1 − δ))α
β(1 + ABK
σ−1
σ
)
RBK
t+1F
BK
t+1
RBK
t F
BK
t
= (1 + ABK)
(β(AK + 1 − δ))α
β(1 + ABK
σ−1
σ
)
p
BK
t+1F
BK
t+1
p
BK
t F
BK
t
= (1 + ABK)
(β(AK + 1 − δ))α
β(1 + ABK
σ−1
σ
)
One can apply the same logic to the shadow banking sector growth rates with the only
difference of having ASB as productivity for the sector. As such, the relative growth rate
of one with respect to the other are:
F
BK
t+1 /F BK
t
F
SB
t+1/F SB
t
=
1 + ABK
1 + ASB
(p
BK
t+1F
BK
t+1 )/(p
BK
t F
BK
t
)
(p
SB
t+1F
SB
t+1)/(p
SB
t F
SB
t
)
=
1 + ABK
1 + ASB
σ + ASB(σ − 1)
σ + ABK(σ − 1)
which is increasing in the ratio of productivities ABK/ASB.
Off-balance sheet resource constraint.
Given a fraction ξ, of funds F
BK
t
, with ξ ∈ (0, 1), we have:
F
BK
t = ξF BK
t + (1 − ξ)F
BK
t
=
Z ZBK
0
fBK,t(j)dj +
Z Z
0
fX,t(j)dj
266

by symmetry: f
BK
t
(j) = ξF BK
t
ZBK
f
X
t
(j) = (1 − ξ)F
BK
t
Z
Off-balance sheet solution:
max
{fX
t
(j)}
r
X
t
(j)ι
X
t
(j) − R
BK
t
f
X
t
(j)
sub



ι
X
t
(j) = f
X
t
(j)
r
X
t
(j) = p
BK
t

IX,t
IBK,t− 1
η

ιX,t
IX,t− 1
σ
gBK =
I
BK
t
F
BK
t
+ 1
=
1
F
BK
t
h
(
˜I
BK
t
)
η−1
η + (I
X
t
)
η−1
η
i η
η−1
=
1
F
BK
t
h
(ABK,tξFBK,t)
η−1
η + (AX,t(1 − ξ)FX,t)
η−1
η
i η
η−1
+ 1
=
h
(ξABK,t)
η−1
η + ((1 − ξ)AX,t)
η−1
η
i η
η−1
+ 1
ξ is endogenous and can be computed from the optimal pricing conditions: RBK
t =
σ−1
σ
r
X
t
(j), and RBK
t =
σ−1
σ
r
BK
t
(j)(1 − s). Divide the two expression in order to find the
optimal expression for ξ:
r
BK
t
(j)
r
X
t
(j)
=

ιBK,t
ιX,t − 1
σ

˜I
BK
t
I
X
t
!− 1
η + 1
σ
which leads to: ξ
∗ =
1
1 + (1 − s)
−η

AX,t
ABK,tη−1
.
267

Table C.1: Gross real and nominal growth rates for each sector of the economy
Consumption / Wages Real Investments
Real gC = (β(1 + AK − δ))α = gw gK = β(1 + AK − δ)
Nominal gCˆ = gC ; gwˆ = gw gKˆ = g
α
K
Banking Shadow Banking
Real gIBK = 1 + (1 − s)ABK = gFBK gISB = 1 + ASB = gFSB
Nominal gIˆBK
=
(1 + (1 − s)ABK)(β(AK + 1 − δ))α
β

1 + ν
1+ν ABK gIˆSB
=
(1 + ASB)(β(AK + 1 − δ))α
β

1 + σ−1
σ ASB
C.3 Additional empirical evidence
Figure C.2: Proportion of bank loans relative to the total amount of loans issued by financial
institutions (in blue), and shadow banks (in red) in the United States over the period 1960q1 –
2019q4
Notes: Banks loans stand for the total on-balance sheet loans of U.S. chartered depository institutions. Shadow Banking total loans is obtained as the sum of all the major non-bank financial institutions. See Table C.3 in the chapter’s
appendix. Total Financial Institutions loans stand for the fraction of loans issued by U.S.-chartered banks out of the total amount issued by the domestic financial sectors. All data are at quarterly frequency, and deflated by GDP implicit
price deflator.
26

Table C.2: Effect of funding competition on deregulation in the United States over the period 1980q1-2008q4
Deregulation index (∆Dt+4)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Funding competition 0.734*** 0.677** 0.941** 0.894** 0.910** 0.857* 0.835** 0.824** 0.849** 0.794**
(∆xt−1) (0.171) (0.268) (0.439) (0.413) (0.445) (0.443) (0.354) (0.354) (0.358) (0.331)
Fund competition – LT effect 1.108*** 1.012** 1.382** 1.328** 1.350** 1.276* 1.092*** 1.068*** 1.105*** 1.016***
β/(1 − θ) (0.269) (0.432) (0.658) (0.629) (0.660) (0.656) (0.397) (0.388) (0.405) (0.369)
Deregulation index (∆Dt) 0.337*** 0.331*** 0.319*** 0.327*** 0.326*** 0.328*** 0.235** 0.229* 0.231** 0.218**
(0.089) (0.089) (0.088) (0.086) (0.091) (0.090) (0.113) (0.119) (0.116) (0.106)
Inflation (∆pt) 1.651 0.809 1.033 -0.081 -0.990 2.351 2.505 3.011 3.055
(4.228) (4.645) (4.688) (6.603) (6.877) (4.365) (4.231) (4.247) (3.669)
Output growth (∆yt) 5.194 4.282 3.985 2.240 3.375 3.003 2.853 6.598
(4.195) (3.727) (3.330) (3.382) (3.765) (3.724) (3.500) (4.582)
US real equity price (∆eqUS t ) 0.322 0.324 0.431 1.417* 1.104* 1.090* 0.832
(0.230) (0.241) (0.262) (0.816) (0.610) (0.594) (0.514)
Oil price (∆poil t ) 0.111 0.101 0.221 0.225 0.270 0.337
(0.230) (0.213) (0.264) (0.264) (0.274) (0.303)
US long term i rate (∆lrUS t ) 20.020 27.697 24.648 60.942 61.598
(14.411) (18.067) (15.128) (39.074) (38.008)
Global real volatility (grvt) 2.962 3.202 3.217 3.056
(1.907) (2.099) (2.060) (1.942)
Global real equity price (∆eqUS t ) 0.710 0.699 1.138
(0.718) (0.729) (0.908)
Global long term i rate (∆glrt) -1.190 -1.318
(1.020) (1.015)
Global output growth (∆y∗t ) -10.758*
(6.347)
R-squared 0.192 0.194 0.220 0.229 0.233 0.243 0.335 0.339 0.347 0.380
Notes: ‘Funding competition’ is the ratio between U.S. chartered depository institutions checking, savings, and time deposits and the total amount of outstanding
MMMF shares. The deregulation index is taken as absolute difference with respect to four quarters before. Long term interest rates are quarterly absolute first
difference. All other variables are first difference of the natural logarithm one quarter before. The deregulation quarterly variables are obtained by linearly interpolating the original annual indicator. The long term effect of funding competition is computed from the specification of an ARDL with partial adjustment. Errors
are corrected following Newey-West. All data are quarterly. Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Sources: U.S. deposits and MMMF
shares: Financial Account of the Fed, Z.1 tables (quarterly frequency). Deregulation index: Philippon and Reshef (2012). Output growth: U.S. Bureau of Economic
Analysis. CPI: OECD accessed through FRED. Oil price: West Texas Intermediate, Federal Reserve of St. Louis. U.S. real equity price, U.S. long term i rate, and
global variables: GVAR dataset, Mohaddes and Raissi (2020).
269

C.4 Data sources
Data on financial institutions are taken from the Federal Reserve – Financial Accounts.1
In terms of variables creation, All the major non-bank or non-depository financial institutions are included under the definition of shadow banking sector: money market funds,
mutual funds, closed-end funds, exchange-traded Funds (ETFs), Government-sponsored
enterprises, agency and GSE-backed mortgage pools, special purpose vehicles (SPV),
finance companies, real estate investment trusts (REIT), security brokers and dealers,
property-casualty insurance companies, life insurance companies, private pension funds.
Table C.3 provides the references for the actual Financial accounts series to be retrieved.
Data on Off-balance sheet operations are taken from the Enhanced Financial Accounts of
the Fed.
“U.S. chartered depository institutions" comprehend mostly banks of different sort: national commercial banks, state-chartered commercial banks, federal savings banks, statechartered savings banks, cooperative banks, savings and loan associations, and international banking facilities (IBFs) established by U.S.-chartered depository institutions.
Private depository institutions adds to the previous category: credit unions, foreign
banking offices in the U.S., banks in U.S.-affiliated areas to the previous category of U.S.
chartered.
All series are deflated by using the GDP Implicit price deflator from OECD – Main Economic Indicators, and retrieved from FRED - Federal Reserve of St. Louis. Table name: USAGDPDEFQISMEI. Output data: U.S. Bureau of Economic Analysis retrieved from FRED
database.
The deregulation index is taken from Philippon and Reshef (2012) and made available
on the website of Professor Philippon. The series is extended for four years from 2006
to 2010q2 in order to provide a longer time span. After 2010q2, a more accurate analysis
1What used to be called Flow of Funds.
270

would be needed in order to account for the Dodd-Frank Act to be in place.
271

Table C.3: Data sources utilized in both the descriptive statistics and empirical analysis
Series name Series code
Private Depository total liabilities + equity FL704194005Q
U.S. chartered Checking deposits FL763127005Q
U.S. chartered total Time and Savings deposits FL763130005Q
SBS - Property causality insurance companies FL514090005Q
SBS - Life Insurance total assets FL544090005Q
SBS - Private Pension Funds total assets FL574090005Q
SBS - MMMF total liabilities MMMFFAQ027S
SBS - Mutual Funds total assets LM654090000Q
SBS - Closed-End Funds total assets LM554090005Q
SBS - ETF total assets LM564090005Q
SBS - GSEs total assets FL404090005Q
SBS - Agency and GSE-Backed mortgage pools AGSEBMPTCMAHDFS
SBS - Issuers of ABS total assets FL674090005Q
SBS - Financial Companies total assets FL614090005Q
SBS - Real Estate Investment Trusts total assets FL644090005Q
SBS - Security Brokers and Dealers total assets FL664090005Q
Loans - Total financial sector FBLA
Loans - U.S.-Chartered depository institutions USCDILA
Loans - Mutual Funds MFSLNCBA
Loans - GSEs GSELA
Loans - Agency and GSE-Backed mortgage pools AGSEBMPTCMAHDFS
Loans - Issuers of ABS IABSLA
Loans - Finance Companies FCLA
Loans - Real Estate Investment Trusts REITTMA
Loans - Security Brokers & Dealers SBDOLAA
Loans - Property-Casualty Insurance companies FL513065505Q
Loans - Life Insurance companies LICLA
Loans - Private Pension Funds PPFTMA
Notes: All resources refer to the U.S. Federal Reserve - Financial Account Z.1 tables. Data in levels at quarterly frequency.
272

Abstract (if available)

Abstract This dissertation on macroeconomics and macro-finance is composed of three chapters. The first chapter is "Identifying the Effects of Sanctions on the Iranian Economy using Newspaper Coverage". It contributes to both the time series and the sanctions literature by proposing a new identification and measurement strategy to study the effect of international sanctions on the Iranian economy over the period 1989-2019.

The second chapter is “Inequality and the Rise of Finance”. It studies the causes behind the rise of the financial sector observed in the United States from the 1980s. The theory proposed claims that the growth of the financial sector can be seen as an endogenous rise of non-bank financial institutions reacting to secular macroeconomic forces. An exogenous variation of the labor share generates an endogenous rise of wealth inequality. In turn, this leads to a larger demand for safe assets, a compression of real interest rates, and the entry of other financial intermediaries.

The third chapter is "The Political Economy of Banking and Shadow Banking Competition". It takes the larger inflow of funds towards the financial sector as given, and it provides a theory for the emergence of financial deregulation from the 1980s. When other non-bank financial institutions are able to attract the funds that investors demand by exploiting a regulatory and technological competitive advantage, traditional banks are predicted to lobby harder on policymakers to level the playing field. The model also predicts an endogenous rise of financial innovation to stay ahead of the competition.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Three essays in international macroeconomics and finance

PDF

Essays in macroeconomics

PDF

Growth and development in Africa: macroeconomic and microeconomic perspectives

PDF

Growth, trade and structural change in low income industrializing economies

PDF

Essays on business cycle volatility and global trade

PDF

Essays on Japanese macroeconomy

PDF

Three essays on macro and labor finance

PDF

Essays on macroeconomics and income distribution

PDF

Essays on firm investment, innovation and productivity

PDF

Two essays on financial crisis and banking

PDF

Essays in uncertainty and aggregate economic activity

PDF

Essays in international economics

PDF

The political economy of R&D collaboration: micro- and macro-level implications

PDF

Inter-temporal allocation of human capital and economic performance

PDF

Essays on interest rate determination in open economies

PDF

Essays on government intervention in financial crises

PDF

Three essays on heterogeneous responses to macroeconomic shocks

PDF

Essays on sovereign debt

PDF

Early-warning systems for crisis risk

PDF

Essays in corporate finance

Asset Metadata

Creator Laudati, Dario (author)

Core Title Essays in macroeconomics and macro-finance

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Economics

Degree Conferral Date 2024-05

Publication Date 04/15/2024

Defense Date 03/26/2024

Publisher Los Angeles, California (original), University of Southern California (original), University of Southern California. Libraries (digital)

Tag banking,deregulation,finance,Inequality,Iran,macroeconomics,macro-finance,OAI-PMH Harvest,Political Economy,safe assets,sanctions

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Quadrini, Vincenzo (committee chair), Betts, Caroline Marie (committee member), Kurlat, Pablo Daniel (committee member), Li, Wenhao (committee member), Rancière, Romain (committee member)

Creator Email dario.laudati1@gmail.com,laudati@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC113872125

Unique identifier UC113872125

Identifier etd-LaudatiDar-12810.pdf (filename)

Legacy Identifier etd-LaudatiDar-12810

Document Type Dissertation

Format theses (aat)

Rights Laudati, Dario

Internet Media Type application/pdf

Type texts

Source 20240415-usctheses-batch-1140 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu