University of Southern California Dissertations and Theses

Context is everything: An empirical study of institutions, instability and economic growth

(USC Thesis Other)

Context is everything: An empirical study of institutions, instability and economic growth

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFO RM ATION TO U SE R S
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may be
from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Bell & Howell Information Company
300 North Zeeb Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/521-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CONTEXT IS EVERYTHING:
AN EMPIRICAL STUDY OF INSTITUTIONS, INSTABILITY
AND ECONOMIC GROWTH
by
Nauro Ferreira Campos
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Political Economy and Public Policy)
December 1997
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 9 835119
UMI Microform 9835119
Copyright 1998, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES. CALIFORNIA 90007
This dissertation, written by
N (\\J R . 0 C
under the direction of h.l.%...... Dissertation
Committee, and approved by all its members,
has been presented to and accepted by The
Graduate School, in partial fulfillment of re
quirements for the degree of
DOCTOR OF PHILOSOPHY
/ Date
DISSERTATION COMMITTEE
^ ’ ’ ’ " ........................................ Chairperson
.........................
............
/fS fw A ' J Z . / Z c i - —
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
To my father.
ii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
"We cannot see, feel, touch, or even measure
institutions; they are constructs of the hum an mind."
Douglass C. North, 1990.
"Yes, but if 'social capital' is to be more than a
buzzword, something more than mere relevance or
even importance is required. (...) There needs to be an
identifiable process of 'investment' that adds to the
stock, and possibly a process of 'depreciation' that
subtracts from it. The stock of social capital should
somehow be measurable, even inexactly."
Robert M. Solow, 1995.
iii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ACKNOWLEDGMENTS
I have had the honor to work closely with two unique scholars, Professors
Jeffrey Nugent and Richard Easterlin. W ithout their friendship and support, this
dissertation never would have been written.
Jeffrey Nugent's encouragement from my first year of graduate school,
and the refuge provided by his development economics classes, have been
greatly appreciated. He was largely responsible for pushing me in the direction
of institutions, for which I will be eternally grateful. I thank him for his guidance,
constant support, and invaluable suggestions at various stages of this research.
Over the years, I have had the pleasure and the privilege of working
closely with Richard Easterlin, and continue to marvel at his breadth and insight.
He supported me in my choice of a time consuming and risky dissertation,
constantly challenged my abilities and assumptions, and helped me make the
im portant connections. His enthusiasm for the dissertation and its relevance
never waned, and bolstered my own.
University regulations should prohibit anybody to defend a Ph.D.
dissertation on institutions or Latin America if Abraham Lowenthal is not in the
committee. He has been an endless source of advice and encouragement.
I would also like to thank David Lopez-Lee. The first prospectus of this
dissertation was written for his research methodology class. He has been a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
careful reader and editor, demonstrating an eye for detail but always conscious
of the broader implications.
I thank James Robinson for the always candid advice and encouragement.
W ithout those long conversations over short coffees, this dissertation would
have been much less.
I would also like to thank my old advisors and supervisors. From Tokyo
and from Baltimore, Joao Carlos Ferraz and Stuart Leslie sent comments and
words of encouragement. Pablo Guerrero and Eduardo Wiesner, from
Washington, showed me that institutions are much more than just constructs of
the hum an mind.
I would also like to thank Joao Campari, Jeff Cohen, Yael Duthilleul,
Gaston Gelos, Giselle, Kastello, Feisal Khan, Vic Saddi, Joe Schumpeter, Jenny
Tessendorf, and Cadu Young, for their friendship and support throughout. I
wish everybody was this lucky.
Last, but far from least, without Cristina nothing would have been
possible, plausible, or worthwhile.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLE OF CONTENTS
Chapter 1. When the Devlops Meet the Ninsts 1
1. Forgetting macro 5
2. Defining institutions 8
3. Identification issues: which institutions? 9
4. How institutions affect economic development? 12
5. Pessimism as sin 16
6. The absence of politics 19
7. Institutions with a human face 21
8. Where are we and where are we going? 23
9. Organization 24
Chapter 2. A Survey of the Evidence on Institutions and Development
(When Both are Defined Broadly) 26
1. Difficulties in cross-country studies 28
2. Problems in measuring institutions 31
3. The evidence 35
4. Conclusion 64
Chapter 3. An Empirical Analysis of the Relationship between
Institutional Development and Economic Growth 67
1. A Comparative Index of Institutional Development (CIID) 69
2 .The impact of institutions on economic growth 75
3. What explains institutional development? 85
4. Conclusions and suggestions for future research 91
Data appendix to Chapter 3 105
Chapter 4. On the Causal Relationship between Political Instability,
Institutional Development and Economic Growth 114
1. Methodological and data issues 118
2. The pattern of causality 124
3. The relationship between SPI and economic growth 129
4. Conclusions and suggestions for future research 133
Bibliography 154
vi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ABSTRACT
The objective of this dissertation is to evaluate, empirically, the
relationship between institutional development, socio-political instability, and
per capita economic growth. The study focus on the experience of the
developing countries between 1960 and 1990.
This study differs from others on three basic aspects. First, while the
relevant empirical literature does not seem concerned with assessing the role of
institutions from within a well-established theoretical framework, in this study I
carefully enlarge the standard economic growth model to accommodate my own
index of institutional development. Second, I am concerned here not only with
the effects of institutions but also with its determinants. Third, this study
approaches political instability and institutional development as two intrinsically
related phenomena.
I reach three main conclusions. First, I find that my index of institutional
development is robust, positive and statistically significant when incorporated
into the standard economic growth model. Second, I identify inequalities in the
distribution of wealth and policy distortions as the main determ inants of
institutional development (in Latin America), even after controlling for initial
conditions. Third, I conclude that socio-political instability (as usually measured
in the literature) has no long-term causal effect on economic growth, except in
the Sub-Saharan Africa countries.
vii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 1.
When the Devlops Meet the Ninsts1
Exactly ten years ago, Professor Paul Streeten wrote that structural
adjustment is w hat economic development is all about. If I have to paraphrase
him, I would say that transition is what economic development is all about.
In reality, the process of economic development is a collection of different,
although not always distinct, transitions. Among them we find: the transition
from an economy characterized by a dominant subsistence sector, to one with a
diversified production and employment structure; from an economy with low
savings and investm ent rates, to one with high domestic savings and investment;
from one economy plagued by incomplete and disconnected markets, to one
with well-functioning, competitive and integrated product and factor markets;
from one economy defined by technological backwardness, to one extensively
using appropriate techniques; from an illiterate and unskilled labor force, to an
economy with mobile and skilled labor; from a population showing high fertility
and mortality rates, to one which these rates have been significantly lowered;
The title alludes to a 1973 article by Professort Axel Leijonhufvud. "Life Among the Econ”. where he
meticulously describes the hierarchy and day-to-day interaction am ong the different econom ic tribes. In
this article, the developm ent econom ists' tribe is called Devlops. Here, the new institutional economists
tribes is called Ninsts.
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
from one economy where international trade is limited, to one where it is
diversified. Naturally, this list is not exhaustive.
Two other transitions are of particular relevance for our discussion. The
first is the transition from a society with a concentrated distribution of income
and wealth, to one where the national product is distributed increasingly
according to effort, m erit and talent.
The second is the transition from an ineffective to an effective State. In this
case, effectiveness is a function of how well the needs, tastes, and preferences of
the population are embodied, translated and reflected into the functioning of the
government, into the organization of the state itself, and into the formulation and
implementation of its policies. From the development theory point of view, a
state that is not representative is not effective.2
This State transition is to be observed in the process of consolidation of the
economy's institutional framework. At the beginning, there was light, but also
there was social and political instability. This early instability suggests an
"evolutionary" process of trial and error from which preferred institutional
forms will be selected, and remain for relatively long periods of time.
: Bardhan argues that "if one takes a broader concept of development to incorporate general well
being of the population at large, including some basic civil and political freedoms, a democracy
which ensures these freedoms is, almost by definition, more conducive to development on these
counts than a non-democratic regime" (1993, p. 47.) Another source advancing this position states
that "Political choices, embodied in constitutions, laws, policies, and conventions, set the rules of
the game and empower their arbiters. The making and implementation of these choices is the
labor of democracy. If these choices are not contested and adjusted over time, or if they are
subverted by corruption, innovation and hence development grind to a halt in favor of powerful
and protected interests opposed to change" (IRIS, 1996.)
2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
These two transitions, of inequality and of the State, are of great relevance
for both the economics of development and the economics of institutions.’ Their
relevance can be exemplified by the African tragedy, by the recent difficulties of
most former communist countries, by the Latin American "empty box"
syndrome, and by the imperfect ethnic accommodation of Asian countries.
Although I do not think I have to remind the reader that these are
generalizations, some specific examples should help clarify. In the case of Sub-
Saharan Africa, the extent and intensity of social and political instability (a
consequence, as it is, of the imperfect ethnic accommodation in national
territories demarcated by European metropolises) has impeded the flourishing of
productive activities, and thus, economic growth.4 The former communist
countries are states in search of an institutional framework. Informal rules, habits
and customs, do indeed change slowly.’ Fanzylber's empty box in Latin America
refers to that if we construct a matrix with rows for high and low economic
growth and columns for equal and unequal distribution of income, and
distribute Latin nations accordingly, the only empty cell we find is that with high
growth and equal distribution.0 As for the Asian countries, think of India: an
economy growing only modestly with a sophisticated state that has been
’ Economics of institutions and new institutional economics (NIE) are terms here used
interchangeably. They refer to the works of, among others, Nobel pnze winner economists
Ronald Coase, James Buchanan and Douglass North. See Rutherford (1994) for a thematic
overview of their work, as well as a comparison with the “old" institutional economics.
See Levine and Easterly (1996), and Canova and Bertocchi (1996.)
See Shleifer (1995), McMillan (1997) and Dewatripont and Roland (1997).
See Fanzylber (1990).
3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
successful in isolating itself from ancestral caste divisions. It is a democracy that
can hardly be called representative.
As for the field of development economics, the importance of these two
crucial transitions (inequality and State) is relatively well understood. W hat
seems to be lacking is a fuller appreciation that is precisely upon these transitions
that a bridge can be built towards the economics of institutions.
The objective of this chapter is, therefore, to discuss the problems and
limitations I perceive in the economics of institutions. I argue that the recognition
and attempted solution of these problems is of great importance to the good old
development theory (with its macro, empirical and normative emphases.)
These problems, I offer, are the following: (1) The inattention to issues at
the macro or national level of analysis. (2) The problems in defining and
operationalizing the notion of "institutions." (3) The problems of identification:
which institutions affect the process of development? (4) And very much related,
how do these institutions affect the process of development? (5) The excessive
pessimism. (6) The insistence on investment in physical capital and the disregard
for the possibility that the effects of institutions on development may take root
by way of hum an capital. (7) The absence of politics, as reflected in the
inattention to characteristics of political regimes and in the reluctance to produce
an argument that approaches social and political instability as an indication of
institutional immaturity.
See Bardhan (1996) for a discussion.
4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A note of caution. The first problem, NTE's inattention to issues at the
macro or national level of analysis, serves a double purpose here: it is also the
major motivation for this dissertation work. The feature of the economics of
institutions particularly problematic from a developm ent viewpoint is the lack of
empirical, normative w ork at the macro level.5 The problems discussed below all
depend, for their validity, on this fundamental caveat.
1. Forgetting macro
The point I w ould like to make in this section is simply that the lack of
empirical, normative w ork at the macro level in the economics of institutions is
widely recognized as a limitation by new institutionalists themselves. One of the
prominent voices in this literature has noted recently:
Furthermore, in its current stage the strength of the new theory is the
studv of outcomes and structures at the m arket or micro level, and the
studies of life within organizations at the so-called micro-micro level.
There have been few attempts to merge macro-economics and the
economics of institutions and develop a theory of macro-economic
relationships centered on transaction costs and allowing for variations in
institutional arrangem ents (Eggertsson, 1994, p.20).
A related argument is made by Nugent, who argues that if NIE is to have a lasting effect on
the economics of development, "systematic empirical analyses will have to be accomplished"
(1997, p.8). Although he also identifies seven "problems", his analysis is not restricted (as ours is)
to macro level issues. As a consequence, his list includes elements such as "interdependencies
between different institutions" and "heterogeneity of theoretical tools due to the multilevel
character of institutions."
a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This is to say that the economics of institutions has been very valuable when
focusing on im portant microeconomic issues9 , but too timid and, consequently,
has attained only limited success when dealing with the national (or macro)
economic performance of LDCs. Some of the possible reasons behind this are
explored under the other "problems" discussed below.
From a development perspective, the lack of normative implications is
discouraging. For example, in the preface to a recent collection of the best
empirical research from the economics of institutions, the editors advert that
"[T]he scope of the series is comparative and historical rather than international
or specifically American, and the focus is positive rather than normative."
(Alston, Eggertsson, and North, 1996, p. v). An edited volume on the best of the
empirical research, that is comparative, international and norm ative in scope, I
believe, will have to wait.1 0
The lack of macro level, normative, empirical research is widely and
openly recognized. What does this mean, in general? It means that we are not
able to answer straightforward questions, for example, like: Is the United States
today more institutionally developed (in any sense of the w ord) than, say,
Mozambique? And what does it mean to the good old developm ent theory? It
According to Bardhan, "In recent years two strands of non-Walrasian economic literature have
developed well-articulated endogenous theories of institutions, and they are both getting to be
prominent in the new microeconomics of development" (1989, p. 1390). See Lin and Nugent
(1995) for a comprehensive review of this micro-oriented literature.
1 0 On the absence of normative concerns, Alston notes that "the research on institutions by many
social scientists is either highly descriptive or so abstract as to render it useless for policy" (1996,
p.25).
6
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
means that there is a scarcity of results that development economists tend to
value. In particular, questions like "Which institutions should we change if we
are to prom ote development ?" remain very much unanswered.1 1
The problems discussed below provide an overview and, hopefully, an
explanation for these difficulties. First, the debates on the precise definition of
"institution" have been endless. Because there is no agreement on w hat is and
what is not an institution, it becomes virtually impossible to pinpoint which ones
have an effect on the process of economic development. Moreover, the state is
more often than not seen as the house of all evils. Yet, it is clearly the most
convenient item to launch a comparative and inter-national research program
upon. Nations w ithout states are an oddity: the state is an institution (or a group
of institutions) that is common to all and thus invites, quite naturally,
comparative analysis (recall that only China has TVEs). This reluctance has
som ewhat hindered a full acceptance of NIE ideas (that is, by development
economists.) In particular, because by not pursuing this macro, state-centered,
empirical line of inquiry results in ignoring altogether a num ber of very
im portant issues, like the neglect that institutions may have a "hum an (capital)
face" and that social and political instability can be conceptualized as an
indication of institutional immaturity. Now that the "problems" are reasonably
linked, let us turn to each one of them.
" Leave alone the subsequent question, "how can we change these development-promoting
institutions?"
7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2. Defining institutions
One first difficulty is in terms of the definition of "institution."1 2 In order
to differentiate the m any available definitions it may be useful to consider the
distinction between institutions and organizations. This distinction is the basis
for North's extremely influential work. He sees institutions as being completely
separated from organizations. Institutions, he says, are "the rules of the gam e"
in a society, and they are composed of formal rules, of informal constraints and
of their enforcement characteristics (North, 1997.) Ij Further, certainly referring to
the informal aspects, he notes that "we cannot see, feel, touch, or even m easure
institutions; they are constructs of the human m ind" (1990, p.107.) Organizations
are very different, they "are the players: groups of individuals bound by a
common purpose to achieve objectives" (1995, p. 23.)
For empirical w ork (at the macro level), this distinction is too rigid.1 4 This
has been increasingly recognized by applied researchers. For example, a recent
report on how to ameliorate the problem of corruption in Africa, argues that:
See Menard (1995) and references therein.
1 3 For a different view, see Schotter who argues that "we care about what the agents do with the
rules of the game, not what the rules are" (1981, p.155).
1 4 There are many who believe that this separation is exaggerated. For instance, Harris, Hunter
and Lewis point out that "it is the focus upon the interaction between institutions and
organizations in some of the most recent NIE writing that makes the NIE approach specially
relevant for students of long-run Third World development" (1995, p.6, emphasis added).
8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Corruption, then, is a failure of governance. What do we mean by
"governance?" It refers to the fact that a people manages its affairs
according to its own norms of fairness and legitimacy, embodied in laws,
procedures, policies, conventions, and organizations designed to carry
them out. One could refer to all of these as "institutions." The quality of
these institutions is an im portant determinant of both economic and
political development. (IRIS, 1996)
We adopt the broad concept of institutions as encompassing both the
rules of the game and its players. This is the direction we follow below. At the
macro level, de-emphasizing this distinction seems to speak to the heart of this
issue because, as Adelman and Morris note, "institutions m atter greatly because
they determine which government policies are likely to be adopted, and which
institutions will be strengthened, introduced or weakened"(1989, p.1429).
3. Identification issues: Which institutions?
Aside from the issue of a proper definition, there is also the m atter of the
identification of which institutions affect the process of developm ent and how.1 3
Below we turn to the "list of key institutions" offered by various important
authors.
According to North, "the inability of societies to develop effective, low-
cost enforcement of contracts is the most important source of both historical
stagnation and contemporary underdevelopment in the Third W orld" (1990,
Bardhan (1996) discusses this problem at length.
9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
p.54). Olson stresses "different legal and organizational arrangements and
economic policies" (1996), and Reynolds points out that "the single m ost
important explanatory variable is political organization and the adm inistrative
competence of government" (1983, p.976).
Easterlin argues that "the spread of m odem economic growth has been
associated with favorable institutional conditions", and that "... everywhere
establishment of the rule of law, enforcement of contracts, political stability, and
elimination of arbitrary seizure or taxation of property by despots or others was
essential" (1996, p. 56). Further, his argum ent for the inclusion of formal
universal schooling as an important item in the list of explanatory variables is an
important contribution.
Abramovitz and David put forward "social capability" as the factor that
limit a country's ability to catch-up. It is a concept that
has to do with those attributes, qualities, and character of people and
economic organization that originate in social and political institutions
and that influence the responses of people to economic opportunity. It
includes a society's culture and the priority it assigns to economic
attainment. It covers the economic constitutions under which people
live, particularly the rights, limitations, and obligations involved with
property, and all incentive and inhibitions that these may create for
effort, investment, enterprise, and innovation. It involves those long
term policies that govern particular forms of organization or activity ... ,
and the policies that may support or restrict such organizations. And it
covers the policies that provide for the public provision of social services
and those that support the accumulation of capital by investment in
infrastructure and by public education and research (1996, p. 50-51).
10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Sachs and W arner argue that economic growth and the convergence of
levels of per capita income requires “reasonably efficient economic institutions.
Poorly managed economies -such as those with the absence of secure property
rights, autarkic trade policies, inconvertible currencies, and so forth — are
unlikely to experience convergence no m atter the underlying production
function or initial level of human capital" (1995, p.5).
And finally, Solow argues that “even the merest beginning of Eastern
Europe's transform ation has made unmistakably clear how much economic
development depends on the presence of the institutional and attitudinal
infrastructure of a m odem capitalist economy. This includes such things as a
worked-out and generally accepted framework of property rights, enforcement
of contracts, and a whole slew of m arket institutions, including financial
institutions that can gather savings, evaluate loan quality, and control risk."
(1997, pp. 71-72, emphasis added).
There seems to be a consensus on the need for clearly defined and
enforced property rights. It should be emphasized that good institutions do not
arise as manna from heaven. Indeed, as noted by North, "it is the polity that
defines and enforces property rights" (1995, p.20). Therefore, it would seem
appropriate that political economy considerations be allowed to play a greater
role in explaining the macro level determ inants and effects of institutions.1 0
1 0 As Bates notes: "Taking political factors into account thus provides explanation for the
direction and magnitude of the departures from the status quo that economic institutions make
possible and yields insights into the source of variability in their performance. The new
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Doing so, would also rather naturally lead the study of institutions in a more
normative direction: if the institutions that m atter can also be changed, more
attention should be devoted to how best to encourage the development of
growth-enhancing institutions.
4. How institutions affect economic development?
Institutions affect the process of economic development in many ways: by
ameliorating the free rider problem (Olson, 1965), reducing transaction costs
(Coase, 1960, and Williamson, 1985), and curtailing rent-seeking (Krueger, 1974).
Also, by solving coordination problems institutions contribute to reducing the
overall level of uncertainty in the economy (Schotter, 1981.) The latter is
particularly important for our purposes.
Coordination problems arise when there are gains to all "players" from
coordinating their choices of strategies, and in doing so, avoiding a mutually
inferior outcome. The classic example is the choice of which side of the road to
drive. It does not matter which side is chosen (right or left), as much as that all
players "choose" (and effectively drive in) the same side. Social conventions,
rules and (traffic) laws play a role in solving this problem. Although uncertainty
is not completely eliminated, it is certainly reduced.
institutionalism originates in economics. To fulfill its agenda, however, it must move into the
study of politics. It needs to take into account the allocation of power in society and the impact of
the political system on the structure and performance of economic institutions" (1995, p.44).
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Four our purposes, the analogies draw by Bardhan are even more
appropriate. He discusses the well-known proposition that under competitive
market conditions it does not matter whether capital hires labor or labor hires
capital: "who hires whom essentially depends on the capacity to be the residual
claimant in production, and that in turn depends on the capacity to bear risks ..."
(1989, pp.1393-94.) One key word here is "residual": institutions affect
development because in reducing risks and uncertainty, they help determine
(together with technology) the actual rates of return. The other key word here is
"claimant": institutions affect development because they help decide who has a
legitimate claim on the "residual" and who does not. I believe this is precisely
what So low had in mind when he spells out that the commonly found "well
defined" adjective before "framework of property rights" is to be understood as
"worked-out and generally accepted" (1997, p. 71.)
Institutions affect development in raising representativeness. The most
important quality in any given institutional framework, vis-a-vis promoting
economic development, is the degree to which the tastes, needs and preferences
of the population are translated into the functioning of the government, the
organization of the state, and the formulation and implementation of public
policies.
Notice that this point of view calls attention to both the input and output
(or outcome) sides of the institutional development process. According to this
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
identification, the mark of a successful process of institutional development does
not reside solely on the optimal extent of state intervention, or on its ability to
credibly pre-commit to policy decisions (Rodrik, 1992), or on the efficient
provision of public goods. Instead, it goes beyond these by requiring the state
and other political coalitions to have "an 'encompassing interest' in economic
performance of the country as a whole" (Bardhan, 1996, p .ll). By the same token,
it considers as im portant not only the security of property and contractual rights,
but also the methods and means by which a society assigns or distributes these
rights (Shleifer, 1995). Finally, this concept of institutional development gives no
more weight to efficiency in the provision of public goods, as to which and how
many public goods are provided and to the way in which society arrives at this
choice.
Among the advantages of this point of view are the fact that it can be
empirically implemented, and that it is flexible enough to accommodate the
different concerns of institutional economics.
Although certainly not the only way of identifying institutional
development, our approach points to the need to examine the possibility that
institutions that are democratic and transparent would significantly improve
macroeconomic performance in the long-run. Institutions m atter when they
provide a voice.1 '
1 The concept of “voice" we have in mind is, of course, that of Hirschman. The relationship
between "representativeness" and economic growth is formalized in Boone (1995).
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Differences in educational systems offer a handful of examples. Extending
educational opportunities to a broader population basis, if organized according
to the principles above, should increase the competitiveness of advancem ent
through the educational system and, eventually, improve the functioning of
labor markets. This policy choice assures that only the best and the brightest, in
scholastic performance, will proceed. In other words, education should be
competitive and merit-driven, and not determined by family wealth, ethnic
origin, or other considerations. The point can be illustrated by contrasting the
experiences of South Korea and Venezuela (World Bank, 1993, p .199). Public
expenditure in education in 1985 was 4.3 percent of GNP in Venezuela, but only
3 percent of GNP in South Korea. The crucial difference, however, w as the share
of the education budget allocated to prim ary education: while in South Korea,
prim ary education received 83.9 percent (2.5 percent of GNP), in Venezuela it
received only 31 percent (1.3 of GNP). The observed outcome in Venezuela, and
indeed in most of Latin America, is a mixture of high levels of illiteracy,
inefficiencies in that students spend more years in school to obtain a degree, and
the perpetuation of inequalities by over-spending in higher education (Londono,
1995)'8 .
1 8 The literature on macroeconomics of populism in Latin America provides vet another
perspective. For instance, Cardoso and Helwege argue that “Far from providing an indictment of
redistribution efforts, the history of populism makes conspicuous the paucity of genuine
redistribution programs in Latin America. (...) Classical populists distributed the gains from
growth among the politically enfranchised" (1991, pp.59-61).
15
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Finally, stressing this quality of institutions suggests the need for an
analytical framework in which it should play a role similar to that of
technological progress (i.e., the efficiency parameter in a production function). It
should work in a similar m anner in that it can be both "disem bodied" or
"embodied" in the accumulation of the factors of production. We just
exemplified how institutional developm ent could be embedded in the formal
educational system, the main locus of hum an capital's production1 * . How it
affects the accumulation of physical capital is even more straightforward
(property rights). One can also imagine a similar relation in the case of capital
markets and financial systems in that, if they exist, their influence on
development might depend on the extent to which they would allow different
kinds of firms, both small and large, take advantage of investm ent
opportunities2 0 .
5. Pessimism as sin
From the point of view of development economics, a source of
dissatisfaction with the general state of institutional economics is the pessimistic
'* Throughout this paper, we follow the tradition in economics of restricting the definition of
human capital to the outcomes of the formal education system. We recognize, however, that there
are other components (particularly, on-the-job training and health improvements), and that these
might be even more important in the developing countries' context. See Becker (1993) for a
classical treatment.
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
tone of much of the latter literature. Because of the importance of the notion of
"path dependency" and common references to "institutional impediments", the
NIE would seem to breed pessimism that countries can break out of their
institutional traps. On the other hand, since developm ent economics has in the
past made great strides by trying to find ways to flatten the Kuznets curve, to
escape poverty traps and various kinds of vicious circles, shouldn't development
economics push and stretch NIE so as to find ways to break path dependency
and to avoid the continuation of underdevelopment as an institutional failure?
Moreover, the way the empirical literature seems to read the notion of
"path dependency" is that institutions change only very slowly over time. A
problem with this reading is that it justifies an unfortunate research practice:
namely, the use of averages for different time lengths in the same regression
equation. For example, in trying to assess whether institutions are an important
determinant of observed cross-country differences in per capita GDP growth
rates, some studies regress the average of these rates between, say, 1960 and
1985, on an index of institutional development that is averaged between, say,
1980 and 1985.:! If institutions do not change, this is very reasonable. But what if
they do?
; o See Roubini and Sala-i-Martin (1992) and De Gregorio and Guidotti (1995) for empirical
evidence that highlights the negative contribution of financial development to economic growth.
:i In the World Development Report 1997. results based on regression analysis on panel data
from 94 developed and developing countries for 1964-1993 are used to conclude that "countries
with good economic policies and stronger institutional capabilities grow faster" (1997, p.13). The
dependent variable is the average, or long-term, growth rate of per capita GDP between 1964 and
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Interestingly, this alleged institutional resilience is not as common an
argument in the theoretical literature as many would like to believe. An
important distinction in Douglass N orth's work is that between formal and
informal constraints. He notes indeed that the informal constraints change very
slowly. These are beliefs, behavioral patterns, mental maps, and informal rules
that are not only constructs, but also inhabitants of the human mind. Few exist
outside. But North also points out that the formal rules can be changed
"overnight" (even if their enforcement characteristics presumably can not). Most
of the measurements we have available today refer, obviously, to the formal
rules.
The formal rules capacity to change brings us back to the "state
transition." In some comers of the economics of institutions, the state is
perceived as the source of all evils, as the ultimate enemy that has to be
minimized and squeezed in order to provide the maximum possible space for the
"market." My reservation at this point is that ideology seems to be playing too
large a role. If we believe there is a need for research work at the macro or
national level, that is empirical and normative, then this way of perceiving the
state will not help us much. The state is an institution (or a group of institutions)
1993. The quality of policy is measured by a "policy distortion index": a composite based on the
share of trade in GDP, currency overvaluation, and extent of divergence between local and
international prices (1997, p. 170). Institutional capability is proxied by an index of the quality of
the bureaucracy. As far as I know, the first year for which the data for the quality of the
bureaucracy is available is 1982, not 1964. Notice, however, that the documentation in the Report
is not detailed enough to allow a more judicious verdict at this time.
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that is common to all countries and thus invites, quite naturally, cross-country
comparative analysis.
6. The absence of politics
The pessimistic view of the state is one reason why — after three Nobel
Prizes—I believe we are still in need of research in the NIE tradition that is
empirical, at the macro level, and normative. As a consequence, a num ber of
research venues are still open and relatively under-explored. Two quite
im portant among these venues are socio-political instability, and political
regimes.
As for the relationship between socio-political instability and institutional
development, for those unfamiliar with these two literatures it has to come as a
surprise that they have very much evolved separately.” A possible (and perhaps
the simpler) connection is evaluated in Chapter 4, namely, that high levels of
political instability are associated with low levels of institutional development.
Nevertheless, we feel that a formulation on social and political instability as an
indication of institutional immaturity is feasible, relatively simple and will
generate some high returns. If for no other reason, because it will force us to pay
more attention to the often neglect specification of the duration of the relevant
“ However, some first and original insights to understand the relationship between SPI and
institutional development are found in Dhonte and Kapur (1997.)
19
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
effects. We have in m ind questions like: can the deleterious effects of political
instability on per capita GDP growth be found in the long-run? How long does it
take to a major "spike", in terms of political instability, to translate into reduced
growth rates?
The second issue that is relatively under-explored is the role of political
regimes. Although there is a rapidly growing literature on the economic
consequences of democracy, what is difficult to find there is the argument that
democracy is a part, or one element, of the relevant development-promoting
"institutions". What we have in mind here is an "institutional black box": our
first task is or should be to identify what are the different "elements" inside the
box, and then investigate whether there are distinct patterns of complementarity
and substitutability am ong them. This is clearly an ambitious idea, and 1 am
convinced we do not have all the elements in place today to pursue it (what I
have in mind are two needed elements: conceptual frameworks that can be
empirically implemented in a straightforward manner, and second, an amount of
data covering significant periods of time and national experiences, that is also
conceptually relevant.) On the other hand, a worrisome and distinctive trace of
the meager empirical literature on institutions and developm ent is the tendency
to compartmentalize. Rarely we find a discussion of the relationships between
the various different elements found "inside the black box." I think the way the
state is treated is one among the reasons for this neglect.
20
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7. Institutions with a human face
The subject of this section is, in my opinion, one of the simplest and at the
same time one of the most common omissions in the economics of institutions. I
argue that the literature has insisted on the "investm ent in physical capital"
channel, and has disregarded the equally plausible possibility that the effects of
institutions on economic development may take root by way of hum an capital.
I say simple because in theory, there is no reason to focus exclusively on
incentives to investment in physical capital. Indeed, one of North's most famous
parables is the one about piracy (1968.) Individuals compare the returns to the
acquisition of the skills necessary to be a good pirate with the returns to the
acquisition of skills necessary to be, say, a good shoe maker. Two centuries ago,
the returns to choosing being a pirate, adjusted for risk, were higher in many
countries. (I hope the countries with a higher share of shoe makers in the labor
force are someday found to have grown faster, or at least to have done so in a
more sustainable way.) Murphy et al. (1991) offer a similar (and up-dated)
argument: countries with large numbers of lawyers grow slower than countries
with large numbers of engineers.
One possible reason for this insistence on investm ent in physical capital —
and attendant disregard to the possibility that the effects of institutions on
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
economic development m ay take root by way of human capital - is the emphasis
on the well-defined property rights framework. Property rights m atter for assets,
and arguably, they matter even more for tangible assets. The accumulation of
physical capital to take place requires a minimum level of security, that the
returns to the investment are returning to the investor. The hum an capital story
is potentially a bit different: it appears "to be much more a m atter of quality than
of quantity. People acquire new skills rather than more of the old skills" (Solow,
1997, p. 77.)
Despite these difficulties with human capital, it is nevertheless surprising
that the focus of the few empirical studies has been almost exclusively on the
accumulation of physical capital. The data used to reflect the qualities of the
institutional framework are, more often than not, indicators that reflect the risks
and costs faced by a foreign investor. They reflect the probability of governm ents
repudiating contracts, the quality of their bureaucracies, the am ount of
corruption, and so on and so forth. The point is simply that this is too young a
literature to discriminate on the basis of convenience, instead of on the basis of
the issues that are thought to really matter. If from an pure economic growth
perspective, one can legitimately doubt whether the effort to incorporate hum an
capital is worthwhile, this is not a plausible doubt from the point of view of pure,
good old, development economics.
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8. Where are w e and where are we going?
One basic motivation for this study comes from a possible parallel
between the study of technological change and that of institutional development.
On the research agenda of economics, institutions today occupy a rather similar
position to the one technology did forty years ago. Although Abramovitz and
Solow were clearly not the first economists to emphasize its importance, they
pioneered in at least two fundamental ways. First, they courageously dismissed
the profession's belief that the topic should be better left to others, in their case,
to engineers. Second, they understood that without an explicit and cogent
attempt at quantification, there would be neither a marshaling of talent to
research the topic, nor any substantial progress. They knew the profession
needed some m easure of its ignorance.
After three Nobel prizes, it would be difficult to find today those who
believe (or w ould openly say) that institutions should be better left to others.
However, nothing like the impressive marshaling of talent working on the topic,
the profession's enthusiasm, and the sequence of major breakthroughs that
punctuated the study of technological change in the 1960s, are to be seen in the
case of institutions. Not only does the profession still seem to be looking for the
size of the residual or a measure of its ignorance, but also the links between
institutions and economic growth remain very much under-explored.
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I firmly believe that we should not accept at face value argum ents that
open with the notion that the effects of institutions on grow th and development
"are hardly controversial" (North, 1997.) We need, perhaps with some urgency,
to amass empirical evidence that these effects are clearly distinguishable within a
standard theoretical framework. To do so, there are crucial questions that need to
be addressed (I like to believe that some of these questions were at least properly
raised in the presentation above). This is quite a tall order and it will take time
and effort before we start to see the first breakthroughs.
9. Organization
The main objective of this study is to measure the level of institutional
development, across countries and over time, in order to evaluate its effects on
the growth record of the developing countries. In this chapter we motivated the
study bv discussing some of the limitations of the new economics of institutions
from the point of view of a developm ent economist. The main conclusion is there
is great need for research that is macro and empirical and normative.
Chapter 2 reviews the empirical literature on the relationship between
institutions and economic performance, at the macro level. It finds that the
number of concepts in use vastly exceeds the supporting data. Consequently, the
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
chapter is able to bring together works dealing with, among other themes,
institutions, property rights, the rule of law, democracy, and political instability.
In Chapter 3 I propose an index of institutional development and evaluate
whether it contributes to explain economic performance in a sample of Latin
American countries (for which I also investigate the determinants of the
institutional development index itself), as well as in a larger sam ple of
developing countries.
Using the Granger causality framework, Chapter 4 examines the
relationship between socio-political instability, economic growth and
institutional development in a large sample of developing countries. The chapter
includes the study of the relationship between instability and investm ent (the
focus of existing research), but goes beyond it by also focusing on the much less
explored relationship between instability and hum an capital.
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 2.
A Survey of the Evidence on Institutions and Development
(When Both are Defined Broadly)
Institution is a broad concept that reflects multi-dimensional phenomena,
and as such is elusive conceptually, and more so in face of quantitative work. In
taking stock of the empirical literature we start by recognizing that different
studies concentrate on different aspects or dimensions of institutions. Some look
at characteristics of political regimes, others at the level of instability these reflect.
Still other studies examine the effectiveness of the rule of law, or the various
qualities of the bureaucracy. What is truly striking is that the number of
institutional dimensions examined far exceed the data available for capturing
these dimensions.
The objective this chapter is therefore to review the empirical literature on
the relationship between institutions and economic development. To take stock
of the empirical evidence seems timely, but in doing so we should not be limited
to studies that focus on “institutions" in general, but rather should include
studies that focus on different aspects of institutions. Below we discuss studies
that focus on, inter alia, political institutions, civil liberty, institutional reform,
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
democracy, rule of law, political repression, political instability, and civic
attitudes. Although the authors label their subjects in different ways, a reason for
putting all these together is that the underlying data (to m easure these subjects)
is by and large the same. Approximately 70% of the studies reviewed below use
the “Gastil indexes" to measure one of the concepts just listed. The remaining
30% of the studies use other data sources or own classifications (usually dum m y
variables).
Evidence from studies about institutions can be divided into two broad
groups, collections of case studies and cross country econometric evidence. The
emphasis given to the latter here is because it represents a m ore direct way to
obtain the inputs for a set of generalizations or stylized facts.1 We are searching
for corroboratory and systematic evidence that allow for reasonably secure
generalizations over large numbers of national experiences and over significant
periods of time.
The chapter is organized as follows. Section 1 discusses some of the most
common problems in the empirical w ork on cross-country comparisons. Section
2 discusses difficulties faced by those attempting to establish empirically that
institutions do indeed matter for economic performance. Section 3 reviews a
large num ber of individual studies, which are presented in chronological order.
Section 4 concludes.
1 A synthesis of the large number of case studies is beyond our objectives. Recent collections
include the four volumes edited by Diamond, Linz and Lipset (1989), and the ones edited by
Williamson (1993) and Haggard and Webb (1994).
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1. Difficulties in cross-country growth studies
Having m ade clear that the present review focuses on the econometric
evidence, it is im portant to discuss the many problems and reservations involved
in these types of studies. A significant proportion of the empirical evidence can
be labeled "cross section growth regressions". It seems safe to say that the typical
cross sectional growth study first calculates the average growth rate of per capita
GDP over, say, 1960-1985, then averages a num ber of conditioning variables and
runs the appropriate regression. A well-known problem with this approach is
that the degree of adherence to a given, or explicit, theoretical fram ework is
rather variant among these studies and depends upon the choice of conditioning
variables as well as upon the way they enter the statistical model.: In one
extreme, perfect adherence, the empirical exercise is a test of a growth model. In
the other, when the connection is nonexistent, the objective becomes to disclose
empirical regularities and differences in the determ inants of growth in order to
influence future theoretical and policy work.
Some have advanced the idea that these studies constitute the empirical
basis for the endogenous growth literature', and furthermore, that the major
: This point is vigorously argued by Srinivasan (1994).
’ The development economists' reception to this literature has been half hearted (Bardhan, 1994).
On the other hand, Krugman has advanced the very radical understanding that the discipline has
been "resurrected" at the hand of these "simple models of increasing returns" (1993). Although
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
difference between these and the neoclassical literature of the 1960s is precisely
"the importance that empirical studies have had this time around" (Sala-i-
Martin, 1994). In any case, these studies present problems, other than the
adherence to a theoretical framework.4
What are the main problems in cross-country growth studies? Levine with
Renelt (1991,1992) and Zervos (1993) list five major areas of concern: aggregation
and sampling, interpretation of the coefficients, causality, measurement, and the
absence of sensitivity checks.
The first problem refers generally to sampling and, specifically, to
selection bias. One of the questions is whether countries are the appropriate units
of analysis and, if so, whether the large samples often employed by these studies
can be safely considered as being drawn from a common population (Haberger,
1986). I think the m ost recent literature satisfactorily addresses these concerns. It
is important to bear in mind that their objective is to bring to light patterns of
much closer to the first view, I would suggest that some potential positive spillovers are now on
the horizon (other than the ones related to trade and technology that Bardhan identifies). One of
them is a healthy revival of the empirical tradition in development economics, specially when
bringing back in some of its grand themes, notably, income distribution, human capital and a
productive role for government (Fishlow, 1995). Another is the emphasis on the role of socio
political instability and, to a lesser extent, democratic institutions. This, I believe, puts us much
closer to treating developing countries as having "both interests and passions" and farther from
the tenets of the neoclassical counter-revolution of the early 1980s, from which Lai's suggestion
that a "courageous, ruthless and perhaps undemocratic government is required to ride
roughshod over these newly-created special interest groups" (1983, p. 33) is still a disturbing
example.
‘ Many have sternly criticized these studies by suggesting that they are no place for the
methodological perfectionist, however "There is enough arbitrariness in these figures to make the
purists among us shrink at the thought of touching such data. Given the scarcity of reliable
primary information, the purist position can almost lead to abstinence from empirical work
altogether. It can also result in silence on some of the most pressing policy problems (even though
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
correlations, or empirical regularities, or even more explicitly, factors that can be
held responsible for observed inter-national differences in grow th rates or, less
often, in development records. Phrased in this fashion, the nation emerges as the
natural unit of analysis. Moreover, the fact that growth differentials among
regions within a nation are significantly smaller than the growth differentials
among nations and the existence of central governments contributes to the
defense of this choice (of the country as the unit of analysis). The question
whether countries are sampled from the same population is ameliorated by the
inclusion of continent dummies and by differentiating between developing and
developed countries. Nevertheless, it seems that room for improvement still
exists and more attention to the problems of selection bias should be paid.
The question of the interpretation of the coefficients was just addressed
when discussing the role of the theoretical model. If one considers these to be
resultant from structural or behavioral relationships, the danger is eminent.
However, it should be clear this is increasingly less often observed.
The third problem area is causality (according to Levine, Renelt and
Zervos) or simultaneity (Przeworski and Limongi, 1993.) There are two issues
here. The first is that the literature has already paid due attention and has more
recently been mostly engaged in m ethods that contemplate the problem of joint
endogeneity and reverse causation3 . Notwithstanding, studies which are less
some who are purist on empirical matters seem willing enough to rely almost entirely on pure
theory to pronounce on general policy recommendations)" (Sen, 1994, p. 318).
5 Londregan and Poole (1990) is often pointed as the pioneering work in this regard.
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
sensitive to these issues still can be justified on the basis of prior or outside
knowledge: Burkhart and Lewis-Beck talk about the positive effect of
developm ent on democracy as one of the few "iron laws" in political sociology
(1994). The second issue is that the role of theory in ameliorating this problem
can hardly be over emphasized.
The claim that in these studies there is an absence of sensitivity checks on
the results' robustness is a valid one. There are indeed few indications that this
type of check is done at all, although there are invigorating and clear signs that
these are becoming more and more frequent.
2. Problems in measuring institutions
The measurement problem is particularly evident in the empirical
literature on institutions. As em phasized before, a major reason why one can
include a large num ber of studies on democracy, political instability, civil
liberties, governance structures, institutional features, is precisely because most
of them employ a common and limited set of underlying data0 . A more careful
handling of the m yriad of indicators, with the preoccupation of making clear
This is captured by Armijo, Biersteker and Lowenthal: "Although market libertarians and
modernization theorists do not always use the same terms to describe the phenomena, they both argue
that economic liberalization (whether it is described as economic reform, market-oriented development
policies, a capitalist economy, or economic freedom) and political democratization (whether described as
political freedom, democracy, or the lightly government ways) are in some ways reinforcing" (1994, p.4).
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
what they measure or proxy for, will certainly bring additional strength to the
existing empirical results.
Measures of institutional development share the following shortcomings.
First and foremost, time coverage is poor. Most indicators are not available for
years prior to 197V, and therefore they are not able to fully capture changes. The
resilience of institutions is a quality reserved for the "informal elements"
component of the current definition: organizations and formal rules do indeed
change over time or even, North perhaps exaggerating, "overnight".
Second, a number of dimensions of institutions have been identified in
the literature. However, data capturing these is simply not available. All existing
indexes are in this sense incomplete.
Third, since institutional development is multidimensional, any single
index requires aggregation, the results of which may be sensitive to arbitrary
choices of weights. This problem is still being solved as different authors test
different ways of aggregating their data. It should be noticed, however, that
sometimes aggregation is performed at the coding stage. Below we discuss one
of these indexes that are a "composite without components" (another reason for
Bv basing their index on the widely available data on the components of the monev supply,
Clague, Keefer, Knack and Olson (1996) constitute a significant exception. Yet, the Gastil indexes
used by Scully start in 1973, those used by Knack and Keefer (1995) only from 1972 on in some
cases, and from 1980 on in other cases. The indicators Mauro (1995) used are from Business
International (now incorporated into The Economist Intelligence Unit) are available from 1971
on, and, finally, the indicator constructed by Gwartney et al. (1996) is available only from 1975 on.
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the lengthy discussion of the Gastil indexes is that they are the most widely used
measure of the quality of the institutional framework.)
Gastil's civil liberty index captures the rights of the individual against the
state, that is, freedom of expression, freedom of the press and other media,
various forms of censorship, the right to a fair trial and the political
independence of the judiciary, the number of political prisoners, and the degree
of literacy. Countries are ranked on a seven point scale: w hen ranked as 1, a
country has a political system which adheres scrupulously to the rule of law and
constitutionally protect and enforce freedom of expression. Countries with a
score of 2 aspire to this level, but due to internal struggle and violence, ignorance
or limited channels of the media they are given a lower rank. This scale goes on
and it should suffice to say that in a country with a score of 6, the rights of the
state are given legal precedent over individual rights, and when criticism is
heard, it is so in limited (controlled) intensity. Finally, a score of 7 indicates the
complete absence of public debate over the actions taken by the governing elite.
The civil index reflects mainly the independence of the judiciary and the
freedom of the press, while the rankings of the political liberty index reflect the
degree to which citizens have control over those who govern them.
Consequently, the latter focuses on the nature of the competition among political
parties and the likelihood of their respecting the outcomes of this process.
Roughly speaking, the civil index reflects rights against the state, while the
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
political liberty index reflects rights against government. Countries are ranked in
a seven-point scale: from one (the highest degree of political liberty) to seven (the
lowest). A score of 1 is to describe a political process in which “the vast majority
of the polity is enfranchised with the right and the opportunity to elect those
who govern" (Scully, 1987, p. 603). A country scoring 2 is one where even though
there is an open political process that works imperfectly due to structural
constrains, those who govern can be voted out of office. A country scoring 6 is
expected to be a political regime "without the formality of elections or those
states which offer a single list of candidates and in which voting is largely a
ritualistic celebration of the leadership" (p. 603), while a score of 7 indicates an
absolute tyrannical regime, without any legitimacy at all.
Among the reasons for its popularity are that the index has been produced
annually for a large number of countries and territories since 1973, that it goes
beyond dichotomous classifications and that it is constructed with a
complementary 'index of political rights' (Gastil, 1991). In addition, a meticulous
analysis of its reliability does not point to any major problems (Bollen, 1993).
These two indexes also receive theoretical support from Dahl (1985) and his
earlier work on the importance of institutional guarantees for political
democracy.
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3. The evidence
Arguably the earliest contribution to the literature is that of Lipset (1959).
The central hypothesis of this work is that a high level of economic development
is a favorable condition for democracy and the objective is to unveil the complete
set of such requisites, or pre-conditions. Lipset starts by classifying a sample of
thirty "European and English-speaking Nations" into "stable democracies" (13)
and "unstable democracies and dictatorships" (17). The criteria is "the
uninterrupted continuation of political democracy since World War I and the
absence over the past twenty-five years of a major political movement opposed
to the democratic 'rules of the game'". In the Latin American context, the sample
of twenty countries is sub-divided into 'democracies and unstable dictatorships'
(7) and 'stable dictatorships' (13). Economic developm ent is measured by a
series of indexes/ The time coverage was dictated by data availability and refers
to the period after World War II. To test his hypothesis, he calculates the
averages for each indicator in each continent and examine whether these means
scores lower as you move from the more to the less democratic nations. The
hypothesis of compatibility between democracy and development is largely
confirmed. Nonetheless, it should be pointed that the way it deals with
* These are related to wealth (per capita income, persons per motor vehicle, and doctors,
telephones, radios, and newspapers all per 1,000), industrialization (percentage of males in
agriculture and consumption of energy per capita), urbanization (per cent of population in cities
over 20,000, over 100,000 and in metropolitan areas) and education (percentage literate and
primary, secondary and higher education enrollments per 1,000 inhabitants).
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
democracy is arbitrary and many have worked, since then, trying to alleviate this
problem.'’
Adelman and Morris (1967) provide an interesting comparison because
they use a much larger number of indicators for social, political and economic
factors (41), a much larger number of observations (74 countries between 1950
and 1964) and a much more sophisticated statistical technique (factor analysis).
However, their research design and results are very similar to the ones obtained
by Lipset. Their long-run analysis is carried out for 74 developing countries,
while their short-run analysis is accomplished by controlling for levels of
economic development (defined by the authors as lowest, intermediate and
high). They obtain in their short-run analysis that authoritarianism helped
countries in the low and intermediate levels of development.1 , 1
Dick (1974) employs a research design and strategy that is similar to
Lipset's and asks if there is any meaningful relationship between the form of
government in less developed countries and their rates of economic growth. The
first step in his answer is to classify a sample of LDCs into authoritarian (defined
as a regime where one political party or persons or committee maintains control
See among others, Taylor (1972) and Bollen (1980,1993).
0 They argue that, at the early stages, the evolution of participant political institutions is "of
negligible importance; it assumes increasing relevance as social institutions become more
adaptable to the requirements of economic growth. The association between more democratic and
better articulated and integrated political systems, on the one hand, and levels of economic
development, on the other, probably arises because both the ability to generate sustained growth
and the evolution of more sophisticated political institutions require fundamental changes in
mentality characteristic of the spread of rational thought patterns. The participant style of life
tends to generate a capacity to adapt existing institutional frameworks to continual economic and
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
through force or propaganda), semicompetitive (when there is either one major
and several minor political parties if these can not win elections, or when the
unique party promotes periodically legitimizing elections) and competitive (where
two or more parties contend for periodically and free elections). The study
covers the period 1959-1968 and the sample has 58 LDCs.1 1 Real GDP per capita
growth rates and real GDP growth rates were used to rank these countries and
then these were compared against the classification of political regimes.
The visual inspection of these cross-tables led Dick to advance as his major
conclusion that the "results certainly do not support, and tend to refute, the view
that authoritarian countries are universally capable of achieving faster economic
growth in the early stages of development than countries having competitive
political systems, or that underdeveloped countries face 'a cruel choice between
rapid expansion and democratic processes' " (pp. 823-824).
Bilson's main hypotheses is that "the loss of economic freedom through
welfare state socialism or national economic planning will lead to a decline in
personal freedom" (1982, p.94). Capitalism is best at securing a series of rights
that once disseminated will keep government in check, maximize national wealth
and bring about a civil society permeated by efficient democratic institutions. His
two main hypotheses are that the loss of freedom (civil liberty) is related to the
extent of government economic intervention, and the other is that the degree of
social change. This malleability of social structure is essential both to successful entrepreneurial
activity and to effective political modernization" (p. 172).
37
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
repression of these liberties is related to the "concentration of the politics
industry" in the country.
The article presents empirical evidence on the relationship between
economic and political factors and the repression of civil liberties for a cross-
section of 184 countries and territories for the year 1979. These countries are
classified into capitalist, capitalist-socialist, capitalist-statist, socialist and
territories (the 'economic' axis) and into multi-party, dom inant party, one party
and territories (the 'political' axis). The empirical foundation for the study is
given by the Gastil index of civil liberty (previously discussed.)
In order to test the hypothesis that civil liberty is positively correlated
with the degree of economic freedom, Bilson estimates the predicted value of the
civil liberty index conditional upon the country economic and political
characterization. Notice that for this first step the only innovation is in phrasing:
he calculates means and standard deviations and, from them, establishes the
appropriate confidence intervals. From this point on, visual inspection is used to
conclude that "civil liberties decline as the system become less democratic and
less capitalist" (p. 101).
Bilson's second hypothesis relates the degree of repression of civil liberties
"to the 'concentration' of the politics industry in the country". In its
operationalization he tries to explain the variation in his indexes by a number of
" Plus one aggregate: the 14 countries of Francophone Africa for which disaggregated data were
not available.
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
economic variables1 2 . This test is the first among those reviewed here that uses
regression analysis and, arguably, the first to explicitly bring the issue of
causality to the forefront. In order to test this second hypothesis it requires an
am ount of data that was not available for the original 184 countries and this
reduces the sample to 55. The regression results shows only two of the
independent variables as having a significant impact (at the 5% level) on civil
liberties, namely real income per capita and the ratio of wage and salary
payments to GNP.
Weede (1983) offers a rather complete survey of the economic literature on
the 'incompatibility proposition', that is, that political democracy at lower levels
of development is a luxury good and, as such, will contribute to stagnation
rather than to growth. The author tests the following set of hypotheses: "first, the
more democratic the political system, the smaller the economic grow th rates of
nations are likely to be. Second, the negative impact of democracy on economic
growth should be stronger in LDCs than worldwide. Third, the impact of
democracy on economic growth should be less negative where the state plays a
minor economic role, but more negative where the state plays a major economic
role" (Weede, 1983, p.24). These hypotheses are tested for 94 countries (74 LDCs),
1 2 Namely, real income per capita, population, the growth rate of per capita GNP from 1970 to 1976, and
the ratios of central government expenditures, exports, agricultural income, wage and salary payments, to
GNP.
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the time span is determined by data availability and seems to cover the period
1960-19791 3 .
The first hypothesis is confirmed for the whole sample, and its
qualification (hypothesis 2) is rejected when the sample is restricted accordingly:
"among LDCs, (...) the negative impact of political democracy on growth rates is
so much weakened that its very existence cannot be ascertained according to
conventional criteria of significance" (p. 32). Weede's third hypothesis is tested
by re-running the regressions for a sub-sample of 35 LDCs and DCs where
government revenue as a percentage of GDP exceeds 20%. Nevertheless, the
hypothesis is confirmed.
Weede's main results are that "focusing on LDCs only, where theoretically
and morally any incompatibility between growth and democracy should be the
most serious problem, leads to the verdict that the negative impact of democracy
on growth is too small to be of either statistical or political importance. Certainly,
these results provide no excuse for authoritarianism, neglect of hum an rights or
repression in the name of economic progress. (...) Whether the overall impact of
democracy on growth is weak, there is a subgroup of nations where democracy
contributes to slowing down growth. In nations where the state controls much of
the economy, whether more or less developed, political democracy is a major
barrier to economic growth. Elsewhere this is not so at all." (pp. 35-36).
: 3 The dependent variable was economic growth (measured alternatively bv growth rates of GNP per
capita and growth rates of GDP) and the list of regressors induded political democracy (from Bollen,
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The main objective in Kormendi and Meguire (1985) is to provide
exploratory empirical evidence on a group of hypothesized macroeconomic
determinants of growth, among them civil liberty (the Gastil index), in 47
countries between 1950 and 1977. To purge the Gastil civil liberty index of any
spurious cardinality and to limit the extent of possible measurement error in the
index through grouping, a dichotomous variable was defined which took the
value of 1 if the Gastil index was either one or two, and zero otherwise. This
variable was then included in three general equations: first in the list of
regressors of economic growth1 4 , next in this list of regressors of economic
growth augmented by the investment-income ratio and, finally, it was included
as a regressor when the dependent variable was the investment-income ratio
itself.1 3
Their results are that the effect of civil liberty on economic growth is
positive, but only marginally significant when the investment ratio is excluded.
When the investment ratio is included, the civil liberty dummy channels a great
fraction of its power through the newcomer, suggesting "that the effect of civil
liberty on growth operates mainly through the investment channel rather than
the return channel" (p. 156). This last qualification should be retained because it
1980), the level of economic development (in 1965), investment, human capital (primary and secondary
school enrollments in 1960) and the military participation ratio (1965).
1 4 The basic group of regressors in Kormendi and Meguire is the following: per capita income at the
beginning of the period, the mean population growth rate, the standard deviation of real output growth,
the standard deviation of money supply shocks, the mean of money supply growth, the mean growth of
the ratio of government spending to output, the mean growth of exports as a proportion of output, the
mean growth in the rate of inflation, the mean investment to income ratio, and civil liberty (from Gastil).
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
is, in my opinion, one of the greatest contributions of this article: the puzzling
finding that civil liberties have such a dramatic impact on investment but a
relatively modest impact on grow th1 0 .
Vemieris and G upta (1986) introduce income distribution and socio
political instability as argum ents in the savings function for 49 countries for
which data were available (a cross-section with observation dates ranging from
1957 to 1969). They conclude that socio-political instability's "negative influence
is stable, large and statistically significant" (p. 879). Surprisingly, they find that
the middle 40% of the population is the "highest savings-producing group on the
average and at the m argin" (p.880). The method they employ to build the index
of socio-political instability is an important innovation. They use discriminant
scores and the associated probability (that a country is stable) of a function that
includes protests, demonstrations, deaths, and regime type (a dum my variable
for democracy).
Kohli's research question is "what is the impact of democratic rule on
economic growth, income distribution, and the m anagem ent of foreign debt?"
(1986, p. 155). In order to answer it, he selects ten developing countries and
divides them in two groups of equal size according to their Bollen (1980)
1 5 The reason for their inclusion is that investment has been identified as one of the few robust
variables in explaining cross-country economic growth (Levine and Renelt, 1991,1992).
° Gupta (1988) and Pourgerami and Assane (1992) present interesting extensions and refinements
of this model.
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
measure of democracy, for the year 1965.1 ' He compares annual GDP growth
rates for 1960-1970 and for 1970-1982, the direction of change in income
inequality between 1960 and 1970 and the average external debt service as a
percentage of exports for 1980-1982. By visual inspection, Kohli reason that “the
economies of many democratic Third World countries have grown at satisfactory
rates" (p.156), although "over the two decades, the democracies grew at about 5
per cent annually, while the selected authoritarian countries grew at about 6 per
cent" (p. 158).
Scully (1987) argues that the extent of freedom is greater under common
law than under civil law (Muslim law and marxist-leninist law are also
examined). The Gastil measures of political and civil liberties are used again to
test the hypothesis that the choice of law determines the extent of liberty in 167
countries - classified by type of legal system - for the period 1973-1984.1 S
The test of the hypothesis that the choice of law determines the extent of
liberty proceeded by stepwise regression and by probability estimates of the
incidence of freedom and tyranny under common and civil law systems. Scully's
’ The democratic group is composed of Costa Rica, India, Malaysia, Sri Lanka and Venezuela
("Group A"), while the other brings together Argentina, Brazil, Egypt, Morocco and the Republic
of Korea ("Group B").
1 5 This particular article is also important because it evaluates the reliability of the Gastil's
indexes. This is accomplished by correlating them with rankings of thirty human rights or
freedoms (compiled by Humana and measuring signatory compliance with the "United Nations
Declaration of the Rights of the Man"). He concludes that "some 25 of Humana's separate
rankings of human rights are statistically associated with Gastil's ranking of political liberty at the
95 percent level of significance, while 27 out of 30 statistically associated with the ranking of civil
liberty. These statistical results may be interpreted as rather firm support for the accuracy of
Gastil's description of the actual levels of political and civil freedoms in nations throughout the
world" (p.604).
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
most robust result is that "the finding of statistically greater chances of freedom
and lower probabilities of tyranny under common law is invariant to the choice
of the probability model" (p.613). Further, he argues that when the rule of law is
unpredictable and uncertain, transaction costs are higher.
Sloan and Tedin (1987) study the relationship between regime type and
five public policy outcomes.1 U Twenty Latin American countries are classified
according to their regime type between 1960 and 1980, in democratic,
bureaucratic-authoritarian, communist, traditional authoritarian, and
transitional. A simple regression model is formulated where this classification
and regime duration (in years) are regressed on GDP per capita, among other
indicators. Their main result is that "democratic regimes had the second-best
record in promoting economic growth and agricultural production, in avoiding
external debts, in controlling military expenditures, and in making advances in
the health field" (p.121). Democratic regimes in Latin America, between 1960 and
1980, have better growth records than traditional authoritarians', but worse than
bureaucratic-authoritarians'.
In another contribution, Scully (1988) uses measures of the institutional
framework to explain the growth rates of per capita GDP and efficiency
measures for 115 developed and developing market economies over the period
Namely, domestic economic performance, agricultural production, military spending, external
debt and domestic social performance.
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1960-1980. “ The results when each of these variables is regressed alone on
economic growth was that all show the expected signs and are significant at the
1% level (except "POL OPEN", which was significant at the 5% level). The results
are similar, and m uch more interesting, when all of them are concurrently
regressed on economic growth, except that the sign of "POL OPEN" turned out
to be negative. His m ain result is the following: "It was found that the choice of
the institutional fram ework has profound consequences on the efficiency and
growth of economies. Politically open societies, which bind themselves to the
rule of law, to private property, and to the m arket allocation of resources, grow
at three times (2.73 to 0.91 percent annually) the rate and are two and one-half
times as efficient as societies in which these freedoms are circumscribed or
proscribed" (p.661).
Marsh (1979, 1988) draws upon major sociological theories of economic
growth to establish a set of hypotheses, tested against the average rate of GNP
per capita in two periods, 1965-1984 and 1970-1978, for 55 LDCs. The framework
■ ° The variables that capture the effects of the institutional framework on economic development
were the Gastil indexes of political and civil liberties, and Gastil's classification of economic
systems (in capitalist, mixed-capitalist, capitalist-statist, mixed-sodalism, or socialist) and of
levels of economic freedom (in high, medium-high, medium, low-medium, and low). These last
two indexes were jointly used as indicators of economic liberty. These variables were made
operational in the following manner: if the ranking of political liberty is less than 2 (more than 5)
the dummy variable POL OPEN (POL CLOSE) takes on one, otherwise it is zero; if the ranking of
civil liberty is less than 2 (more than 5) the dummy variable INDIV RIGHTS (STATE RIGHTS)
takes on one, otherwise it is zero; and if the ranking of economic liberty is less than 2 (more than
5) the dummy variable FREEMKT (COMMAND) takes on one, otherwise is zero.
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
includes, as control variables, the level of economic development at the
beginning of the period and the share of gross domestic investment in GDP.:|
The indicators used to assess the effect of political and institutional factors
on economic growth are the two Gastil indexes and a measure of competition
among political parties (the likelihood that two random ly selected members of a
country's national legislature belong to different political parties in 1965). In the
results for the 1970s none of these three variables was significant, although civil
rights and party competition have the expected negative coefficients. A similar
result obtains for the unrestricted sample, the difference being that only party
competition has a negative coefficient. Among the various results, it is
worthwhile to stress that Marsh hypothesizes that "the more political
competitiveness and the better the human rights record in LDCs, the slower the
economic growth rate would be. We found this not to be the case when the
proper control variables are used. The consistently non significant effects of
democracy, political and civil rights on economic growth means that while these
variables do not improve a country's growth rate, at least they do not necessarily
impede it" (1988, pp. 68-69, italics in the original).
His nine hypotheses are the following: a greater rate of economic growth is expected from
greater human capital investment in schooling and literacy (HI), a more extensive the spread of
mass media (H2), greater government revenues and public investment (H5), a larger military
participation ratio in the population (H6), and a higher military expenditure as a percentage of
GNP (H7). By the same token, a lower rate of economic growth is expected from greater ethno-
linguistic heterogeneity (H3), a country's greater economic dependency in terms of export partner
concentration and direct foreign investment (H4), a higher level of democracy and human rights
(H8) and the more unequal a society's income distribution (H9).
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Vorhies and Glahe (1988) investigate whether political liberty brings about
social development. The former is measured by the Gastil civil liberty index and
the latter by a group of six indicators." They operationalize this hypothesis by
regressing civil liberties on each of the six indicators, individually, for a
maximum of 150 countries. These correlations were found to range from 0.179
for the non-agricultural share to 0.328 for life expectancy, and they conclude that
"for the world the rank of political liberty may have a significant though small
relationship with a nation's level of social development" (p.50). Next, they repeat
this procedure for the five sub-samples, of different sizes, that the Gastil
classification of economic systems allows for: "in Socialist, Mixed Socialist and
Capitalist-Statist countries political liberty appears to have virtually no
relationship with social development. (...) On the other hand, in the Mixed-
Capitalist and the Capitalist nations there is a significant relationship between
political liberty and social development" (p.52-53).
Arat's contribution (1988) is to test whether the level of socio-economic
development (as m easured by energy consumption per capita) has any effect on
the level of democracy- Arat builds a new index of democracy, combining the
extent of political participation, competitiveness of the political system and a
measure of the coerciveness of government, for the period 1947-1977, in a large
sample of 130 developing and developed countries.
Namely, GNP per capita, average life expectancy, literacy rate, infant survival rate, income
equality index, and the percentage of non-agricultural GDP).
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The estimation procedures encompasses pooled, time-series and cross-
sectional data, and the results sustain the view that "increasing levels of
economic development do not necessarily lead to higher levels of democracy,
even for the less developed countries" (p.30). Moreover, they lend support to the
notion that "democracy is not a one-way ladder that countries climb as their
economy and social structures develop.(...) a positive correlation between
economic development and democracy displayed by the cross-sectional data, but
not confirmed by the longitudinal data, supports the argument of economic
development as a necessary but not sufficient condition of democratic
development" (p. 33-34).
Pourgerami (1988, 1992) tests the hypothesis that the level of economic
development determines the type of political institutions and the type of political
institutions impacts the rate of economic growth. The author includes measures
of democracy (the Political Repression Index), investment, education, a dum m y
variable for culture (one if the country shows a "cultural tradition tolerant of
diversity, conflict, and compromise", 1988, p.128, and zero otherwise), union
membership, share of public welfare expenditures, type of economic system, and
economic development (measured by the level of the Physical Quality of Life
Indicator). The sample contains 92 developing and developed countries for the
period 1965 to 1984. His decomposition analysis show that among the
preconditions for the emergence of democracy, the type of economic system and
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
culture "show strong and highly significant positive effects which are largely
transmitted directly" (p.136). Moreover, "development affects democracy
directly and indirectly via education and investment. The positive causal
association between democracy and growth is transmitted both directly and
indirectly via [unionization and welfare expenditures]" (p.139).
Grier and Tullock (1989) evaluate the effects of a series of macroeconomic
variables, among them the level of political repression, on the economic grow th
record of 113 developing and developed countries. They pool data for 89
developing countries for the years between 1961 and 1980.2 3 The exercise is first
carried out for non-weighted observations and then for population-weighted
continental sub-samples. In the first case, the coefficient for lack of civil liberties
is negative and significant in Africa and in the Americas, but non significant and
negative in Asia. The population-weigh ted results are rather similar, except for
Asia, where the coefficient is not significant and positive.
Remmer (1990) classifies eleven Latin American countries according to the
characteristics of their political regime, between 1982 and 1988, as old and new
democracies and authoritarians. She provides evidence that, when controlled for
the magnitude of the debt burden at the outbreak of the Debt Crisis (1982), "the
puzzle of the 1980s (...) has not been the fragility of democracy, but its
They regress the growth rate of GDP on initial per capita real GDP, mean growth of the
government share of GDP, mean population growth, standard deviation of real GDP growth,
mean inflation rate, mean change in inflation, standard deviation of inflation, a dummy for OPEC
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
surprisingly vitality in the face of overwhelming economic constraints" (p.335).
She tests for differences in six performance indicators (GDP grow th, government
deficit over GDP, inflation rate, the ratio of debt to exports, the evolution of real
wages and of the unem ploym ent rate) and concludes that no statistical difference
can be identified between democratic (old and new) and authoritarian regimes.
McMillan, Rausser and Johnson are interested in the effects of institutional
reforms on economic grow th (1991). The indicators for reform are the Gastil
indexes of Civil and Political Liberties and the exercise is carried out using
regression analysis on panel data for 56 countries betw een 1973-1985. The
dependent variable is a three-year moving average of annual per capita GDP
growth rates. Among their explanatory variables the ones that deserve particular
attention are the ones related to the timing and m agnitude of institutional
changes2 4 . Their major result is that the "coefficients for liberal political rights and
free civil liberties are associated with higher economic grow th" (p. 14) and that
these benefits are realized with a delay. For political rights, until the second year
the result is negative, while after the fifth it turns positive, "indicating an
increase in growth in the fifth and subsequent years following a reform in
political rights" (p. 15). A similar result obtains for civil liberties. Sensitivity
analysis focused on the lag structure and it did not reveal any major problem.
membership and a dummy for lack of political liberties (including countries in the two most
repressive categories of the Gastil index of civil liberty).
These are operationalized by setting up a series of dummy variables in the following manner: they
assume the value of one if the observation has had a political right scale value less than the historically
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Barro (1991), Like Grier and Tullock, is interested in evaluating the effects
of a large num ber of different factors on the economic growth record of 98
developing and developed countries, between 1960 and 1985. These factors are,
in his first model, GDP, secondary and primary school enrollment ratios at the
beginning of the period. Later, he adds adult literacy, student-teacher ratio in
primary and secondary schools, the average real governm ent consumption to
real GDP (excluding defense and education) and dummies for economic system,
for Latin America and for Africa. He also includes, as measures of political
instability, the average number of revolutions and coups and the average
number of assassinations per million population per year. When these last two
variables are regressed on the average growth rates of per capita GDP, they
turned out negative and significant. The result holds whether the period is 1960-
1985 or 1970-1985.
The article from Benhabib and Spiegel (1992) is an exercise in growth
accounting. Their work is concerned with the way we theorize about the role of
human capital in economic development and how this can be better translated
empirically. They m odel how levels of human capital can influence per capita
income growth along the transition path, and they offer two possible
mechanisms: an endogenous growth one and a catch-up one. The former states
that the Solow residual is a direct function of the levels of educational
high for one year, zero otherwise. The series are established by changing the number of years after the
historical high, up to five years. The same procedure is performed for the civil liberties index.
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
attainment, while the latter holds that the ability of a country to adopt new
technologies from abroad is a positive function of its domestic stock of hum an
capital. As for their empirical results, it is interesting to note that their political
instability coefficient turns out to be insignificant in explaining differences in
growth rates but is im portant in explaining investment rates (although only
when the human capital variables are omitted).
The empirical evidence presented by Ozler and Rodrik (1992) focus on
private investment rates and include two sets of variables to proxy for their
"political transmission mechanism". The first is the proportion of the population
that lives in cities and the second is the Gastil Indexes of Political and Civil
liberties. These are entered in levels and in interaction with the authors' m easure
for external shocks (the LIBOR rate). One of the most interesting result is that
when these sets enter in levels they are never significant, but the conclusion are
very different when they enter in interactions: "urbanization tends to magnify the
consequences of an external shock on private investment (...) and the effect of an
external shock on investm ent is larger in countries with more restricted political
systems (...) Conversely, increased political liberties dampens the effect of
negative shocks" (p. 157, italics in the original).
Alesina and Rodrik test the notion that "in democracies, where voter
preferences presumably influence government policies, more concentrated
wealth distributions are conducive to lower rates of economic growth" (1992, p.
5 2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
23). They use a sample of 67 countries for the period 1960-1985, sub-divide them
into democracies and non-democracies, and find strong confirmation for their
main hypothesis (the results are significantly different between samples). The
potential for reverse causation is also minimized: their 2SLS results confirm and
strengthen the notion that "equality is conducive to growth, and this effect
operates primarily in democratic regimes" (p. 40).
Persson and Tabellini (1992) use a similar specification but instead of sub
dividing the sample, they choose to include a dummy variable for democracy.
Their results suggest that "a more unequal size distribution of income is bad for
growth in democracies, while more land concentration is bad for growth
everywhere" (p. 18). They formulate a general politico-economic equilibrium
growth model that formalizes the notion that growth is an increasing function of
a more equal income distribution as well as of a higher average level of basic
skills. They provide additional empirical substance to the argum ent that
democracies with initially unequal income distributions show a worse growth
record than the ones with more even distributions. The rationale is that the poor
vote for higher taxes on capital (broadly defined), thereby generating a
disincentive to investment.
In related work, Persson and Tabellini (1994) regress measures of income
inequality, school enrollment and democracy on the average growth rate of real
GDP and investment for 49 developed and developing countries between 1960
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and 1985. Sensitivity analysis indicates that the results are not affected by
measurement error, sampling, reverse causation or om itted variables. They
concluded that "democracies on average grow faster and have a higher initial
level of per capita income" (p. 613). Democracy has a positive and significant
effect on growth but only when income inequality is low at the beginning of the
period. Moreover, the findings "suggest that the effect of equality on growth
may indeed operate through a political mechanism" (p. 613).
Pastor and Hilt (1993) investigate the effects of democratic institutions on
private investment in seven Latin American countries between 1973 and 1986.
Although the major result is that democracy exerts a significant and positive
influence on private investment, the article also deserves note for the indicator of
democracy used and the rationale it offers for the expected effects. They argue
that democracy "may be consistent with the long-term stability of social
institutions necessary to reassure economic agents and foster investm ent and
growth; authoritarian regimes in contrast, are inherently unstable due to their
insensitivity to public preferences with regard to economic policy" (1993, p.492).
The confidence in the lasting effects of democracy derives from the "popular
accountability of democratic regimes [that] affords them a much deeper
legitimacy" (p.503).
They decide on four different indicators for the role of political variables
on private investment behavior: the 10 point democracy index from Gurr, labor
5 4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
income share, a binary variable for the country's adoption of an IMF economic
stabilization program, and the BERI index (a composite of quantitative
assessments of political characteristics, such as size of bureaucracy, currency
convertibility and contract enforceability). The evidence seems to be robust to
changes in specification: the coefficients of all the political variables, have the
expected signs and are significant (although, the BERI index is significant at the
10 or 20% level in three specifications and the IMF variable is significant at the
107c level in one of them).
Chatterji, Gilmore, Strunk and Vanasin (1993) analyze the role of political
and cultural factors on the growth record of 85 LDCs between 1960 and 1985.
After assembling a simple growth model in order to assess the empirical
plausibility of the convergence hypothesis (whether the growth process tends to
equalize incomes per capita across countries)2 3 , they briefly review the literature
and find that consensus has been built on a list of "conditions for convergence"
(savings ratios, human capital and government spending) that, surprisingly,
excludes non-economic variables. In order to bring in these variables they rank
85 LDCs on the basis of the Gastil index of political liberty and the Human
Freedom Index (from UNDP). In this scale, one is assigned to the country with
the highest degree of political freedom (Costa Rica) and 85 to the lowest (Angola,
Burma and Congo). The finding that a greater degree of political freedom
corresponds to higher growth in real GDP per capita is judged by the authors' as
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
being robust. Moreover, evidence supports the result that convergence “to a
lower level of real GDP per capita in the long run" (p. 2033) is more likely to
occur for countries with low levels of political freedom.
Bhalla (1994) considers the relationship between political and economic
freedoms, and economic growth and social welfare. His largest sample comprises
79 countries covering the period 1973-1990. One of his most valuable
contributions is to differentiate the effects of political and economic freedoms.
The former is measured by a manipulation of the Gastil political and civil
liberties indexes, while latter notion is implemented by openness to international
markets and by the black market premium on the domestic currency. In a
simultaneous equations frame, he finds a strong positive relationship between
both freedoms and economic growth. His results for social welfare (as proxved
by infant mortality and secondary school enrollment) are that political and civil
liberties, as well as education, are strongly significant and positive, but grow th is
not significant in the enrollment equation.
The Barro and Lee (1994) contribution is the analysis of a set of five
influences that they found discriminate relatively well between slow and fast
growing countries (what they call “losers and winners"). Winners have higher
initial levels of hum an capital, higher investment rates, smaller governments, less
"government-induced market distortions", and less political instability. Political
instability is measured by the average number of successful and unsuccessful
2 5 A study along similar lines is that of Wolf (1994).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
revolutions per year and also by a dummy variable for "countries that
participated in at least one external war between 1960-85 and by an estimate of
the fraction of time over 1960-85 that the country was involved in an external
war" (1994, p.284). This exercise is carried out for 85 developing and developed
countries between 1965 and 1975, and for 95 developing and developed countries
between 1975 and 1985. The results support the view that "political instability is
harmful for growth" (p.295).
Perotti (1994) argues that there are three main channels in the relationship
between income distribution and economic growth: imperfections in capital
markets, voting on fiscal policy, and political instability. Each of these offers a
different set of testable implications. When the favored channel is the capital
market, these implications are: (1) for a given degree of capital market
imperfection, a more equal distribution of wealth should be associated with a
higher rate of investment; (2) for a given distribution of wealth, a lower degree of
capital market imperfections should be associated with a higher rate of
investment; and (3) the importance of the distribution of wealth in affecting the
rate of investment should decrease as the degree of imperfection in capital
markets decreases. The voting on fiscal policy approach yields two testable
predictions: (1) the rate of investment decreases when governm ent transfers
increase; and (2) governm ent transfers increase with the distance between the
average and the m edian voters. Finally, the political instability approach
5 7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
typically yields these hypotheses: (1) the rate of investment is a negative function
of socio-political instability; and (2) socio-political instability is a positive
function of the inequality in the distribution of income.
Perotti first estimates and then compares the different models that
correspond to each channel. He uses 70 countries for the period 1960-1985 to
show that all three implications of the capital markets approach are confirmed
(OLS), that the two predictions from the political instability approach are bom
out by the data, and that the fiscal policy approach results are "conspicuously
inconsistent with (indeed, opposite to) the theory".
Using cross-sectional and pooled data over the period 1960 to 1985,
Helliwell (1994a) evaluates the two-way linkages between democracy and
economic growth. Data availability forces the use of different sample sizes (125,
98 or 90 developing and developed countries) and operationalization is
accomplished in the following manner: "Economic development is measured by
average per capita real income [and] the measure of democracy is obtained by
transform ing measures of political and civil liberties published annually by
Gastil [combined into a Freedom Index]" (p. 227).
His main result is that the effects of income on democracy are robust and
positive. Particularly interesting is that the results show "that the strong
correlation between democracy and income is not simply due to the fact that the
OECD countries are richer and more democratic than most other countries" (p.
5 8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
228) and, consequently, they "cast some doubt on the notion of there being a
threshold level of income above which there is a sharp increase in the probability
of a country being democratic" (p.230).
In addition, the problem of joint endogeneity is dealt with by adjusting for
the simultaneous determination of income and democracy. Since re-estimation
"raises rather than lowers the estimated effect of income on democracy, we can
be sure that the positive effect of income on democracy is not due to positive
feedbacks from democracy to economic growth" (p. 235). This result refers to the
effects of the Bollen and Gastil measures of democracy and personal freedoms on
economic growth, assessed for 90 countries between 1960 and 1985, in a
comparative growth framework in which growth of per capita GDP depends
negatively on initial income levels, as implied by the convergence hypothesis,
and positively on rates of investment in physical and hum an capital. The results
pointed to a very weak negative effect of democracy on subsequent growth. Yet,
Helliwell notes that this weak negative result obtains in the presence of
investment and schooling and this drives him to investigate whether the effects
of democracy on grow th are not direct, as implied by the above results, but
indirect, and more precisely, through education and investment2 '’
He uses the Bollen democracy index for 1960 as one instrumental variable,
and finds that "the democracy index is seen to have a positive effect on
Many of the papers reviewed in this section discuss the investment channel (see also Easterly
and Rebelo, 1993). The education channel is modeled in Saint-Paul and Verdier (1993).
59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
subsequent schooling and investment rates, with the estimated effect becoming
smaller and less significant when account is taken of the effects of initial income
levels" (p. 243). Helliwell indicates that the net effect (indirect minus direct) is
likely to be positive: "even more optimistic is the further result that the economic
cost of democracy drops, and may even be eliminated, when account is taken on
the positive effect that democracy appears to have on subsequent education and
investment, both of which tend to increase economic growth" (p. 246).
Muller and Selligson (1994) use a sample of 27 developing and developed
countries to estimate whether civic attitudes and macroeconomic variables
influenced the level of democracy over the period 1981-1990. Two noteworthy
results are that the larger the income share of the top quin tile, in 1970-1980, the
lower the level of democracy is, in 1981-1990 (this effect is significant in all of
their specifications). Another interesting result is that GDP per capita in 1975,
explaining civic attitudes that characterize a democratic culture, is negative in
most specifications, but always insignificant.
The major contribution in Keefer and Knack (1995) is that, by using newly
available and apparently more direct measures of the institutional framework,
they obtain stronger positive results in terms of its impact on economic growth
and investment2 '. They carry out the analysis for 97 developed and developing
countries between 1974 and 1989. One of these new indicators is the Business
Environment Risk Intelligence index (also used in Pastor and Hilt, reviewed
60
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
above). The BERI index results from the summation of four variables:
bureaucratic delays, nationalization potential, contract enforceability, and quality
of infrastructure. The other indicator used is the International Country Risk
Guide. The ICRG index results from the sum m ation of rankings of the quality of
the bureaucracy, extent and intensity of corruption in government, adherence to
the rule of law, risk of expropriation, and repudiation of contracts by
government. Both indicators are provided by private consulting firms advising
multinational companies on country risk. The authors critically evaluate the
usefulness of using revolutions as a proxy of political instability and also of using
the Gastil indexes as a proxy for property rights (their main concern).
The authors' first analyze the correlations between these two indicators
and revolutions, num ber of assassinations and the Gastil indexes. The
correlations they find are low (indicating that these new variables indeed bring
additional information), except in the case of Gastil versus BERI. Next they assess
the effects of the BERI and ICRG indexes on economic growth.2 3 The results
supports the superiority of these indicators since they emerge with the right sign
and as significant (particularly the BERI index), even in the presence of the "old"
measures.
' Knack (1994) carries a similar exercise along the lines of the convergence controversy.
2 5 On the basis of the "standard" endogenous growth model, they regress GDP in 1960, secondary
and primary school enrollments in 1960, and average annual government consumption on GDP,
the average annual frequency of coups and revolutions, the average annual frequency of political
assassinations, and the deviation of investment price level (relative to the U.S. and from the
sample mean) on the average GDP growth rate for 1974-1989 as well as on the average share of
private investment on GDP.
61
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Keefer and Knack (1995) offer three valuable lessons. The first is that
political revolutions and the Gastil indexes are very rough proxies and, as such,
have natural limits in their applicability. The second is that their pair of variables
support the view that "institutions that protect property rights are crucial to
economic growth and investment". The third is that conditional convergence
emerges when institutional quality is controlled for.
Mbaku (1994) is a significant contribution to our discussion because he
uses different measures of economic development (GNP per capita, the Physical
Quality of Life Index, and the UNDP's Human Development Index). The main
result, from his empirical analysis, is that political democracy, as measured by
the Bollen Index, is positive and significant for the whole sam ple (117 countries).
This is independent of whether the indicator for developm ent is GNP per capita
or the HDI. Mbaku then divides the sample according to income levels.2 '1 Some
problems emerge from this exercise: when GNP is the dependent variable, the
sign of the coefficient for democracy is negative, although insignificant, in the
high income group. And it is positive but still insignificant in the middle and low
income samples. W hen the PQLI is the dependent variable, the sign for
democracy is negative but insignificant for low and m iddle income countries,
and positive in the high income countries (although it is significant only if
investment is not in the equation). Finally, when the HDI is the dependent
^ In high, middle and low with 28, 50 and 39 observations, respectively.
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
variable, the sign for democracy is always positive (independent of income
groups), but it is significant only for the high and low income countries.
Barro (1994) assembles a sample of 89 developed and developing
countries, for the period 1960-1990. He concludes that the direct effect of
democracy on growth is weakly negative but improvements in the standard of
living substantially increase the probability that political freedom will increase.
Interestingly, the variables proxying for standard of living are the three
components of the UNDP's Human Development Index. The article also raises
the issues that there exists "a nonlinear relation in which more democracy
enhances growth at low levels of political freedoms but depress growth when a
moderate level of political freedom has already been attained" (p.26). Barro
argues that understanding the indirect effects of democracy on development is
the most important implication for future research.
Burkhart and Lewis-Beck (1994) employ a sample of 131 developed and
developing countries, for the years 1960-1985, to study the two-way causality
between democracy and development. Their basic model proposes that the level
of democracy in a given year is a function of its previous year's level, of
economic development (their proxy for which is energy consumption per capita),
and of two interaction terms between economic development and the relative
position of the country in the "world-system" (that is, if it is core, semiperiphery,
63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
or periphery). They conclude that development Granger-causes democracy, but
that democracy does not Granger-causes development.
Finally, Feng (1995) differentiates empirically between the effects of
democracy and political instability on economic growth. His starting point is to
isolate the discrete forms of political instability (regular versus irregular
government changes), and to study whether their consequences are significantly
similar. Feng's work suggests that they are not. He uses a cross-section of 97
countries, for the years 1960-1980, and a simultaneous equations framework to
uncover an ambiguous total effect of democracy on growth. On the one hand, the
direct effect is negative and significant. On the other, the indirect effect
(democracy, through the probability of regular and irregular government
changes, on economic growth) is large, significant and positive. He thus
concludes that "whether democracy positively or negatively contributes to
growth is jointly determined by its impact on government changes which in turn
also affect growth and democracy" (p. 28).
4. Conclusions
First of all, a note of caution is in order. Hirschman has recently suggested
(1994) that the relationship "between political and economic progresses" is
characterized by an "on-and-off connection". The difficulties in their study arise
64
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
from the alternation between their interdependence and autonomy. This problem
is even more serious for development economics because of its inherent neglect
of political aspects or, as Hirschman put it earlier, its tendency to perceive
developing countries as having "only interests and no passions" (1981).
There are several points that emerge from this analysis. First, this review
indicates that the empirical work provides robust and increasingly sophisticated
support for a positive answer to whether institutions fosters economic growth
and economic development.
Second, although this effect seems well grounded, the understanding of
the transmission mechanism (to be generous) or the rationale for this positive
impact (to be not) still remains to be more extensively investigated. The good
news is that the evidence suggests human capital and investment as strong
candidates in the transmission process. They are thus natural starting points for
future research. It can not go unnoticed that, regarding these indirect effects, the
rationale for hum an capital and investment is in very different stages. The recent
emphasis has been on the role institutions play in protecting expectations
(because of well-defined property rights), and consequently on their beneficial
effects on the accumulation of physical capital. Very little has been done in terms
of conceptualizing that institutions can play a role because of their effects by way
of human capital.
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A third point is that closer attention to recent theoretical developments by
empirical researchers should be encouraged. Theory is both a mean of solving
the transmission paradoxes and a mean of supporting the empirical findings
more firmly. Furthermore, many of the problem s identified in the empirical work
arise from the fact that these are concepts that do not lend themselves to easy
operationalization. The measurement problem is particularly severe in the sense
that one is able to include a large num ber of studies on democracy, political
instability, civil liberties, and institutional features because all of them employ a
common and limited set of underlying data. Once again, one of the main lessons
from this review is that a more careful handling of the myriad of indicators, with
the emphasis on making clear what they measure or proxy for, would bring
additional strength to the existing empirical results.
And fourth, regional studies of this relationship are in very short supply
indeed. The few exceptions seem to focus on Latin America and regional
diversification of future work should be a priority. In the wake of the Asian
Miracle, the absence of specific studies on South and Southeast Asia is simply
astonishing (Helliwell, 1994b, is the honorable exception).
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 3.
An Empirical Analysis of the Relationship between
Institutional Development and Economic Growth
The objectives of this chapter are to construct an index of institutional
development that reflects the concerns voiced earlier (chapter 1), and to examine
whether this index plays a role in explaining cross-country and over time
differences in economic performance.
As mentioned before, a fundamental characteristic of development-
promoting institutions is representativeness. Inspired by Hirschman's work, we
claim that the central and common characteristic of development-promoting
institutions is that they give agents a voice, a stake in the system. By doing so,
they increase the appropriability of benefits or, conversely, reduce the am ount of
rent-seeking. More specifically, the degree to which the tastes, needs and
preferences of the population are translated into the organization of the state, the
functioning of the government, and the formulation and implementation of
public policies are identified here as the most important institutional quality vis-
a-vis economic development.
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Based on this notion, in Section 1 a com parative index of institutional
development (CUD) is constructed for nineteen Latin American countries from
1960 to 1986. In Section 2 CUD is incorporated into an otherwise standard model
of economic growth (the results obtained for the sample of Latin American
countries are reinforced by similar results obtained for the developing countries
as a whole, also discussed.) Since CIID is a composite index, Section 3 explores
the determinants of CIID and re-estimates the grow th model taking these into
account. Section 4 concludes.
There are several important reasons behind the choice of Latin America as
the main focus of this exercise (in what follows they are not discussed in order of
importance.) First, nowhere else have the shifts in development strategy and its
attendant structural reforms been as striking. Second, since sustaining these
reforms appears to remain a serious challenge in Latin America, success in
extending and sustaining them would seem to require an especially delicate
balance with respect to the role of the state. While in some respects, the state
needs to be strengthened so as to take on new tasks (Edwards, 1995), in other
respects its role may have to be diminished and changed so as to allow greater
play for the market (Wiesner, 1994; Naim, 1995). Third, among regions of
developing countries, it is only for Latin America w here the required data for
measuring and endogenizing institutional developm ent are readily and
68
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
consistently available. Given the interest in examining human capital as another
possible link betw een institutions and economic growth, it is relevant to note that
it is Latin America where it has been suggested that institutional development
can contribute positively to economic development only if it succeeds in
realizing more fully the region's human capital potential (Londofio, 1995). Last
but by no means least, it is in Latin America, with its relatively high level of
resource endowm ents, very considerable growth rate differences from one
decade to another (and especially disappointing growth rates since the late
1970s), that the case for examining the role of institutions in explaining growth
rate differences would seem paramount.
1. A Comparative Index of Institutional Development (CIID)
The objective of this section is to justify the need for, and construct, a
Comparative Index of Institutional Development (CIID) based on generally
observable indicators which, collectively at least, are expected to capture some
essential differences in institutional environments across countries and over
time.
To be sure, there are other indexes that capture aspects of institutional
development, such as those found in Scully (1988), Mauro (1995), Knack and
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Keefer (1995), Clague, Keefer, Knack and Olson (1996) and Gwartney, Lawson
and Block (1996). Yet, for the following reasons, there is need for a new index.
First and foremost, the alternative indexes cover a very short period of
time. Since none of them is available for years prior to 1971l, none is able to
capture the effects of the substantial differences in either institutional orientation
or economic growth between the 1960s and 1990s (e.g., the reversal in growth
patterns that followed the oil shocks, and the debt crisis).
Second, the available alternatives are incomplete in that they fail to
consider the "input side" of the process. For instance, the indicators used by
Knack and Keefer focus exclusively on the efficiency of the provision of public
goods (an outcome), but exclude important factors, such as the choice of the
specific public goods to be provided and the way in which that choice is made.
Likewise, the Contract-Intensive Money (CIM) measure of Clague et al. (1996) is
perhaps too closely related to credit flows, financial intermediation and
investment to make convincing their argument that the causality goes from CIM
to the rate of growth.
By basing their index on the widely available data on the components of the money supply,
Clague, Keefer, Knack and Olson (1996) constitute a significant exception. Nevertheless, the
Gastil indexes used by Scully start in 1973, some of the indexes used by Knack and Keefer (1995)
start in 1972 and others do so only in 1980. The indicators Mauro (1995) used are from Business
International (now incorporated into The Economist Intelligence Unit) are available from 1971
onwards. Finally, the indicator constructed by Gwartney et al. (1996) is available from 1975
onwards.
70
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Third, existing measures have been subject to an implicit selection bias in
that country coverage has been restricted to countries where foreign investment
is, or has been, important. Only for such countries has there been sufficient
incentive to invest in the development of the information necessary for
constructing such indicators/
A fourth and final reason there is a need for a new index is
methodological. Since institutional development is multidimensional, any single
index requires aggregation, the results of which may be sensitive to arbitrary
choices of weights. For example, Mauro's index of institutional efficiency was
obtained by averaging nine indicators, thereby arbitrarily weighting each of
them equally.3 Although the arbitrariness of weighting the individual
components of the aggregate index can be reduced through the use of principal
components and other methods4 , most of these authors (those except Knack and
Keefer) have chosen not to use them, generally on the mistaken grounds that the
underlying variables were highly correlated.
: For example, the set of indicators Mauro used for Latin America are available for Argentina,
Chile and Mexico, but not for Bolivia.
The Economic Freedom Index constructed by Gwartney et al. (1996) has 17 components. They
offer three versions of the index, each based on a different aggregation scheme. One gives equal
weight, and the two others are based on survey responses of two different types: “experts on
economic freedom" and "experts on particular countries" to rank these 17 components over time.
Another method used to construct composite indexes is the Borda ranking technique. See
Thomas and Wang (1996) for a recent example of indexes of "distortions" and "interventions",
constructed using the Borda technique.
71
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
To overcome the latter shortcoming, the construction of CIID is based on
the principal components method.3 The underlying variables of the index are
shown in Table 1 (all data are taken from Gurr, 1990.) These are: the
competitiveness and openness of executive recruitment, constraints on chief
executives, competitiveness and regulation of participation, legislative
effectiveness and selection, and limitations on the economic scope of government
actions. Each of these eight indicators is scored according to the coding indicated
in Table 1." The principal components yields a set of loadings for each individual
country from which a tim e series of values of CIID is generated for the period
1960 to 1986r, the decade averages of which are shown in Table 2.
Before examining the country and decade-specific values of the CIID
index presented in Table 2, it is important to recall that virtually all such indexes
have been known to provide classificatory anomalies. For example in their index
of economic freedom for 1980, Gwartney et al. (1996, 64-65) ranked Guatemala
above Uruguay which in turn was above the United States, while Paraguay's
One use of the method of principal components is to identify a small number of "latent
variables" in a way that as much information (variance) as possible from the original or
underlying variables is retained. It estimates linear combinations of these original variables, all
orthogonal to each other (components), with the property that the components are "uncovered"
in decreasing order bv the amount of the total variance in the original variables thev capture
(Greene, 1990, 271-273).
Admittedly, this index may do short shrift to other components of what Trebilcock (1995) refers
to as the "institutional matrix", such as the quality of the bureaucracy. Naturally, if these were
available I would include them among CIID's underlying variables.
On average, the first of the principal components captures 70 percent of the variance.
72
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
score was greater than Germany's. According to the ICRG's quality of
bureaucracy index, the indicator for Dominican Republic was three times the
corresponding one for Uruguay and the BERI measure for contract enforceability
is higher for Ecuador than Argentina. The fact that many of the indexes of
institutional development are subjective no doubt helps explain w hy the indexes
are often much more comparable over time within a single country than between
countries. However, even the non-subjective CIM index of Clague, Keefer, Knack
and Olson (1995, Appendix C) produces some distinct surprises in that CIM is
higher for Swaziland, Zimbabwe and Venezuela than for the United Kingdom or
Switzerland and for Botswana, Jamaica and Malawi than for Switzerland or
Belgium.
This is not to say that the use of these indexes should be avoided at all
cost. This is a young, im portant and thriving literature and we m ust accept the
inherent difficulties in constructing a precise measurement of institutional
development. The point is simply one way to increase the belief that institutions
matter, is to attempt to demonstrate that an aggregate index helps explaining
differences across countries and over time in economic performance. The quality
of the data will certainly improve, but we are convinced that a key factor in this
improvement is precisely the building up of a consensus that institutions matter
(even when measured imperfectly.)
73
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Since CIID is also based on subjective indicators constructed by G urr from
different, country-specific secondary studies, not surprisingly it produces some
anomalies across countries. The best light to look at the index is therefore to
recognize that the index (CIID and all others discussed above) does a better job
across time than across countries. As can easily be seen, there is substantial
variation in the index across countries at any particular time period. For
example, for the early 1980s, CIID varies from a low of 0.02 for Nicaragua to 2.88
for Ecuador. The index also shows considerable variation over time: substantial
increases occurring between the 1960s and early 1980s in Brazil, Colombia,
Dominican Republic, and Paraguay, substantial declines in Chile, Costa Rica,
Jamaica, Mexico, Nicaragua, and Uruguay and substantial fluctuations in Bolivia,
Ecuador, Guatemala, Honduras, Peru and El Salvador. While the reader may be
surprised by the sharp decline for Chile into the early 1980s, it should be recalled
that the early 1980s witnessed the height of that country's military dictatorship
and preceded m any of its subsequent economic reforms. Chile's CIID m ust
surely have increased substantially since that time, especially with the successful
implementation of the "growth with equity" strategy of the early 1990s
(Raczynski, 1995).
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2. The Impact of Institutions on Economic Growth
In order to answer the question "Do institutions matter?", in this section
we add our index of institutional development (CUD) to an otherwise standard
Solow growth model, and then use that model to explain variations in growth
rates across 19 Latin American countries and the last three decades. The use of
the Solow model is motivated primarily by the fact that it contains a shift
parameter that "reflects not just technology, but resource endowments, climate,
institutions and so on" (Mankiw et al., 1992, 410-411, italics added), thereby
making explicit the link between institutions and economic growth. Additional
advantages of the Solow model in this context are: (1) the comparisons it affords
with the many other studies which use this framework8 , (2) its ability to test
some other important hypotheses such as (a) that income per capita should be
positively related to savings and negatively related to population growth rates,
and (b) that countries converge to their steady-state levels of income per capita,
Several authors have found the Solow framework, at least in modified form, to be superior to
the endogenous growth models. For example, Cardoso and Fishlow (1992) find that an
augmented version of the Solow model incorporating the external sector to be superior. Even
among those that embrace the endogenous growth framework, there is open recognition that its
"silence with respect to the underlying model" is an important shortcoming (De Gregorio, 1992,
69). Levine and Renelt (1992) provide evidence that the econometric results from endogenous
growth studies tend not to be robust. Finally, it should be noted that we are unaware of other
studies on the role of institutions and economic growth that attempt to establish this relationship
from a standard theoretical framework.
75
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and (3) that it works especially well (with convergence properties fulfilled) in
samples of relatively homogeneous countries like those of Latin America.
The central piece in this model is, of course, a production function with
positive and diminishing marginal products and constant returns to scale. It
relates output (V ) to a pair of essential inputs, capital and labor (K and L), and to
an efficiency parameter (A) representing either technology or institutions. If the
function is of Cobb-Douglas form, output in period t is:
(1) Y t = Kia(AlLl)i~a 0 < a < 1 .
It is assumed that the technological or institutional progress is labor-augmenting
and that the rates of growth of population (u), technological or institutional
progress (g) and depreciation (S) are constant and exogenous for any period*.
If k is the capital-labor ratio and y is income per worker, the assumptions
about the growth of population and technology imply that, in the steady-state
(i.e., when the various quantities grow at constant rates), k. would converge (for
small values of n, g and < 5 ) to a value k* given by:
9 See also Dixit (1976). Artus (1993). and Barro and Sala-i-Martin (1995).
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(2) £* = [ ]^~a)
n+g + 5
Substituting (2) into the production function and taking logs, the steady-
state income per worker becomes:
(3) ln[— ] = ln.T +gt + a ln(s) — ln(/7+g + d > )
L, 1 - a 1 - a
This yields the well-known hypotheses of the Solow model: the higher the
rate of saving, the richer the country; and the higher the rates of population
growth, labor-augmented technological change and depreciation, the poorer the
country. The m odel not only predicts that income per capita in each different
country will converge to its steady-state value, but also yields estimates of the
speed at which this convergence occurs. Let y* be the steady-state level of income
per worker from (3), and y. be the actual value at time t. In the neighborhood of
the steady-state, an approximation of the speed of convergence p is given by:
dln(y,)
(4) — ^ = /ffln(yf_1*)-(ln(y,_I)]
where P = - (n+g+5) (1- a).
77
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Notice that the speed of convergence depends on the determinants of the steady-
state as well as on the level of income per worker at the beginning of the period.1 0
There are good reasons to expect that this model would perform better
when tested for samples that are relatively homogeneous (i.e., that share similar
steady-states.) For example, Mankiw, Romer and Weil (MRW, 1992) obtain quite
different param eter estimates for OECD countries than for non-oil exporting
LDCs. They found evidence of "unconditional convergence" am ong the OECD
countries1 1 , but little or no such evidence for their larger — and much less
homogeneous— samples. Given that our sample is limited to Latin American
countries, this framework would seem quite appropriate.
Our first step is to assess the specification provided in MRW (1992). They
assume that g and S are constant across countries (because the technology
frontier is universal and data on country-specific depreciation rates are not
available) and take the values 2 and 3 percent, respectively.1 2 They also assume
that the effects of population growth (n) and savings (s) are independent of
1 0 The concept of conditional convergence does not imply any tendency for the dispersion (or
variance) of per capita incomes to decrease (the latter is often referred to as " cr convergence)
(Barro and Sala-i-Martin, 1995, 383-387).
" The importance of homogeneous samples can be appreciated by pointing out that Barro and
Sala-i-Martin (1995) present similar findings (unconditional convergence) for the states in the
U.S., regions within Europe, and prefectures within Japan.
1 2 This is obviously a very strong assumption, as correctly pointed out by Srinivasan (1994, 271).
Yet, since this is not directly relevant to our present purposes, and for convenience, the same
78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
country-specific factors (captured by the stochastic term), thereby justifying OLS
estimation. The MRW specification, therefore, requires that a coefficient for
initial incomes [from (4)], a constant, and a stochastic term should all be added to
equation (3) above. Table 3 shows the results for this specification, obtained with
the use of their data for Latin America.1 ’
In general, this specification fits the data quite well. The adjusted R’ is
only slightly smaller than the one obtained by MRW for their sub-sample of
twenty-two OECD countries (of 0.65). This seems to support our claim that Latin
America may be more homogenous (in terms of determinants of the steady-state)
than the world as a whole or even Asia or Africa alone. In other im portant
respects, however, the results are mixed.1 4 On the one hand, the convergence and
the investment coefficients both have the expected signs and are significant. On
the other, the coefficients on human capital and growth of population have the
"wrong" signs, although neither is significantly different from zero.1 3
assumption is made here. Helliweil discusses studies that use different values of g, "thus giving
international transfers of knowledge a key role to play in the convergence process" (1994, 237).
1 3 The sample is the same set of 19 countries listed in Table 2. It should be noted that for none of
these Latin American countries was the quality grade of the Penn World Tables Mark 5.6 data
used here as low as "D" (which is reserved for data based on extremely scarce primary sources
and said to be of inferior quality).
u The small sample size may also adversely affect these results. Nevertheless, problems are
common in growth studies which include human capital, as well documented by Benhabib and
Spiegel (1994) and Pritchett (1996).
” These last two results (on human capital and population) are very important because they could
constitute a case against the use of the Solow model for this particular region. This would not be
so, however, if in fact education has been characterized by over-investment in higher education
79
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
To help overcome these shortcomings in the application of the model to
Latin America we make the following modifications: (1) the inclusion of dum m y
variables for the three decades (1960s, 1970s and 1980s) and for each country in
the sample, (2) the use of a better proxy for human capital'0 , and (3) the
introduction of our index of institutional developm ent (CIID).
Our next step is to assess how the MRW specification from Table 3 would
behave with the above modifications in the hum an capital measure, the inclusion
of dummy variables, and subsequently the inclusion of CIID. The data sources
and descriptive statistics are presented in Table 4 and the results in Table 5.
The specifications in the first three columns of Table 5 test for
unconditional convergence (column [1 ]), and for conditional convergence in the
basic version of the Solow model (column [2]) and in the hum an capital-
augmented version (column [3]). Notice that the coefficients for convergence
(initial income, In Y0), and the investment share in GDP (In l/GDP) in columns [2]
and/or an inefficient allocation of resources. These possibilities suggest the need for better
measures of educational investment in the labor force than gross enrollments rates. Other than
the usual shortcomings, gross enrollment data are particularly problematic for Latin America.
Morley shows that, between 1986 and 1989, Brazil failed to graduate 78 percent of those who
entered primary school, Guatemala 64 percent, Honduras 57 percent, and Bolivia 50 percent
(1995, 60 et seq.).
Here we use the Barro and Lee data (1993), and define human capital as the average years of
schooling of the population over 25 years old. The other comparable option would be the Nehru
et al. data set, but two of the countries in our sample (Guatemala and Dominican Republic) are
not covered there.
80
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and [3], are significant and of the expected sign. These results are quite robust
since they obtain regardless of subsequent changes in specification.
With respect to the coefficients on population (In n+g+5) and grow th of
human capital (HKGROWTH), our results only marginally im prove upon the
ones in Table 3 (i.e., those obtained with the data and specification of MRW).
Although the coefficients on population and hum an capital are not significant, at
least they have the expected signs (this holds irrespective of the specification).
Beginning w ith column [4], the results reflect the inclusion of our index of
institutional developm ent (In CIID). Its inclusion raises the value of the adjusted
R: and its coefficient is positive and significant. Even when investm ent is
excluded from the model, as shown in column [5], the coefficient of CIID remains
positive and significant. More importantly, notice that in colum n [6 ], when we
include our hum an capital variable (HKGROWTH), the coefficients of both
institutions and hum an capital are now positive and significant, whereas, before,
the effect of hum an capital alone was not significant. 1 One possible explanation
lies in the possibility that, if left unchecked by appropriate institutions, those
with more hum an capital may be motivated to take advantage of rent-seeking
To further explore the relationship between human capital and institutional development
various interaction terms and non-linearities were introduced, taking us farther from the Cobb-
Douglas form and closer to the translog form (Lau, 1996). Yet, the introduction of quadratic terms
for CUD and HKGROWTH and a CIID-HKGROWTH interaction term failed to increase the
explanatory power or to alter the results in any meaningful way.
81
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
opportunities, thus implying that human capital accumulation may be
counterproductive to economic growth. On the other hand, with appropriate
institutions, the incentives for rent seeking activities may be reduced and those
for productive activities increased, thereby raising the attainable rate of economic
growth (Murphy, Schleifer and Vishny, 1991).1 8
This is quite a meaningful finding: only after CIID is included, does the
effect of hum an capital become positive and significant.‘ g In contrast to existing
studies which emphasize a nexus between institutional development and per
capita income growth operating through physical capital accumulation, the
results presented here suggest that a similar nexus operating through human
capital formation m ay be stronger.
Given that CIID is itself a very specific aggregation of the eight individual
indicators (shown in Table 1), a question which arises naturally in such a
situation is whether the aggregate CIID performs better, worse or the same as its
Another interesting possibility is provided by Azariadis and Drazen (1990) who formalize the
Bovvman-Anderson-Easterlin argument that there are threshold externalities associated with the
accumulation of human capital. That is, that economic growth "should be correlated with human
investment relative per capita income, with high rates of growth being associated with the prior
attainment of especially high levels of human investment relative to per capita income"
(1990,519). See Behrman (1993) for the relevance of these considerations in the Latin American
context. Notice that the recent empirical literature incorporates this possibility, even if indirectly,
by preferring to use secondary enrollment data as a proxy for human capital.
While the accumulation of human capital is also a central concern of some contributors to the
endogenous growth literature (Lucas, 1988; Barro and Sala-i-Martin, 1995; and Behrman, 1995),
the reasons and mechanisms are very different from those suggested here.
82
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
individual components. Notably, however, the aggregate index has a level of
significance exceeding that of any individual component alone.2 0
Judging by the values of the adjusted coefficient of determination and the
robustness of the results, the preferred specification is that in column [6 ]. Notice
that the value of adjusted R2 is 0.64 which is quite comparable to that obtained
by MRW for their OECD sample (with a reported adjusted R2 of 0.65) and
considerably higher than those obtained for their LDC samples (adjusted R2 of
0.46).2 1
The coefficients of initial income (convergence) and investment have the
expected signs and remain highly significant across the different specifications.
Indeed, their significance is raised slightly by the inclusion of CIID and the
accumulation of human capital. Our results with respect to CIID strongly suggest
that institutional development should be considered among the variables which
have to be controlled in explaining the continent's postwar growth experience.
Institutions do matter for economic growth and when both CIID and human
capital appear in the same equation, the explanatory power and significance of
each is increased. The latter suggests the existence of an im portant and
2 1 If in stead of CIID, we induded all its individual components (from Table 1) in the spedfication
in column [6] (Table 5), we find that none of them is superior to the coeffident of CIID itself. Also,
using the components instead of CIID generates a lower adjusted R ‘ .
; 1 Notice, however, that these results differ from those of MRW in that theirs are from a cross-
section of countries.
83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
heretofore neglected nexus between institutions and growth through hum an
capital.
In order to evaluate whether the results for CIID depend on the fact that
we are looking exclusively at Latin America, the model was re-estimated using a
larger sample of developing countries. To do so, some changes are needed.
Because of the rather stringent data requirements (in particular, a panel for
human capital), the maximum number of countries we were able to use is 65
(Table 6 .) The other significant change is that to save on degrees of freedom, we
decided to use regional dummies, instead of country dummies.
The results are presented in Table 7. Notice first that, comparing columns
[1 ] and [2 ], we obtain the convergence result only after we include the regional
dummies. Differently from Latin America, the coefficient for (In n+g+S) is
significant, at least for the first specifications, and has the expected sign
throughout. The same does not happen to hum an capital suggesting that further
investigation is needed (to study whether this is just a Latin American issue, or
is observed in other regions.) More importantly, the results for our variable of
interest at this point, CIID, clearly indicate that when we use a broader group of
countries we obtain similar results to the ones we obtained above. As already
noticed, this larger sample can not be used for the analysis of the determ inants of
CIID due to its stringent data requirements.
84
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3. What Explains Institutional Development?
Having established that institutions indeed matter (with respect to
economic growth), it becomes important to further examine both how this role is
played and the direction of causality. These are the objectives of the present
section.
In order to identify the factors that account for differences in institutional
development, we face two challenges: (1 ) the lack of a rigorous theoretical model,
and (2 ) the corresponding paucity of relevant empirical exercises.
While, as indicated above, there is a substantial literature on the im pact of
institutional variables on economic growth”, there is much less on the impacts of
economic growth on institutional development. Three fine exceptions'' are (a) an
article by Bilson (1982) explaining variations in the Gastil index of civil liberties
across 55 countries in 1979, (b) a paper by Helliwell (1994) evaluating the two-
way linkages between democracy and growth for a panel of 125 countries for the
years 1960-85, and (c) a paper by Porter and Scully (1995) explaining two-way
feedbacks between economic growth and constitutional changes.
" Such as Scully (1988), Knack and Keefer (1995), and Mauro (1995).
85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Although the Bilson study on civil liberties used seven im portant
explanatory variables2 4 , only two (namely, per capita real income and the ratio
of wages and salary paym ents to GNP), turned out to have effects significantly
different from zero. Yet, both had the "wrong" signs.
Helliwell (1994), on the other hand, combined Gastil's civil liberties and
political rights indexes into a measure of democracy, and pooled data on 125
countries for the period 1960-85 to evaluate the two-way linkages between
democracy and growth. His main findings were that the level of income per
capita and secondary school enrollment both have positive and significant effects
on the level of democracy, and that the degree of democracy is significantly
higher in OECD countries, significantly lower in six oil-dependent countries of
the Middle-East, slightly lower in Africa and slightly higher in Latin America
than in the remaining countries of Eastern Europe and Asia (1994, 228). He also
finds that the direct effect of democracy on grow th is negative, but that its
^ We selectively emphasize contributions that had a greater impact in the economics literature. It
should be clear, however, that there are a number of political science studies that should also be
considered. In this vein, see Burkhart and Lewis-Beck (1994) and references therein.
These were (1) the level of real income per capita (expected to have a positive sign), (2) the size
of the country (proxied by its population), (3) the ratio of exports to GNP (positive effect), (4) the
ratio of central government expenditure to GNP, (5) the ratio of agricultural income to GNP (both
(4) and (5) having effects of ambiguous direction), (6) the ratio of wages and salary payments to
GNP (expected to have a positive effect because a "high labor income ratio is typically associated
with a skilled and literate work force and with a relatively even distribution of income", p. 105),
and, (7) the growth rate of per capita GNP over the period 1970 to 1976 (expected to have a
negative effect due to the fact that, according to Olson, "rapid economic growth is a disruptive
and destabilizing force that leads to political instability", quoted in p. 106).
86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
indirect effect (through education and investment) is positive and somewhat
stronger.
Porter and Scully (1995) explain how the needs to offset the diminishing
returns to factor accumulation in the neoclassical growth model give rise to
attempts in the political market to change the "rule space". Although they stop
short of a full-fledged attempt to empirically apply such a model, they illustrate
its application by the need to make national markets more efficient in the
nineteenth century U.S.. This need was realized by the federalization, and hence
harmonization, of many state and local rules and regulations and was reflected
in the acceleration of new legislation. Under certain conditions (including the
efficiency of the political allocation of rights, an idea very close to the spirit of
CIID in this paper), the allocation of rights and obligations through such rule
changes can be of lower cost than those brought about through private contracts
and transactions.
In this light, our approach to the identification of the determinants of
institutional development starts by devising three classes of variables, namely,
initial conditions, structural variables and the following other direct
determinants of CIID: Gastil's index of civil liberties (CIVLIB), the interaction
between the per decade average number of revolutions and coups d'Etat
(REVCOUP) and the black m arket exchange rate premium (BMP), the latter a
87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
proxy for overall economic distortions. Our expectation is that the effects on
CIID of the first three variables would be positive and that of BMP negative. For
initial conditions, we include the beginning of decade levels of population
(POPO), human capital (HK 0) and real per capita GDP (In YO). O ur expectation is
that the effects on CIID of the first would be negative and the latter two positive.
Finally, as structural variables1 ’ , we include: the percentage of the labor force in
agriculture (PCTCGAG), the Gini coefficient for land ow nership (Gini land), the
ratio of public expenditures in tertiary education to public expenditures in
primary and secondary education (lNEQT_PS)2 b , and dum m y variables for each
country as well as for the three decades. We expect the direction of the effects on
CIID in the first three cases to be negative and in the latter two cases to be
ambiguous. Interaction terms between some of these structural variables were
also introduced.
Although the results presented in Table 8 are exploratory and, thus,
should be interpreted carefully, they are not without interest. In column [1] of
Table 8 are our initial results for CIID, based on the first three variables
mentioned above (plus the series of dummy variables.) Although all coefficients
J The data are included in the Appendix to this Chapter.
2 0 It is quite well documented for Latin America that the distribution of human capital is one of
the major causes of the persistently high income inequality (Behrman, 1993; Londono, 1995;
Morley, 1995; and Berry, 1996). Further, data availability across countries and over time was a
crucial factor in the choice of using the determinants instead of the income distribution data
88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
have the expected signs, only one of them (BMP) turns out to be statistically
significant. The level of overall economic distortions seems to be a powerful
hindrance to the establishment of a sound institutional framework.
As shown in column [2], introducing the initial condition variables does
not improve the results. Noteworthy is that the effect of BMP is still negative and
significant, and that the level of human capital turns out to be positively related
to the level of institutional development. The three remaining structural
variables are introduced in the specification shown in column [3]. Notice that
both initial per capita income and initial level of human capital are significant
and all have their expected signs. “ Notice also that all previous results,
including the significantly negative effect of BMP on institutional development,
are essentially unaffected by the introduction of these terms. “ s
Given the aforementioned relevance of income inequality in general and
rent-seeking in particular to our CIID measure, the lack of significance of one of
our inequality-related structural variables (INEQT_PS) is som ewhat
disconcerting. Yet, since the effects of these variables need not be independent of
directly. For instance, these data is not available for Bolivia for all three decades of interest in this
study (Deininger and Squire, 1996).
r It does not change these results if vve include each determinant one at a time, or if instead of the
interaction term vve use revolutions and coups separately.
We also experimented with other structural variables but the results were similar. Among
them, urbanization, the size of the middle class (third quintile in the income distribution), levels
instead of rate of change of the labor force in agriculture, and the ratio of public expenditures in
tertiary to primary education (instead of primary and secondary).
89
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
each other, in the specification given in column [4] we include an interaction
term between them. Note that the sign of the interaction term turns out to be
positive (suggesting a reinforcing effect), but more importantly, that the
introduction of this term raises the absolute value of the negative coefficients as
well as the significance of the two separate types of inequality. Finally, notice
that the inclusion of this interaction term also raises the significance levels of
BMP.
The above results allow us to identify some factors contributing to the
observed intertemporal and intercountry variations in our index of institutional
development. The estimates shown in column [5] are deemed best and the basis
for endogenizing our index: inequalities and distortions are the major
determ inants of CUD.
Table 9 presents the results of our efforts to remove a possible source of
bias (simultaneous equation): we re-estimated the production function using the
predicted values of CUD given by the previous analysis. Column [1 ] shows the
results for the per capital income grow th rate which obtains from two-stage least
squares, substituting the predicted value of CUD (from column [2]) for its actual
value in the growth equation. Notice that after eliminating this bias, all the
previous results remain. In addition, the significance of both institutional
development and the accumulation of hum an capital are raised.
90
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Given the possibility that these results obtain because the initial level of
human capital is am ong the first-stage determinants of CUD, in column [3] of
Table 9 we show the two-stage least squares estimates of the parameters o f the
growth equation obtained after excluding the initial conditions variables from
the CDD equation. The results for this streamlined CUD equation are given in
column [4]. As can be seen, this modification does not alter any of the results we
obtained before.
4. Conclusions and Suggestions for Further Research
The objective of this chapter was to advance understanding of the
determinants of per capita growth by bringing institutional developm ent
explicitly into an otherwise standard growth model. In particular, w hen
institutional developm ent is defined (as in CUD) in such a way as to pay equal
attention to the "input side" of institutional development as to the "output side",
it turns out to have a robust, positive and significant effect on the rate of per
capital income growth across countries and over time. With high values of CUD,
the opportunities for rent-seeking would seem to be reduced, thereby raising the
incentive for the efficiency-enhancing exercise of "voice" and inducing greater
energies to be devoted to productive economic activities.
91
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Moreover, the fact that CIID displays considerable variation not only
across countries but also over time suggests that the relevant institutions for
economic growth are not as fixed as they are generally perceived to be. This calls
attention to the possibility that institutions (as opposed to policies) could
deliberately be changed in ways that would raise the overall rate of economic
growth. Moreover, since an im portant reason for the positive influence of CIID
on economic growth is due to decreased opportunities for rent-seeking, the
positive influence of institutional development on growth need not occur at the
expense of greater income inequality.
Another and perhaps even more important finding concerns the identity
of the link between institutional development and economic growth. Both the
positive effect of the level of hum an capital on the institutional development
index (CIID) and the fact that both the magnitude and significance of the effect of
human capital growth on per capita income growth are raised by the inclusion of
the influence of CIID on growth suggest that hum an capital could constitute an
important link by which institutions affect growth .29 While this finding need not
refute the relevance of physical capital formation as another link between
Since CIID and rent-seeking should be negatively correlated, the finding of a significant
negative effect of human capital growth on CIID appears to support Pritchett's conjecture that
"the institutional environment in many countries has been sufficiently perverse that accumulated
human capital has no effect on, or even has lowered, economic growth even though the returns to
schooling have been substantial because schooling has facilitated rent-seeking" (1996, 33). See
also the important work of Murphy et al. (1991), previously discussed.
92
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
institutions and development, given the virtually exclusive attention that has
been given to the physical capital link in the existing (property rights) literature,
this is an important finding. At a minimum, greater priority of attention in future
research should be given to the triadic relationship between institutions, human
capital and real per capita income growth.
The growth analysis also demonstrates that, when the sample is limited to
countries which are relatively homogeneous, considerable support is provided
for the findings of unconditional and conditional convergence and the effect of
the investment rate on the rate of economic growth. All of the relevant measures
for these variables were found to have the expected signs and to be statistically
significant, even after the sim ultaneous equation bias is eliminated. Hence, in the
latter respects the results also confirm the findings of previous studies,
suggesting the relevance of physical capital investment and conditional
convergence, after controlling for a num ber of variables that now includes
institutional development.
The analysis of institutional development dem onstrated that CIID
responds negatively to the overall level of distortions in the economy
(represented by BMP), and inequality in educational expenditures or land
ownership. These findings seem to support the interpretation of CIID as a
measure of institutional developm ent that reflects "stake in the economic
93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
system" and therefore is of direct relevance to the rate of economic growth rather
than to political democracy as it might at first sight seem.
Some suggestions for future research are the following: (1 ) Priority in
future research should be given to the analysis of the triadic relationship between
institutions, human capital and economic growth. This may be particularly
im portant in the context of Latin America because of substantial evidence that
the average level of education is distorted (toward types which yield high
private rates of return due to opportunities for rent-seeking but low social rates
of return). (2) Given the relatively low value of the adjusted R2 s in the equation
for CIID, a second priority for future research should be given to introduce
additional variables which would increase the explanatory power of the model
without increasing complexity unnecessarily. (3) Another im portant extension of
these models would be to improve on some of the measures used as explanatory
variables. Candidates for improvement would include the human capital
measure (for example, to include health indicators), and CIID itself (to include
aspects related to the bureaucracy and judiciary).
94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 1
Variables underlying CIID (Codings)
Competitiveness of Executive Recruitment 0 = power transfers not regulated, 1 = hereditary
succession, designation or a combination of both,
2 = one executive is chosen by hereditary
succession and the other by competitive elections,
3 = selection is the result of a competitive election
matching two or more major parties or
candidates.
Openness of Executive Recruitment 0 = power transfers not regulated, 1 = hereditary
succession (Closed), 2 = hereditary succession
plus executive or court selection of an effective
chief minister, 3 = selection of an effective chief
minister is through elections, 4 = open
recruitment process.
Constraints on Chief Executives. 1= unlimited authority, 2= intermediate
category, 3= slight to moderate limits on
executive authority, 4=intermediate category,
5=substantial limitations, 6= intermediate
category, 7= executive parity or subordination.
Competitiveness of Participation.
Regulation of participation
Legislative Effectiveness
1 = suppressed competition, 2 restricted /
transitional competition, 3 = factional
competition, 4 = transitional competition,
5=competitive.
1 = unregulated participation, 2= factional or
transitional, 3= factional/restricted, 4= restricted,
5= regulated participation.
0=no legislature exists, l=inneffective, 2=partiallv
innefective, 3=effective and independent from
the executive.
Legislative Selection
Scope of government actions.
0= no legislature exists, 1= non-elective,
2=elective.
1 = totalitarian, 2 = intermediate category,
3=segmental plus, 4 = intermediate category,
5=segmental, 6=intermediate category,
7=segmental minus, 8 = intermediate category,
9= minimal.
Source: Gurr (1990).
95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 2
Comparative Index of Institutional Development (CUD)
Latin America and Caribbean, 1960-1986
1960-1969 1970-1979 1980-1986
Argentina 1.49944 1.38509 2.05066
Bolivia 2.24064 0.69048 1.98412
Brazil 1.48126 1.14643 2.41759
Chile 2.59061 1.30931 0.60011
Colombia 0.47855 1.96318 2.68323
Costa Rica 2.56159 1.09701 0.94485
Dominican Rep. 0.66585 1.98754 2.38086
Ecuador 1.40096 0.90150 2.88221
Guatemala 1.26210 2.22693 1.18711
Honduras 1.55841 0.90942 2.64595
Jamaica 2.67529 0.90500 1.05674
Mexico 2.63842 1.47729 0.29184
Nicaragua 2.32605 1.97554 0.02629
Panama 2.56643 0.73104 1.46076
Peru 1.88175 0.57824 2.65715
Paraguay 0.92341 1.63780 2.51256
El Salvador 1.76778 1.08547 2.09535
Uruguay 2.58876 1.03942 0.98831
Venezuela 1.38232 1.72805 1.72805
96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3
Tests for Conditional Convergence in Latin America
(Using data from Mankiw et al., 1992)
Constant 2.409 *
(2.266)
2.259
(1.980)
Ln (YGO) -0.455 *
(-3.965)
-0.438 *
(-3.553)
Ln (I/GDP) 0.492 *
(3.216)
0.5333 *
(2.942)
Ln (n+g+S) 0.054
(0.725)
0.051
(0.667)
Ln (SCHOOL) -0.014
(-0.460)
Adjusted R: 0.55 0.53
Observations 19 19
Note: Dependent variable is log difference GDP per working-age person 1960-
1985, t-statistics are in parentheses. YGO is GDP per working-age person in 1960.
Investment (I/GDP) and population growth (n) are averages for 1960-1985. (g+S)
is assumed to be 5 percent. SCHOOL is average percentage of the relevant
working-age population enrolled in secondary school, for 1960-1985.
* significant at the 5 per cent level.
97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 4
Descriptive Statistics and Data Sources (Latin America, 1960-1986, n=59)
Mean Std Dev Min Max Source
OLS Growth rate of real
GDP per capita
1.28 2.34 -3.48 6.85 Penn World
Tables 5.6
OLS Growth rate of
population
2.42 0.7418 0.3209 3.75 Penn World
Tables 5.6
Average rate of
investment
16.0721 4.8688 7.18 28.47 Penn World
Tables 5.6
Log of initial real GDP
per capita
7.8253 0.5072 6.9412 8.9539 Penn World
Tables 5.6
Population at start of
decade
14,102 23,716 1,145 121,286 Penn World
Tables 5.6
Index of institutional
development
1.6115 0.7347 0.0263 2.8822 Own
calculations
Growth rate of stock of
human capital
1.8592 1.2538 0.1152 5.3756 Barro and
Lee (1993)
Human capital at start
of decade
3.5758 1.2806 1.1550 6.6300 Barro and
Lee (1993)
Ratio of public exp. in
tertiary to primary plus
secondary
0.3222 0.2069 0.0303 1.4000 UNESCO
Statistical
Yearbooks
Gini coefficient for land
holdings
0.8503 0.0603 0.6940 0.9400 Adelman
and Fuwa
(1994)
Black market exchange
premium
0.3121 0.4965 0 2.9550 Barro and
Lee (1993)
Civil liberties 3.0818 1.7301 0.5000 6 . 0 0 0 0 Barro and
Lee (1993)
Revolutions 0.2175 0.3191 0 1.3 Barro and
Lee (1993)
Coups d'Etat 0.0713 0.1143 0 0.4000 Barro and
Lee (1993)
Rate of change of labor
force in agriculture
-1.545 1.066 -4.700 0.624 FAO
Yearbooks
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 5
OLS Results, Solow-MRW
(Dependent variable is OLS Growth of real GDP per capita)
(1 )
(2 ) (3) (4) (5) (6 )
Constant 0.408 ***
(2 .8 6 6 )
0.208784
(0.966)
0.181036
(0.811)
0.33565
(1.629)
0.41494 *
(1.904)
0.29024
(1.449)
ln (YO) -0.0497 **
(-2.574)
-0.0489 * *
(-2.613)
-0.0466 * *
(-2.416)
-0.064 ***
(-3.497)
-0.062***
(-3.172)
-0.061***
(-3.474)
In (I/GDP) 0.02445 * *
(2 .1 0 1 )
0.0237 * *
(2.008)
0.0262 **
(2.423)
0.0244 **
(2.343)
ln (n+g+5)
-0.047432
(-0.823)
-0.051800
(-0.883)
-0.03696
(-0.691)
-0.031783
(-0.555)
-0.04685
(-0.903)
HKGROWTH 0.001249
(0.595)
0.003686
*(1.836)
ln (CIID) 0.0077 **
(2.564)
0.0073 **
(2.260)
0.009 ***
(3.159)
Adjusted R: 0.5152 0.5528 0.5439 0.6174 0.5609 0.6438
Note: In parentheses are t-statistics. (YO) is GDP per capita at the beginning of
the decade. Population growth (n) is OLS growth rate, per decade. I/GDP is
average investment rate per decade. Following Mankiw et al., (g+(5) is assumed
to be 5 percent. HKGROWTH is the rate of change in average schooling years of
population over 25 years old (from Barro and Lee, 1993). CIID is author's index
of institutional development. Decade and country dummies included in
estimation (but not shown).
* Significant at the 10 per cent level.
** Significant at the 5 per cent level.
*** Significant at the 1 per cent level.
99
permission of the copyright owner. Further reproduction prohibited without permission
Table 6 .
Sample of 65 Developing Countries
ASIA (11) LATIN MIDDLE-EAST AND SUB-SAHARAN
AMERICA (19) N. AFRICA (10) AFRICA (26)
Bangladesh Argentina Algeria Benin
Indonesia Bolivia Egypt Burkina Faso
India Brazil Iran Botswana
Korea, South Chile Iraq Cameroon
Malaysia Colombia Jordan Central Africa Rep.
Myanmar Costa Rica Kuwait Ghana
Pakistan Dominican Rep. Morocco Gambia
Philippines Ecuador Syria Guinea Bissau
Singapore El Salvador Tunisia Kenya
Sri Lanka Guatemala Turkev
>
Liberia
Taiwan Honduras
Jamaica
Mexico
Nicaragua
Panama
Peru
Paraguay
Uruguay
Venezuela
Lesotho
Malawi
Mali
M auritius
Mozambique
Niger
Rwanda
Sudan
Senegal
Sierra Leone
South Africa
Togo
Tanzania
Zaire
Zambia
Zimbabwe
100
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 7
OLS Results, Solow-MRW
(Dependent variable is OLS Growth of real GDP per capita)
(1 )
(2 ) (3) (4) (5) (6 )
Constant 2.7775 ** 7.3623 *** 7.354 * ** 11.415 * ** 9.5667 *** 11.03 ***
(1.829) (3.522) (2.990) (4.075) (3.140) (3.248)
In (YO) -0.04169 -0.5654 ** -0.667 * ** -1 . 0 0 0 * ** -1.018 *** -1.18 ***
(-0.195) (-2.154) (-2.955) (-3.817) (-3.966) (-3.889)
ln (I/GDP) 1.736 * ** 1.536 *** 2.1281 *** 1.841 ***
(7.122) (6.383) (6.959) (6.057)
ln (n+g+5)
-2 . 1 2 * -2.2581 * -2.4858 ** -2.04597
(-1.745) (-1.919) (-1.699) (-1.443)
HKGROW 0.03859 0.00043
(0.548) (0.006)
ln (CIID)
Dummy for 0.73592 0.28061 0.83746
Asia (1.226) (0.496) (1.326)
Dummy for -0.77664 -1.0905 * * -0.58387
LAC (-1.408) (-2.097) (-0.979)
Dum m y for -1.9063 * -1.76075 * -1.187 *
Africa (-3.322) (-3.269) (-1.755)
Dummy for -0.12909 0.01019 -0.23585 -0.10563 -0.14860 -0 . 0 2 0 0 0
1970s (-0.300) (0.025) (-0.590) (-0.273) (-0.332) (-0.046)
Dum m y for -2.91 * ** -2.641 *** -2.665 ** * -2.483 *** -2.528 *** -2.43 ***
1980s (-6.660) (-6.248) (-6.512) (-6.193) (-5.564) (-5.418)
Adjusted R 2 0.1778 0.2598 0.3011 0.3531 0.3534 0.3997
Note: In parentheses are t-statistics. (YO) is GDP per capita at the beginning of
the decade. Population grow th (n) is OLS growth rate, per decade. l/GDP is
average investment rate per decade. Following MRW (g+i5 ) is assum ed to be 5
percent. HKGROWTH is the rate of change in average schooling years of
population over 25 years old (from Barro and Lee, 1993). CIID is author's index
of institutional development.
* Significant at the 10 per cent level.
** Significant at the 5 per cent level.
*** Significant at the 1 per cent level.
101
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 7 (cont.)
OLS Results, Solow-MRW
(Dependent variable is OLS Growth of real GDP per capita)
(7) (8 ) (9) (1 0 ) (1 1 )
(1 2 )
Constant 7.907 *** 13.252 ** * 5.022 * 12.169 *** 8.859 * * * 12.32***
(2.907) (4.374) (1.719) (3.796) (2.802) (3.456)
ln (Y O ) -0.855 *** -1.312 * * * -0.2715 - 1 . 0 0 2 *** -1 . 2 0 1 * * * -1.48 ***
(-3.415) (-4.425) (-1.063) (-3.242) (-4.460) (-4.431)
ln (I/GDP) 1.8092*** 1.4926 * * * 2.0338 * * * 1.67 ***
(6.768) (5.593) (6.485) (5.311)
ln (n+g+5)
-2.01345 -2.03937 -0.74647 -0.94753 -1.78366 -1.6024
(-1.466) (-1.546) (-0.505) (-0.685) (-1.176) (-1.094)
HKGROW 0.08534 0.03871
(1.098) (0.481)
In (CIID) 0.37553 0.36770 0.9067 ** 0.69271 * 0.8186 * 0.769 *
(0.946) (0.947) (2.143) (1.702) (1.908) (1.829)
Dummy for -0.08975 0.23322 0.43659
Asia (-0.142) (0.348) (0.628)
Dummy for -1.2854 * * -1.1181 * * -0.80158
LAC (-2.418) (-1.987) (-1.338)
Dummy for -2.323 * * * -2.7875 * * -1.726 *
Africa (-4.002) (-4.576) (-2.420)
Dummy for -0.01759 0.14108 0.26929 0.39676 0.07125 0.22188
1970s (-0.040) (0.330) (0.560) (0.880) (0.151) (0.478)
Dummy for -2.405 *** -2.185 * * * -2.452 *** -2.145 *** -2.225 * * * -2.09 ***
1980s (-5.332) (-4.979) (-5.002) (-4.610) (-4.660) (-4.407)
Adjusted R: 0.3028 0.3650 0.1758 0.2859 0.3624 0.4091
Note: In parentheses are t-statistics. (YO) is GDP per capita at the beginning of
the decade. Population growth (n) is OLS grow th rate, per decade. I/GDP is
average investment rate per decade. Following MRW (g+5) is assumed to be 5
percent. HKGROWTH is the rate of change in average schooling years of
population over 25 years old (from Barro and Lee, 1993). CIID is author's index
of institutional development.
* Significant at the 10 per cent level.
** Significant at the 5 per cent level.
*** Significant at the 1 per cent level.
102
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 8
OLS Estimation Results, CUD equation
(Dependent variable is Log of CIID;
Number of observations is 57=3x19)
(1 )
(2 ) (3) (4) (5)
Constant 0.508802
(0.943)
-10.997943
(-1.520)
-7.130958
(-0.992)
0.253676
(0.033)
2.06081
(0.267)
BMP -0.8207 ***
(-3.118)
-0.715945 * *
(-2.674)
-0.767 ***
(-2.982)
-0.80655 ***
(-3.306)
-0.782 ** *
(-3.154)
CIVLIB 0.175158
(1.407)
0.038966
(0.273)
0.001829
(0.013)
0.048117
(0.364)
0.124254
(0.979)
REVCOUP -2.349367
(-1.331)
-2.760597
(-1.567)
-2.795151
(-1.673)
-2.67501
(-1.693)
ln (Y O ) 1.362371
(1.401)
1.68551 *
(1.769)
1.585619 *
(1.758)
1.399613
(1.537)
POPO -0.0000038
(-0.174 )
0.0000079
(0.369)
-0 . 0 0 0 0 1 1
(-0.491)
-0 . 0 0 0 0 1
(-0.460)
HKO 0.557943
(1.414)
0.67181 *
(1.721)
0.295608
(0.717)
0.214968
(0.527)
INEQT_PS 0.091166
(0.124)
-27.99896 *
(-2.044)
-29.426 * *
(-2.123)
Gini Land -0.0821 **
(-2.429)
-0.14271 ***
(-3.280)
-0.147 * * *
(-3.350)
PCTCHGAG -0.896798
(-0.696)
-0.63082
(-0.515)
INEQT.PS x
Gini Land
0.318286 *
(2.053)
0.3339 * *
(2.129)
Adjusted R: 0.0638 0.0841 0.1836 0.2704 0.2443
Note: Numbers in parentheses are t-statistics. CIID is authors' index of
institutional development. BMP is the black market exchange rate premium.
CIVLIB is Gastil's index of civil liberties.INEQT_PS is the ratio of public
expenditure in tertiary education to public expenditure in primary and
secondary education. PCTCHGAG is rate of change of percentage of labor force
in agriculture. Decade and country dummies included in estimation (but not
shown).
* Significant at the 10 per cent level.
** Significant at the 5 per cent level.
*** Significant at the 1 per cent level.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 9: 2SLS Estimation Results
(Number of observations is 57=3x19)
Growth GDP
per capita
CIID Grow th GDP
per capita
CIID
2SLS 2SLS 2SLS 2SLS
(1 )
(2 ) (3) (4)
Constant 0.305327
(1.478)
2.060810
(0.267)
0.231133
(1.132)
12.954153 ***
(3.564)
ln (YO) -0.068274 ***
(-3.578)
1.399613
(1.537)
-0.055116 ***
(-3.089)
ln (I/GDP) 0.021503
(2.008)
0.020837 **
(1.927)
ln (n+g+5) -0.062904
(-1.183)
-0.054306
(-1.016)
HKGROWTH 0.004536
(2.033)
0.004590 * *
(2.026)
ln (CIID) 0.013377 ***
(2.792)
0.013326 **
(2.738)
BMP -0.78124 ***
(-3.154)
-0.84114 ***
(-3.509)
CIVLIB 0.124254
(0.979)
0.19919 *
(1.850)
INEQT.PS -29.42456 * *
(-2.123)
-32.2538 **
(-2.714)
Gini Land -0.14727 ***
(-3.350)
-0.1472 ***
(-3.504)
INEQT_PS x
Gini Land
0.33387 * *
(2.129)
0.3642 * *
(2.717)
HKZERO 0.214968
(0.527)
POPO -0.000010488
(-0.460)
Adjusted R; 0.6266 0.2443 0.6216 0.2519
Note: In parentheses are t-statistics.
* Significant at the 10 per cent level.
** Significant at the 5 per cent level.
*** Significant at the 1 per cent level.
104
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
DATA APPENDIX FOR CHAPTER 3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A.l
Economic growth and investment in Latin America: 1960-1989
OLS Growth of
Real GDP per capita
(% per year)
Investment / GDP (%)
1960-
1969
1970-
1979
1980-
1989
1960-
1969
1970-
1979
1980-
1989
Argentina 2.04 0.75 -2.29 16.23 18.80 14.65
Bolivia 3.72 2.45 -2 . 0 1 20.03 23.22 7.23
Brazil 1 . 8 8 5.60 0.49 18.80 22.46 16.95
Chile 2.28 -1.47 1.42 20.97 16.86 20.95
Colombia 1.91 3.07 1.32 16.27 15.59 15.50
Costa Rica 3.23 3.07 0.16 13.95 17.50 16.93
Dominican Rep. 1.87 3.29 -0 . 1 2 9.76 18.13 17.56
Ecuador 3.49 6.85 -1.65 1.15 25.55 19.54
Guatemala 1.69 2.91 -1.92 8.71 10.67 7.96
Honduras 1.95 2.59 -0.62 4.12 15.50 12.09
Jamaica 4.63 -0.91 0.52 28.47 22.71 14.90
Mexico 3.77 3.26 -1.09 15.62 17.96 15.99
Nicaragua 4.63 -0.82 -3.92 11.13 10.46 12.62
Panama 4.68 1.16 -2.16 19.73 25.83 15.72
Peru 2.85 0.53 -2 . 2 1 18.21 16.33 18.39
Paraguay 1.64 4.64 -2.33 8.15 13.52 17.86
El Salvador 3.20 2.90 -0.30 8.23 9.67 7.14
Uruguay -0.48 1.62 -0.42 9.78 14.91 13.64
Venezuela 2.08 0.94 -1.65 15.50 2 2 . 6 6 15.37
Sources: Data for investment and real GDP per capita are from Penn World
Tables Mark 5.6. Ordinary least-squares growth rates calculated by author.
106
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A.2
Income Inequality in Latin America: 1960-1980
Ratio of Bottom 409c to Top
2 0 %
Gini coefficient
1960 1970 1980 1960 1970 1980
(1 )
(2 ) (3) (4) (5) (6 )
Argentina 0.326 0.280 0.335 0.437 0.411 0.408
Bolivia na na 0.195 na na 0.516
Brazil 0.105 0.105 0.104 0.603 0.641 0.594
Chile na 0.261 0.167 na 0.455 0.530
Colombia 0.249 0.179 0.239 0.562 0.563 0.585
Costa Rica 0 . 2 2 2 0.219 0.258 0.521 0.445 0.475
Dominican Rep na 0.228 0.218 na 0.493 0.433
Ecuador 0.142 0.109 0.279 0.678 0.683 na
El Salvador 0.191 0.328 0.278 0.546 0.465 na
Guatemala 0.559 0.275 0.125 0.299 0.497 0.480
Honduras 0.108 na 0.137 0.618 na 0.549
Jamaica 0.136 na 0.297 0.577 0.413 0.432
Mexico 0.177 0.172 0.213 0.539 0.583 0.480
Nicaragua na na na na na na
Panama 0.242 0.116 0.139 0.500 0.448 0.487
Paraguay na na 0.274 na na 0.451
Peru 0.132 0.115 0.324 0 . 6 6 6 0.5941 0.427
Uruguay 0.320 na 0.296 0.428 na 0.436
Venezuela 0.163 0.191 0.288 0.545 0.477 0.428
107
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A.3
Percentage of population in areas oficially defined as urban:
Latin America: 1960-1990
1960 1965 1970 1975 1980 1985 1990
Argentina 74 76 78 81 83 85 8 6
Bolivia 39 40 41 42 44 48 51
Brazil 45 50 56 61 6 6 71 75
Chile 6 8 72 75 78 81 84 8 6
Colombia 48 54 57 61 64 67 70
Costa Rica 36 38 40 41 43 45 47
Dominican
Rep.
30 35 40 45 51 56 60
Ecuador 34 37 40 42 47 52 56
El Salvador 38 39 39 40 42 43 44
Guatemala 32 34 36 37 37 38 39
Honduras 23 26 29 32 36 40 44
Jamaica 33 38 42 44 47 49 52
Mexico 51 55 59 63 6 6 70 73
Nicaragua 40 43 47 50 53 57 60
Panama 39 44 48 49 50 51 53
Paraguay 36 36 37 39 42 44 48
Peru 46 52 57 61 65 67 70
Uruguay
79 81 82 83 84 85 8 6
Venezuela 67 70 72 78 83 82 84
Source: W orld Bank, Social Indicators of Development 1994.
108
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A.4
Percentage of total labor force in agriculture:
Latin America, 1960-1990
1 1960 1 1965 1970
1975 | 1980 | 1985 | 1990
Argentina 2 0 . 0 18.2 16.4 14.5 13.0 1 1 . 6 10.4
Bolivia 59.1 58.2 55.5 49.2 46.4 43.9 41.5
Brasil
51.9 48.8 45.6 37.9 31.2 27.6 24.3
Chile
29.2 26.9 23.8 19.7 16.4 14.3 12.5
Colombia 51.5 44.5 37.9 36.5 34.0 30.5 27.3
Costa Rica 51.6 46.7 42.2 36.6 30.8 27.1 23.8
Dominican
Rep.
6 6 . 6 63.8 61.2 50.3 45.8 40.7 35.8
Ecuador 57.5 54.1 50.9 44.5 38.6 34.3 30.3
El Salvador 61.6 58.9 56.1 49.2 42.5 39.7 36.5
Guatemala 66.9 1 63.8 61.0 59.1 56.8 54.0 51.2
Honduras 70.1 68.4 66.5 62.7 60.5 57.8 55.0
Jamaica 39.0 34.2 29.5 32.1 31.4 29.2 27.0
Mexico 55.1 50.3 45.2 40.3 36.5 33.2 30.0
Nicaragua 61.8 56.9 51.3 49.1 46.6 42.4 38.4
Panama 50.8 46.1 41.4 36.6 31.7 28.2 25.0
Paraguay 56.0 54.5 52.6 50.7 48.8 47.5 46.2
Peru 52.5 48.7 44.8 43.5 40.0 37.3 34.7
Uruguay 20.7 17.8 15.2 17.1 15.7 14.5 13.5
Venezuela 35.1 30.3 25.6 20.9 16.0 13.3 1 1 . 0
Source: Data for 1960 is from FAO Production Yearbook, 1973, vol. 27; for 1965 is
from FAO Production Yearbook, 1977, vol. 31; for 1970 is from FAO Production
Yearbook, 1982, vol. 36; and for 1975-1990 is FAO Production Yearbook, 1990,
vol.44.
109
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A.5
Gini coefficients for land holdings in Latin America: 1950-1980
Circa 1950 Circa 1960 Circa 1970 Circa 1980
Argentina 86.3 1952 86.7 1960 85.3 1970 na na
Bolivia 93.8 1950 93.8 1950 na na na na
Brazil 83.5 1950 84.5 1960 83.69 1972 83.7 1989
Chile 93.8 1936 na na 93.31 1968 93.3 1989
Colombia 84.9 1954 86.4 1960 85.81 1974 85.8 1988
Costa Rica 89.2 1950 78.2 1963 82.59 1971 82.6 1989
Dominican Rep. 79.5 1950 80.3 * 1960 79** 1971 82.1 1989
Ecuador 86.4 1954 86.4 1954 82.07 1970 na na
El Salvador 82.8 1950 82.7 1961 82.69 1976 81** na
Guatemala 8 6 1950 8 6 1950 86.95 1979 87 1989
Honduras 75.7 1952 75.7 1952 77.62 1967 77.6 1989
Jamaica 82 1943 77 1960 79.8 ** 1970 80.3 1990
Mexico na na 69.4 1960 93.76 1960 93.8 1984
Nicaragua 75.7 1950 80.1 1963 na na na na
Panama 73.7 1961 73.7 1961 80.19 1973 80.2 1989
Paraguay na na na na na na
9 4 ***
1981
Peru 87.5 1950 93.3 1961 91.17 1972 91.2 1985
Uruguay 81.7 1950 82.6 1961 81.61 1967
8 4 ***
1980
Venezuela 90.9 1956 90.9 1956 91.13 1970 91.1 1989
India 52.2 64 61.96 62
Kenya 69.2 67.4 67.4
Philipines 56.4 53.4 50.98 51
South Korea 38.7 31.39
Sources: Source for column 1 is World Handbook of Social and Political
Indicators (1964,1st edition), for column 2 is Persson and Tabellini (1992), and for
columns 3 and 4 is Adelman and Fuwa (1994). If (*) source is World Handbook
(1972, 2nd edition), if (**) it is World Handbook (1983, 3rd edition), and if (***) it
is Statistical Abstract of Latin America (1990, vol. 28).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A.6
Educational Expenditures in Latin America: 1960-1990
Expenditures in
education as % GDP
Education as percentage of
central government
expenditures
1960 1980 1990 1970 1980 1985 1990
(1) (2) (3) (4) (5) (6) (7)
Argentina 2.1 3.6 3.4 11.1 11.3 8.5 na
Bolivia 1.5 4.4 3.0 32.3 32.1 19.8 20.1
Brazil 1.9 3.6 4.6 11.5 4.2 2.2 na
Chile 2.7 4.6 3.7 na 11.9 na 12.0
Colombia 1.7 1.9 2.9 16.9 26.3 24.4 21.4
Costa Rica 4.1 7.8 4.6 26.8 32.4 22.5 20.8
Dominican Rep. 2.1 na na 14.4 12.5 11.5 na
Ecuador 1.9 5.6 2.8 na 37.1 24.5 19.1
El Salvador 2.3 na 1.8 23.6 19.3 16.2 na
Guatemala 1.4 1.9 1.4 16.6 12.1 12.9 11.8
Honduras 2.2 3.2 4.6 20.3 14.8 19.1 15.9
Jamaica 2.3 na 6.1 na na na 12.9
Mexico 1.2 4.7 4.1 16.9 20.0 10.6 7.4
Nicaragua 1.5 na na 19.2 12.8 12.0 na
Panama 3.6 5.2 5.5 20.6 16.0 21.3 15.6
Paraguay 1.3 1.5 1.0 13.9 14.1 12.2 12.1
Peru 2.3 3.1 3.5 21.4 15.3 16.1 22.8
Uruguay 3.7 2.3 3.1 15.8 13.4 8.4 15.9
Venezuela 3.7 4.4 4.1 16.2 20.3 18.6 18.8
Sources: Source for columns 4, 5 and 6 is Cardoso and Helwege (1992). For
colums I, 3 and 7 is UNDP, Human Development Report 1994. For column 2 is
Morley (1995).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A.7
Illiteracy and School Enrollment Rates in Latin America
Illiteracy rates Gro
schoo
ss priir
enrol
tary
ment
Gross secondary
school enrollment
Circa
1970
Circa
1980
1985 1990 1960 1970 1980 1960 1970 1980
Argentina 7 6 5 5 98 105 106 23 44 59
Bolivia na 37 28 23 64 76 84 12 24 36
Brazil 34 26 22 19 95 na 99 11 26 34
Chile 11 9 8 7 109 107 109 24 39 53
Colombia 19 15 15 13 77 108 128 12 25 44
Costa Rica 12 na 8 7 96 110 105 21 28 48
Dominican
Rep.
33 na 20 17 98 100 114 70 21 43
Ecuador 26 20 17 14 83 97 113 12 22 51
El Salvador 43 33 31 27 45 57 71 70 80 18
Guatemala 54 na 48 45 67 87 93 80 14 30
Honduras 43 na 32 27 92 119 101 45 46 59
Jamaica 4 na 2 2 80 104 115 11 22 46
Mexico 26 17 15 13 66 80 99 70 18 43
Nicaragua 43 13 na na 96 99 106 29 38 61
Panama 22 14 14 12 83 107 114 15 31 59
Paraguay 20 13 12 10 110 117 na 63 71 na
Peru 28 18 18 15 80 85 75 13 22 24
Uruguay 6 na 5 4 111 112 106 37 59 60
Venezuela 24 15 14 12 100 94 109 21 33 41
Sources: Data for illiteracy rates are from World Bank, Social Indicators of
Development 1994. Data for gross enrollments are from King and Levine (1993).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table A.8
Percentages of public current expenditure by level of education
in Latin America, 1965-1985.
1965
1975 1985
Prim Sec Ter Prim Sec Ter Prim | Sec | Ter
Argentina 49.8 23.9 16.9 27.0 30.5 30.2 5.5 43.3 30.8
Bolivia 56.4 16.3 17.9 60.3 7.2 15.0 56.2 10.1 23.2
Brasil * 33.5 19.6 20.1 45.4 10.9 22.8 45.9 7.7 19.6
Chile 35.7 17.9 25.6 34.9 13.5 25.2 51.0 19.5 20.3
Colombia 39.6 13.2 24.5 40.5 21.9 24.0 39.2 30.8 22.2
Costa Rica 60.4 16.7 11.5 37.2 22.4 24.4 35.1 22.3 41.4
Dominican Rep. 44.8 15.6 25.4 37.5 20.4 19.6 46.6 19.7 20.8
Ecuador 41.3 21.1 32.3 40.4 29.5 24.0 45.3 35.8 17.8
El Salvador * * 66.6 9.5 11.6 57.5 6.6 23.7 61.9 6.2 14.2
Guatemala 67.0 22.4 3.0 51.3 15.5 19.9 38.2 17.2 19.7
Honduras 71.1 15.2 10.3 68.4 9.6 16.5 49.1 16.7 21.3
Jamaica 48.9 27.7 9.9 33.5 36.9 19.2 29.9 34.0 19.4
Mexico 40.3 12.5 12.7 42.9 31.1 12.6 27.4 26.8 17.6
Nicaragua 60.9 17.3 7.1 55.1 22.4 16.5 43.3 16.7 23.2
Panama 51.6 23.5 8.9 41.1 20.7 12.9 38.7 25.2 20.4
Paraguay * * * 54.2 17.4 20.0 64.8 17.1 16.5 36.6 29.7 23.8
Peru 41.8 18.1 13.5 j 54.6 27.9 2.5 35.6 20.5 2.7
Uruguay 44.9 39.8 15.3 46.7 31.8 17.5 37.7 28.4 22.4
Venezuela 43.2 18.2 19.7 29.3 21.4 34.1 24.5 6.5 43.4
Notes: "Prim" is primary, "Sec" is secondary,and "Ter" is terciary education. If
(*), data is for 1960 (instead of 1965), if (**), it is for 1980 (instead of 1985), and if
(***), it is for 1960 (instead of 1965), and 1970 (instead of 1975).
Source: Data for 1965 and 1975 is from UNESCO, Statistical Yearbook, 1978-1979.
Data for 1985 is from UNESCO, Statistical Yearbook, 1994.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 4.
On The Causal Relationships Between Political Instability,
Institutional Development and Economic Growth
The purpose of this chapter is to investigate the relationship between
socio-political instability (SPI), institutional development, and economic growth.
We first study the causal relationships between socio-political instability (SPI), on
one hand, and economic growth, investment, and hum an capital, on the other,
for a large panel of developing countries. Later, w'e focus on the consequences of
bringing institutional development into this framework.
The effects of SPI on grow th occur because SPI fuels social unrest, disrupts
productive activities, and increases the overall uncertainty in the economy. In
doing so, it undermines the incentives for the accumulation of physical capital,
and reduces the rate of economic growth. On the other hand, poor economic
performance can lead to socio-political unrest. The standard view is that a
feedback system characterizes this relationship: a higher level of SPI decreases
the rate of economic growth, and a diminishing rate of economic growth
increases the level of SPI. Moreover, distilling the lessons from the recent
empirical literature on economic growth, Mankiw argues that one very robust
114
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
finding is that "political instability, as measured by the frequency of revolutions,
coups, or wars, is negatively associated with growth" (1995, 302).
The SPI literature in economics is thriving but young.1 There are still a
number of issues waiting to be investigated. In this chapter, we focus on three of
them: (1) the definition and measurement of SPI, (2) the direction of causality
between SPI and economic performance (the latter encompassing growth,
investment and hum an capital), and (3) the relationship between SPI and
institutional development.
On the first issue, that of definition and measurement, notice that there are
two different understandings of what is meant by socio-political instability. The
first focuses on regular and irregular changes in the composition of the
government, and the second, on harsher aspects, such as revolutions, coups
d'Etat, civil wars and political assassinations. That they overlap (both include
unregulated governm ent changes) does not diminish the suspicion that they
attribute rather different intensities to "instability". While the former constrains it
to relatively tame phenomena, the second places it closer to social chaos.
A related concern is that of the time horizon underlying the measurement
of SPI. It is not uncommon to find studies that proxy SPI by some weighted 20-
year average of the num ber of revolutions, coups and assassinations. We
question whether such long term averages adequately reflect instability. Instead,
1 Various empirical studies dealing with social and political instability were reviewed previously,
in Chapter 2. See also Alesina and Perotti (1996) for a review of this specific literature.
115
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
we conjecture that these results can be better understood as insufficient
accumulation of institutional capabilities, which might be only partly attributable
to SPI. Moreover, here we hypothesize that the negative effect of SPI on
investment, and hence growth, should be even stronger if (as w ould seem indeed
more appropriate to the concept of instability) the data used in such tests were
for shorter periods of time.
On the second issue, that of causality between SPI and grow th, we begin
by acknowledging that the literature, in general, recognizes that causality does
run in both directions. In empirical studies, this shows in the ample use of
simultaneous equations methods. But perhaps because most of these studies
choose to measure SPI by 20 or 25 years averages, it is difficult to find some
direct tests for causality. Below, we use the Granger framework to explore this
relationship in closer detail.
A related issue is the possibility that some aspects of the causality
structure have not been fully spelled out. For instance, the literature is
compelling in pointing out that SPI decreases growth because of its effects on the
accumulation of physical capital. Surprisingly however, one can find very little
on the equally important possibility that SPI is detrimental to grow th because of
its effects on the accumulation of hum an capital. In its most extreme form (a civil
war), SPI at least accelerates the depreciation of the stock of hum an capital.
116
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The last issue is regarding the relationship between SPI and institutional
development. For those unfamiliar with these two literatures, it has to come as a
surprise that they have very much evolved separately. Here we discuss a
possible (and perhaps the simpler) connection: that high levels of SPI are
associated with low levels of institutional development.
To try to overcome these deficiencies, we use two different measures each
representing a different extreme of the definition of SPI, in particular, a m easure
of "severe" SPI and one of "moderate" SPI. Indexes of each such measure are
constructed by country for each five year period between 1960 and 1995 for 98
LDCs. The use of periods as short as 5 years allows us to investigate the effects of
lagged variables and hence to undertake proper causality tests. We investigate
the links between SPI and long-term economic growth, investment and hum an
capital.
The argum ent is organized as follows. In Section 1 we discuss
methodological m atters with special emphasis on the different indexes of socio
political instability. In section 2 we explore causality issues (using the Granger
framework) for the aggregate sample and for each of four separate geographical
regions. In section 3 we subject these results to sensitivity analysis (inter alia, we
investigate whether incorporating an index of institutional development alters
the results.) We derive two main conclusions. First, in the case of our upper-
bound index (which is much similar to other existing treatments of SPI), we
117
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
found that it does not cause economic growth. We find that causality flows in
the other direction: low growth increases the level of SPI. On this basis, we claim
that the Olsonian effect dominates: rapid economic growth involves substantial
structural change, which may undo political coalitions and readjust the relative
power of different interest groups (Olson, 1963.) Second, our results raise a
strong suspicion that the sub-sample of Sub-Saharan Africa countries is the
major, and sometimes sole, driving force behind the empirical results and policy
implications derived in the existing literature on socio-political instability.
Section 4 summarizes the main results and provides suggestions for future
research.
1. M ethodological and data issues
The objective of this section is to discuss the details in the construction of
the SPI indexes, and to present the methodology. We construct two different
measures of SPI, one capturing the traditional severe forms of SPI and the other
based on less severe forms of SPI. While there are clearly many other variants
that could have been explored, it is believed that the two measures capture the
extremes. Hence, if the results should turn out to be similar for the two different
measures, it might well imply that other measures of SPI would be likely to yield
118
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
similar effects.2 The difficulties in measuring a less severe or lower-bound SPI can
be illustrate by considering that just as accurate earthquake measurements
require measurements by devices located near the earthquake and far from the
earthquake, the use of a precise measure to pick up the severe cases of SPI can
complement that of a course measure designed to pick up milder SPI shocks.
While some analysts have sought to combine the two types of indicators
into a single index, since the effects of the two types of instability might well
differ, we choose to keep them separate. For indicators of the measure of severe
SPI, we follow Perrotti (1994) and Alesina and Perrotti (1996) in the use of the
following three indicators: political assassinations, revolutions and coups. These
measures are all taken from Barro and Lee (1993). For indicators of the mild type
of SPI we follow Chen and Feng (1996) and others in the use of indicators (in our
case seven) from Polity III (Jaegger and Gurr, 1996). These indicators are: the
competitiveness of political participation (PARCOMP), the regulation of political
participation (PARREG), the regulation of executive recruitment (XRREG), the
competitiveness of executive recruitment (XRCOMP), the openness of executive
recruitment (XROPEN), the legal (de jure) independence of the chief executive
(MONO), and the operational (de facto) independence of the chief executive
; On the other hand, if the results differ according to the indicator, it would signal that more
attention should be given to some of the alternative measures.
119
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(XCONST).3 Key advantages of these m easures are that they are available for a
large num ber of countries (157), most of them annually since 1960, and that they
are capable of picking up rather subtle changes in both legal and actual practice
changes in the extent to which political actors and processes are subject to
systematic regulation.4 Naturally, the less regulated are such actors and
processes, the greater the potential for social and political change and hence the
higher the value of SPI.
For each of these measures as well as for per capita GDP grow th (OLS
rate), time series data covering the period 1960-1995 are collected for an
unbalanced panel of 98 LDCs listed in Table 1.’ Included are 14 countries from
Asia, 20 from Latin America, 16 from the Middle East and North Africa and 38
from Sub-Sahara Africa.
The duration of the periods used for computing SPI indicators has varied
widely from annual data used in some of the more recent studies using
simultaneous equation methods with panel data to 25 year periods in some of the
pure cross sectional studies. As mentioned before, 25 year periods would seem
far too long to be useful in capturing instability (an essentially short run dynamic
It is dear that a more appropriate lower-bound measure of SPI would include strikes,
demonstrations without violence nor deaths, regional and internal conflicts, free press, etc..
However, data on such variables are not available.
4 Since in Polity III (Jaegger and Gurr, 1996), these indicators are assigned high scores when the
extent of regulations is higher (implying lower SPI), for present purposes the coding has been
reversed.
120
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
measure.) On the other hand, annual data would seem too short in duration to
reflect underlying factors rather than mere transitory phenomena of productive
time lost due to the political changes themselves. For this reason, in this study we
settle on five year, non-overlapping periods, where the observations represent
averages in the respective indexes and real per capita GDP growth rates.
For each of these two measures or boundaries, we construct indexes of SPI
for each five year period through the principal components m ethod (for each
region). For the severe or upper bound SPI (UBSPI) indicator, the weights
resulting from this procedure are: assassinations 0.3162, revolutions 0.6909, and
coups 0.6502. In the case of the mild or lower bound SPI (LBSPI), the weights are:
0.3923 for PARCOMP, 0.1105 for PARREG, 0.4677 for XRREG, 0.4734 for
XRCOMP, 0.3535 for XROPEN, 0.2317 for MONO and 0.4608 for XCONST.
Since both indexes are measures of SPI, one would expect them to be
positively correlated. Yet, since both attem pt to capture rather different aspects
or types of SPI, the correlation between them would not be expected to be very
high. In general, these expectations are fulfilled, except for the M iddle East and
North Africa region where the coefficient is -0.027 (although it is not statistically
significant.) For the other regions, the correlations between the respective pairs
of SPI indexes are all positive and significant at the one percent level.0
' The reason for choosing an unbalanced panel was to keep the exerdse as comparable to the rest
of the literature as possible. The sample we use differs from other studies by very few countries.
121
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
We use the Granger framework to test for the existence and direction of
causality. In other words, with the use of symmetric lags, we attem pt to
determine whether causality is unidirectional, going either from SPI to long-term
growth or from long-term growth to SPI, or if it is two-way. As in the other
studies which attem pt to deal with potential simultaneity, we use panel data
consisting of pooled time series and cross section analysis.* In order to capture
the original uncertainty notion of SPI, the time periods are non-overlapping five
year periods instead of the more usual 25 years period.''
Finally, to allow for the fact that SPI might be determined somewhat
differently in some regions than in others and that the effects of SPI on growth
" The correlation coefficients between our severe and moderate SPI indexes (1960-1995) are the
following: Asia (0.284), Latin America (0.535), Sub-Saharan Africa (0.319), and "All LDCs"
(0.319). Notice that these coefficients are all significantly different from zero at the 1 percent level.
However, for the Middle East and North Africa group its value is (-0.027), although it is not
significantly different from zero (at the 10 percent level.) We hypothesize that this is because of
the interactions between internal and international instability (wars, border closings, etc).
The Granger framework has endured the test of time because of its elegance and strong intuitive
appeal (the notion that an event in the future cannot cause one in the past). Yet, "causation is a
non-symmetric relationship, and there are various ways in which asymmetry can be introduced,
the most important which are controllability, a relevant theory, outside knowledge, and temporal
priority" (Granger, 1987, 49.) For discussion see, among others. Granger (1987), Hsiao (1979), and
Zellner (1988).
Holtz-Eakin et al.(1988) argue that the "pooling of cross-sectional units does have certain
advantages. First, the assumption of time stationarity can be relaxed. The presence of a large
number of cross-sectional units makes it possible to allow for lag coefficients that vary over time.
Second, the asymptotic distribution theory for a large number of cross-sectional units does not
require the vector autoregressions to satisfy the usual conditions that rule out unit and explosive
roots" (p. 1373.)
“ While this greatly reduces the number of observations relative to the large numbers used in
some of the other studies, it also reduces the need for large numbers of own lags which have been
found to be the most significant variables in these studies, hence partially compensating for loss
of degrees of freedom in the estimation process.
122
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
might be different in different regions, we conduct these tests both in the sample
as a whole and separately by region.
A note on the Granger-causality methodology. Consider two time series,
x. and y„ series x, is said to Granger-cause series y. if in a regression of y, on
lagged y's and lagged x's, the coefficients of the lagged x's are jointly
significantly different from zero. More formally,
(1) y = a + Y. y~-L + X '{. x; _. * a.
And x is said to Granger-cause y if all ;/ are jointly significantly different
from zero. In what follows, we use only one lag on the variable .r. Instead of
conducting an F-test for joint significance, the test reduces to the significance of
the coefficient on x (a f-test).
There are two critical issues in applying Granger causality tests.1 0 The first
is the nature of the time lags, that is, their length and frequency. On their length,
Granger admonishes that "using data m easured over intervals much w ider than
actual causal lags can also destroy causal interpretation" (Granger, 1987, p.49).
As for their frequency, or the number of lags used in the statistical test, it should
1 0 We do not know of other studies that use the Granger framework in the instability-growth
context. However, examples of recent studies that use Granger-causality testes in different
contexts are Bahmani-Oskooee et al.(1991), Blomstrom et al. (1996), Conte and Darrat (1988),
Masih and Masih (1995), and Oxley (1994).
123
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
be noted that it is difficult to escape the fact that this choice is arbitrary, usually
to some large extent.
The second crucial issue in applying Granger causality tests, is in regard
to the information set. The Granger test depends on the assumption that the
cause contains unique information about the effect, that is not available
elsewhere. If the information set underlying the test is composed solely of two
series, both of which m ay be affected by a third variable, the test can be rendered
useless.
In light of these concerns, the results presented below are to be
understood as complementary to the ones in the existing literature (reviewed
above) and as descriptive of the behavior of the time series involved. In w hat
follows, we present our Granger-causality results for the relationships between
SPI and economic growth, SPI and investment, and SPI and human capital.
2. The Pattern of Causality
We start by investigating the causality patterns between SPI and economic
growth. Preliminarily, and consistent with other studies, we find a negative
contemporaneous two-way relation between the severe SPI index and economic
growth (as shown in table 3, such a relationship does not obtain for the lower
bound SPI index.) We interpret these as suggesting that the 5-year period is just
124
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
too long a lag to capture the relation between severe SPI and long-term growth.
In other words, there seems to be no relation between severe SPI and long-term
growth: severe SPI has only a short-term effect."
We turn then to Granger causality, properly speaking. One important
result in this context is the differences engendered by the two SPI indexes. For
example, from table 4, there is a strong indication that m oderate SPI Granger-
causes per capita GDP growth (i.e., lowers it significantly), but the same is not
true in the case of the severe SPI measure.1 2
When we break down these results by regions, we find that the
statistically significant notion that moderate SPI Granger-causes per capita GDP
growth seems to be entirely explained by the behavior of the Sub-Saharan
African series. It should also be noticed that the direction of causality (i.e., the
sign of the relevant coefficient) varies quite a bit: for some regions it is negative,
for others it is positive, depending not only on the region but also on the SPI
index we are looking at. As a result, the evidence in support of the popular
hypothesis that the level of SPI causes the rate of per capita income growth
would seem much weaker than generally believed.
As for the Granger-causality relationship flowing from economic growth
to SPI, the results presented in tables 5a and 5b show that only in the case of the
" Conte and Darrat (1988) provide an interesting discussion of long versus short term issues
within causality frameworks.
' * Adding initial income in the previous period, an interaction between lagged initial income and
lagged SPI, or both, does not qualitatively change our results.
125
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
severe SPI index is there evidence of causality. The sign of the coefficient for all
LDCs in our sample suggests that a reduction in the rate of economic growth
Granger-causes a higher level of severe SPI. Yet, and once again, when we break
down this result by region, this relationship obtains only for the Sub-Saharan
Africa sample. With the moderate SPI index (Table 5b), we find no evidence
whatsoever of causality flowing from GDP growth to SPI.lJ
In sum, the causality relationship between GDP growth and SPI seems to
be that an increase in the lower-bound SPI does indeed seem to Granger-cause a
reduction in the rate of economic growth. On the other hand, economic growth
seems to Granger-cause our upper-bound SPI, again with no evidence of reverse
causality. When we conduct these tests separately by region, what emerges is the
strong suspicion that the Sub-Saharan Africa sample is the sole force driving
these results. These rather mixed results lead us to probe further, in particular
into examining the causality patterns between SPI and investment, and SPI and
human capital.
In tables 6, 7a and 7b, we turn to the relationship between SPI and the
accumulation of physical capital. From the results in Table 6, it can be seen that
when assessing whether a high level of SPI (moderate or severe) Granger-causes
investment, the answer seems to be a resounding 'no'. The frail exception is for
" In order to leave just time-series variation to be explained, we included country dummies in
these specifications. This addition does not affect the causalitv results presented here.
126
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the Sub-Saharan Africa sub-sample, w here the coefficient on lagged moderate
SPI is statistically significant, but only at the 107c level.
The results given in Tables 7a and 7b paint a much more incisive picture
on whether investment Granger-cause m oderate or severe SPI. In both of these
cases, the answer is a resounding 'yes'. For all the LDCs in our sample,
contractions in the accumulation of physical capital seem to Granger-cause both
an increase in moderate and in severe SPI. Conspicuous regional differences
again seem to drive this result, the results for Sub-Saharan Africa again standing
out. In this case, there is no suggestion of reverse causality, and the results are
robust to the use of different SPI indexes. The rate of investm ent seems to
Granger-cause SPI, irrespective of it being measured by the severe or moderate
type. Once again, however the sole driving force behind these results seems to be
the Sub-Saharan Africa sample.
As noted before, the SPI literature has not been sufficiently attentive to the
possibility of a human capital-SPI causal relationship. Yet, especially in the case
of the severe SPI, it would be natural to expect that an increase in the number of
political assassinations, higher frequency of coups, or a civil war, could accelerate
the depreciation of the stock of hum an capital. Therefore, we judge that this
would be a good opportunity to investigate this relationship.
Table 8 presents the results evaluating whether the level of m oderate and
severe SPI would Granger-cause the level of human capital (m easured by the
127
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
average number of prim ary school years on population older than 25 years, from
Barro and Lee, 1993). We find there is no evidence whatsoever of such a
relationship. Yet, from tables 9a and 9b, it can be seen that a number of
interesting results obtain for the inverse question. An increase in the level of
human capital is found to Granger-cause a reduction in the level of moderate
SPI, for all LDCs in our sample. The Middle-East and North Africa, closely
followed by Sub-Saharan Africa, is the major contributor to this result. Although
we can not find evidence that the level of human capital Granger-causes severe
SPI in the full sample, this is a result that obtains for our sample of Sub-Sahara
African countries.
In summary, these human capital-SPI results somewhere between the
investment-SPI and the growth-SPI findings discussed above. With the
investment set, it shares the important absence of indications of reverse causality:
SPI seems far from Granger-causing the level of hum an capital, but evidence
suggesting that the level of human capital Granger-causes SPI is relatively
abundant. With the growth-SPI set, it shares the uncomfortable differences
between the two indexes as well as the conspicuous regional differences: for the
pooled sample (all LDCs) human capital seems to Granger-cause moderate SPI,
but not severe SPI. Within the moderate SPI context (Table 9a), the Middle-East
and North Africa sample dominates. Within the severe SPI context, that
determining role is played by the Sub-Saharan Africa sample.
128
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3. The Relationship between SPI and Institutional Development
The objective of this section is to evaluate whether or not the previous
results are robust to modifications in some key estimation features. As discussed
in Section 1, the two critical issues in the Granger framework are the nature of
the time lags (their length and frequency) and the content of the information set.
In what follows, we focus our attention on the latter.
Nevertheless, some tests on the frequency of the time lags were
performed. The results are not affected by the inclusion of two lags of the
"causing variable x" (instead of the one lag results presented throughout this
chapter), nor by the exclusion of one lag of the "caused variable y". We kept the
length of the lag fixed at five years throughout.1 4
The issue regarding the content of the information set refers to the
existence of omitted variables, that affect both growth and SPI, having serious
consequences for the validity of our Granger causality tests. One particularly
promising candidate for such third variable is institutional development. One
reason for it is the relationship between SPI and institutional developm ent itself.
For those unfamiliar with these two literatures, it has to come as a surprise that
4 As noted before, this is because in this chapter the interest is on the relation between SPI and
long-term growth.
129
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
they have very much evolved separately.1 3 Here we evaluate a possible (and
perhaps the simpler) connection: that high levels of SPI are associated with low
levels of institutional development. More specifically, our hypothesis is that, in a
given country, the level of SPI is contemporaneously negatively correlated with
the level of institutional development.
We measure institutional development by an index of "legislative
effectiveness" (from Banks, 1984). The attractive features of this variable are (1)
that it is available for a large num ber of LDCs, and (2) that conceptually it would
seem to be an aspect of institutional development which is especially closely
related to SPI.1 0 The unattractive features are that the data are available only until
1984 and it is a categorical variable that assumes one of four values, from zero to
3.‘ We tried to minimized these drawbacks by using it lagged one period and by
averaging it out.
Table 10 presents some evidence on our hypothesis that high levels of SPI
are associated with low levels of institutional development. All correlation
” However, some first and original insights to understand the relationship between SPI and
institutional development are found in Dhonte and Kapur (1997.) We do not feel the need to
devote as much attention to the much more investigated relationship between institutions and
economic growth.
: o For example, the quality of the bureaucracy is another aspect of institutional development, but
its links to SPI are not as direct or clear.
' "Legislative effectiveness" (LEGEF) is coded zero if no legislature exists. It is coded one for a
ineffective legislature, whether legislative activity is of a "rubber stamp" character,
implementation of legislation is faulty or if it is completely subordinate to the executive. LEGEF is
coded '2' for a "partially effective legislature": if the executive's power substantially outweighs
but does not completely dominate that of the legislature. Finally, a code of '3' is reserved for an
130
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
coefficients are negative and statistically significant. Given the components of
our index of mild SPI, it is not entirely surprising that these correlations are
somewhat higher for this index than for severe SPI.
The next step was to augment the G ranger causality tests with the index of
“legislative effectiveness" (LEGEF.) The most im portant consequence of this
addition is that the resulst obtained previously tend to hold: for "all LDCs" per
capita GDP grow th Granger-causes severe SPI (Tables 12a and 12b), and
moderate SPI Granger-causes (a decline in) per capita GDP growth (Table 11a
and lib.)
It is also notew orthy that for the complete sample of LDCs, LEGEF is
statistically significant only in two occasions. First, it is significant and positive
when the question is w hether severe SPI causes GDP growth (Table lib). Second,
it is significant and negative when the question is whether GDP growth causes
mild SPKTable 12a.)
Once again, we investigate whether or not the results hold when we break
down the sample by geographical region. Regarding the Sub-Saharan Africa
sample, notice two im portant results: one is that this is the only region where
lower-bound SPI seems to Granger cause real GDP growth, even after controlling
for institutional quality (Table 11a.) The other result is regarding the upper-
bound SPI index: LEGEF is significant and has the expected positive sign when
"effective legislature", distinguished by significant governmental autonomy, including its ability
to override executive vetoes of legislation.
131
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the question is whether real GDP growth Granger causes SPI (Table 12a.) In other
words, in Africa it is only after we control for institutional development that
GDP growth seems to Granger cause lower-bound SPI (yet note that growth
Granger causes severe SPI irrespective of LEGEF.)
The fact that there are m any more significant results for Sub-Sahara Africa
than for the other regions, suggests once again that the Sub-Saharan Africa
sample is the driving force behind the Granger-causality patterns. The very few
results for other regions are the following. For the Asia sample, we obtain that it
is only after accounting for institutional quality that growth seems to Granger
cause upper-bound SPI. Notice also that it is only for the Asian countries that
LEGEF is statistically significant and has the expected direction of effect on
LBSPI. One last statistically significant result is that for the countries of the
Middle East and North Africa it is only after we control for institutional
development that a higher level of upper-bound SPI ceases to Granger cause a
higher rate of real GDOP grow th (Table lib).
In sum, we believe that our two main conclusions are not altered by this
robustness exercise. After changing the number of lags and adding an index of
institutional development, the Olsonian effect still dominates: causality seems to
flow from growth to severe SPI, and not the other way around. Further, the
suspicion that the Sub-Saharan Africa sample is driving the results seems to
leave this section rather strengthened.
132
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4 Conclusions and suggestions for future research
The objective of this chapter was to revisit the literature on socio-political
instability and long-term economic growth. Because of our "boundaries
approach" to measuring SPI, on 5 years intervals, we were able to provide a
somewhat explicit depiction of the causality pattern between SPI and growth.
The main conclusions are the following. First, there seems to be at best
only weak and partial support for the two-way causality assumed in much of the
literature. In particular, moderate but not severe SPI seems to Granger-cause a
reduction in per capita GDP growth, and economic growth seems to Granger-
cause a reduction in severe but not moderate SPI. Also, we find that both
investment and hum an capital Granger-cause SPI but not the other way around.
Because of the difficulties in constructing a lower-bound SPI index, and
because the upper-bound SPI index is very similar to other existing treatments of
SPI in the literature, we emphasize the latter results. Severe SPI does not cause
economic growth, and causality flows in the precise opposite direction: low
growth increases the level of SPI. On this basis, we claim that the Olsonian effect
dominates: rapid economic growth involves substantial structural change, which
may undo political coalitions and readjust the relative power of different interest
groups (Olson, 1963.)
133
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The second conclusion is the strong suspicion that the Sub-Saharan Africa
sample is the driving force behind the Granger-causality patterns obtained.
There would seem to be two reasons for this: (1) SPI problems are more frequent
and more severe in Africa, and (2) the Sub-Saharan African sample is larger than
for those of other regions (indeed, comprising approximately 40 percent of our
sample.)
One im portant policy implication is that these results suggest that there is
less reason to believe that social and political instability, by itself, constitutes
such a severe barrier to long-term economic growth that political stability should
be achieved at any cost.
Another interesting implication is that there seem to be some quite
different regional patterns of the relationship between SPI and economic growth.
For example, the African pattern is perhaps one where problems seem to be
more of an institutional, long-term, and structural nature, than just simply
sporadic political instability. The results for the Middle East and North Africa
suggest that a pattern can be found in the interplay between SPI and the
"instability of the external environment." For example, the two SPI measures do
not account at all for the Israeli-Arab conflict. The Asia pattern is marked by
severe SPI causing a decline in the GDP growth rate, and institutional
development having a significant and positive effect when explaining upper-
bound SPI. And finally, the Latin American seems to be a pattern where SPI,
134
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
however measured, has no long-term effect whatsoever on the growth rate (and
vice-versa). At this point, we conjecture that the constant and dramatic changes
in policy ("policy instability"1 8 ) should be brought into the analysis when
focusing on this region.
There are also some suggestions for future research. First, SPI has been a
staple of political science for at least 30 years (Huntington, 1968; Hibbs, 1973; and
Sanders, 1983). Future efforts should be more attentive to this history.
Second, given the substantial difference in results between the two
indexes of SPI used in this study, exploratory research of this sort with other SPI
measures should be encouraged (in particular, with different lower-bound SPI
measures.)
Third, another direction for future research is to experiment with different
causality frameworks, which can accommodate different lag structures as well as
richer information sets (Hsiao, 1979).
And fourth, the regional patterns suggested above are simple conjectures.
We believe that further research that brings into the analysis institutional
development (Africa), policy instability (Latin America), and the instability of the
external environm ent (Middle East) would be able to generate some important
lessons. Because of enorm ous inter-country differences in Asia (the sample
1 9 See Aizenman and Marion (1993.)
135
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
includes Pakistan and Singapore, China and Laos), in this case research should
start at the individual country level.
136
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 1.
Sample of 98 Developing Countries
Asia (14) Latin America
(20)
Middle-East and
N. Africa (16)
Sub-Saharan Africa (38)
Bangladesh Argentina Algeria Angola
China Bolivia Bahrain Burundi
Indonesia Brazil Cyprus Benin
India Chile Egypt Burkina Faso
Korea, South Colombia Iran Botswana
Laos Costa Rica Iraq Cameroon
Malaysia Dominican Rep. Jordan Central Africa Rep.
Myanmar Ecuador Kuwait Chad
Pakistan El Salvador Morocco Congo
Philippines Guatemala Oman Ethiopia
Singapore Honduras Saudi Arabia Gabon
Sri Lanka Jamaica Syria Ghana
Taiwan Mexico Tunisia Guinea
Thailand Nicaragua Turkey Guinea Bissau
Panama United Arab Em. Ivory Coast
Peru
Paraguay
T rinidad-T obago
Uruguay
Venezuela
Yemen Kenya
Liberia
Lesotho
Madagascar
Malawi
Mali
Mauritania
Mauritius
Mozambique
Nigeria
Rwanda
Sudan
Senegal
Sierra Leone
Somalia
South Africa
Togo
Tanzania
Uganda
Zaire
Zambia
Zimbabwe
137
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 2.
Correlation coefficients between
our severe and moderate SPI indexes, 1960-1995
REGION CORRELATION
COEFFICIENT
SIGNIFICANCE
LEVEL
NUMBER OF
OBSERVATIONS
Asia 0.284 .0001 62
Latin America 0.535 .0001 104
Middle East and
North Africa
-0.027 .8169 73
Sub-Saharan
Africa
0.319 .0001 165
All LDCs 0.319 .0001 404
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3.
Socio-political Instability and Economic Growth
(98 LDCs, 1960-1995)
DEPENDENT
VARIABLE
INTERCEPT GDP,., GDP,, LBSPI, LBSPI,, Adj. R-
GDPt 1.9188 **
(2.942)
0.232 * *
(4.990)
0.006307
(0.139)
-0.212923
(-1.789)
0.0579
GDP. 2.4892 **
(3.515)
0.2363 * *
(5.050)
0.005965
(0.131)
0.046616
(0.260)
-0.367 * *
(-2.065)
0.0675
INTERCEPT GDP,, GDP,, UBSPI, UBSPI,, Adj. R2
GDPt 1.2247 **
(4.470)
0.1849 * *
(3.206)
-0.029420
(-0.554)
-0.349981
(-1.862)
0.0351
GDP, 0.9531 **
(3.054)
0.1979 * *
(3.102)
-0.023681
(-0.403)
-0.525 **
(-2.281)
0.323667
(1.488)
0.0431
DEPENDENT
VARIABLE
INTERCEPT LBSPI,, LBSPI,, GDP, GDP„ Adj. R2
LBSPI, 1.4427 **
(8.171)
0.6988 * *
17.198)
0.017109
0.429
-0.005570
(-0.502)
0.5141
LBSPI, 1.4781 **
(8.198)
0.684 * *
(15.682)
0.026971
(0.626)
0.000531
(0.043)
-0.013036
(-1.160)
0.5276
INTERCEPT UBSPI,, UBSPI,, GDP, CDP„ Adj. R2
UBSPI, 0.047144
(0.718)
0.3735 * *
(7.335)
-0.0277 *
(-1.991)
0.1474
UBSPI, 0.118005
(1.621)
0.3709 * *
(7.343)
-0.0266
(-1.859)
-0.0344 *
(-2.166)
0.1659
Notes: t-values are in parenthesis. LBSPI is lower-bound SPI. UBSPI is upper-
bound SPI, GDP is the OLS per capita GDP Growth Rate.
* Statistically significant at the 5 percent level.
** Statistically significant at the 1 percent level.
139
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 4.
Does SPI Granger-causes G DP Growth?
INTERCEPT GDP„ GDP,, LBSPI,, Adj. R2
All LDCs 0.884160 * *
(4.089)
0.195091 * *
(4.358)
0.034480
(0.765)
0.0357
2.558609 * *
(3.905)
0.235449 * *
(5.049)
0.005134
(0.113)
-0.3327**
(-2.762)
0.0695
Asia 3.167412 * *
(5.160)
0.299528 *
(2.509)
-0.073772
(-0.632)
0.0556
2.580185
(1.949)
0.294964 *
(2.460)
-0.065929
(-0.561)
0.131835
(0.540)
0.0471
Latin America 0.232461
(0.549)
0.136115
(1.453)
0.229310 *
(2.055)
0.0354
0.076320
(0.063)
0.138919
(1.441)
0.235975 *
(2.026)
0.025003
(0.119)
0.0264
Middle East &
North Africa
2.210495 * *
(3.012)
0.106884
(0.920)
-0.189765 *
(-2.074)
0.0428
4.147839
(1.930)
0.097620
(0.770)
-0.187396 *
(-2.024)
-0.38354
(-0.929)
0.0385
Sub-Saharan
Africa
0.219911
(0.756)
0.107009
(1.648)
0.077911
(1.101)
0.0105
3.153955 * *
(3.556)
0.137547 *
(2.059)
-0.012702
(-0.182)
-0.6164**
(-3.711)
0.0929
INTERCEPT GDP,, GDP,, UBSPI,, Adj. R:
All LDCs 0.613272 *
(2.493)
0.248576 * *
(4.953)
0.026668
(0.533)
0.208790
(1.199)
0.0581
Asia 3.173282 **
(4.774)
0.290015 *
(2.366)
-0.063754
(-0.503)
-0.4607
(-1.593)
0.1069
Latin America -0.137776
(-0.280)
0.157782
(1.542)
0.226885
(1.747)
0.260559
(0.875)
0.0266
Middle East &
North Africa
2.224822 * *
(2.706)
0.082408
(0.631)
-0.218926 *
(-2.176)
1.19942 *
(2.358)
0.0931
Sub-Saharan
Africa
-0.093419 (-
0.287)
0.189193 *
(2.584)
0.034581
(0.433)
0.040219
(0.147)
0.0270
1 Notes: Dependent variable is OLS per capita GDP Growth Ratp. t-values are in
parenthesis. LBSPI is lower-bound SPI, UBSPI is upper-bound SPI.
* Statistically significant at the 5 percent level
| ** Statistically significant at the 1 percent level.
140
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 5a.
Does GDP Growth Granger-causes SPI?
(Dependent variable is lower-bound SPI)
INTERCEPT LBSPI,., LBSPI,, GDP,, Adj. R:
All LDCs 1.58357**
(9.198)
0.64321 **
(16.273)
0.04001
(1.018)
0.4303
1.3827 * *
(8.145)
0.70016 **
(16.698)
0.01669
(0.406)
-0.00836
(-0.788)
0.5095
Asia 1.7372 * *
(4.476)
0.5763 **
(6.733)
0.07443
(0.877)
0.3927
1.84659 **
(4.453)
0.75044 **
(7.302)
-0.14362
(-1.407)
-0.00872
(-0.272)
0.4389
Latin America 2.43436 * *
(6.73)
0.53534 **
(7.392)
-0.02712
(-0.378)
0.268
1.94099 * *
(5.192)
0.58011 * *
(7.951)
-0.00561
(-0.077)
0.05211
(1.828)
0.3583
Middle East &
North Africa
0.81684 *
(2.099)
0.68933 **
(8.075)
0.12635
(1.42)
0.5367
0.83341 *
(2.528)
0.70157 **
(8.393)
0.09693
(1.20)
0.00366
(0.265)
0.6419
Sub-Saharan
Africa
1.2051 * *
(4.484)
0.77186 **
(10.105)
0.00026
(0.003)
0.5404
1.15902 * *
(4.517)
0.79971 **
(9.668)
-0.00686
(-0.083)
-0.03895
(-1.891)
0.6072
Notes: Periods over which the dependent variable is measured are five-vear
averages, 1960-1995. t-values are in parenthesis. LBSPI is lower-bound SPI,
UBSPI is upper-bound SPI.
* Statistically significant at the 5 percent level.
** Statistically significant at the 1 percent level.
141
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Does GDP
(Dependen
Table 5b.
Growth Granger-causes SPI?
variable is upper-bound SPI)
INTERCEPT UBSPI,., GDP,., Adj. R:
All LDCs 0.02215
(0.345)
0.35717 * *
(6.833)
0.1246
0.09786
(1.354)
0.36623 * *
(7.23)
-0.04094 **
(-2.634)
0.1593
Asia 0.16265
(0.984)
0.46766 * *
(3.137)
0.1555
0.58973
(2.08)
0.4465 * *
(3.06)
-0.10464
(-1.832)
0.1958
Latin America -0.10729
(-0.837)
0.44548 * *
(4.408)
0.1835
-0.14399
(-0.889)
0.4525 * *
(4.379)
0.01551
(0.374)
0.1747
Middle East &
North Africa
-0.23781
(-2.340)
0.27791 * *
(3.772)
0.1883
-0.21544
(-1.722)
0.27217 * *
(3.832)
-0.01066
(-0.495)
0.1968
Sub-Saharan
Africa
0.16678
(1.518)
0.32497 * *
(3.47)
0.0777
0.18918
(1.768)
0.35183 * *
(4.03)
-0.05952 *
(-2.448)
0.1435
Notes: Periods over which the dependent variable is m easured are five-vear
averages, 1960-1995. t-values are in parenthesis. LBSPI is lower-bound SPI,
UBSPI is upper-bound SPI.
* Statistically significant at the 5 percent level.
** Statistically significant at the 1 percent level.
142
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 6.
Does SPI Granger-causes Investment?
INTERCEPT IN V „ INV,., LBSPI,, Adj. R:
All LDCs 2.66286 * *
(7.502)
0.95794 **
(21.724)
-0.15494 **
(-3.604)
0.7348
2.82157 * *
(3.683)
0.95615 **
(20.708)
-0.13461 **
(-2.935)
-0.07203
(-0.620)
0.7496
Asia 1.91661 *
(2.169)
1.06238 **
(8.771)
-0.09676
(-0.721)
0.8403
1.93491
(1.056)
1.05014 **
(8.626)
-0.09904
(-0.735)
0.06303
(0.204)
0.8344
Latin America 4.25176 **
(4.503)
0.86292 * *
(9.630)
-0.13706
(-1.624)
0.6159
4.33989 * *
(2.880)
0.84232
(9.064)
-0.14182
(-1.637)
0.05575
(0.301)
0.5853
Middle East &
North Africa
4.15469 * *
(4.203)
1.01984 * *
(9.326)
-0.31957 * *
(-2.853)
0.7314
6.78391 * *
(3.085)
1.03131 **
(9.578)
-0.35006 **
(-3.085)
-0.46458
(-1.360)
0.7440
Sub-Saharan
Africa
2.40571 * *
(5.143)
0.84084 * *
(12.672)
-0.08111
(-1.329)
0.6911
3.76778 * *
(3.210)
0.75302 * *
(10.179)
-0.00564
(-0.084)
-0.29589
(-1.657)
0.6939
INTERCEPT
INV,,
INV,., UBSPI,, Adj. R'
All LDCs 2.34316 * *
(5.608)
0.95037 **
(19.140)
-0.11782 *
(-2.396)
0.07989
(0.491)
0.7382
Asia 2.10651
(2.053)
0.98298 **
(7.714)
-0.02467
(-0.176)
-0.16239
(-0.423)
0.8410
Latin America 5.13825 * *
(4.486)
0.81549
(7.974)
-0.14007
(-1.498)
-0.10976
(-0.421)
0.5362
Middle East &
North Africa
4.54168 * *
(4.212)
1.03034 * *
(9.028)
-0.34551 **
(-2.987)
0.62161
(1.565)
0.7198
Sub-Saharan
Africa
1.82034 * *
(3.229)
0.78451 **
(9.769)
0.01344
(0.183)
-0.07063
(-0.263)
0.6767
Notes: Dependent variable is Average Investment Rate, t-values are in
parenthesis. LBSPI is lower-bound SPI, UBSPI is upper-bound SPI.
* Statistically significant at the 5 percent level.
** Statistically significant at the 1 percent level.
143
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 7a.
Does investment Granger-cause SPI?
(Dependent variable is lower-bound SPI)
INTERCEPT LBSPI,., LBSPI,., INV,, Adj. R:
All LDCs 1.58357 **
(9.198)
0.64321 **
(16.273)
0.04001
(1.018)
0.4303
1.86384 * *
(9.163)
0.67156 **
(16.042)
0.01588
(0.394)
-0.02520 * *
(-4.213)
0.5243
Asia 1.73720 * *
(4.476)
0.57630 **
(6.733)
0.07443
(0.877)
0.3927
1.97120 * *
(4.576)
0.74030 **
(7.201)
-0.13656
(-1.345)
-0.00869
(-0.949)
0.4438
Latin America 2.43436 * *
(6.730)
0.53534 * *
(7.392)
-0.02712
(-0.378)
0.2680
2.25192 * *
(4.338)
0.57146 * *
(7.754)
-0.00911
(-0.124)
-0.01232
(-0.617)
0.3461
Middle East &
North Africa
0.81684 *
(2.099)
0.68933 * *
(8.075)
0.12635
(1.420)
0.5367
1.45039 * *
(3.616)
0.67689 * *
(8.295)
0.08192
(1.047)
-0.02712 *
(-2.462)
0.6637
Sub-Saharan
Africa
1.20510 * *
(4.484)
0.77186 * *
(10.105)
0.00026
(0.003)
0.5404
1.57146 * *
(4.546)
0.74600 * *
(8.920)
0.01701
(0.213)
-0.02572 *
(-2.172)
0.6094
Notes: Periods over which the dependent varia
averages, 1960-1995. t-values are in parenthesis
UBSPI is upper-bound SPI.
* Statistically significant at the 5 percent level.
* * Statistically significant at the 1 percent level.
?le is measured are five-year
LBSPI is lower-bound SPI,
144
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 7b.
Does investment Granger-cause SPI?
(Dependent variable is upper-bound SPI)
INTERCEPT UBSPI,, INV_, Adj. R2
All LDCs 0.02215
(0.345)
0.35717 * *
(6.833)
0.1246
0.31240 *
(2.399)
0.35245 * *
(6.9)
-0.02242 * *
(-2.711)
0.1604
Asia 0.16265
(0.984)
0.46766 * *
(3.137)
0.1555
0.65195
(1.734)
0.44078 * *
(2.967)
-0.03108
(-1.445)
0.1747
Latin America -0.10729
(-0.837)
0.44548 * *
(4.408)
0.1835
0.00242
(0.006)
0.44537 * *
(4.381)
-0.00650
(-0.283)
0.1741
Middle East &
North Africa
-0.23781
(-2.340)
0.27791 * *
(3.772)
0.1883
-0.30219
(-1.473)
0.27332 * *
(3.79)
0.00328
(0.275)
0.1941
Sub-Saharan
Africa
0.16678
(1.518)
0.32497 * *
(3.47)
0.0777
0.53254 * *
(2.853)
0.29906 * *
(3.322)
-0.03890 * *
(-2.633)
0.1496
Notes: Periods over which the dependent variable is m easured are five-vear
averages, 1960-1995. t-values are in parenthesis. LBSPI is lower-bound SPI,
UBSPI is upper-bound SPI.
* Statistically significant at the 5 percent level.
* * Statistically significant at the 1 percent level.
145
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 8.
Does SPI Granger-causes Human Capital?
INTERCEPT
PYR,,
PYR,., LBSPI,, Adj. R:
All LDCs 0.26848 * *
(7.579)
1.00222 **
(14.485)
-0.01359
(-0.191)
0.9512
0.25672 **
(3.765)
1.00287 **
(14.394)
-0.01575
(-0.220)
0.00344
(0.337)
0.9501
Asia 0.16267
(1.944)
1.01656 **
(6.594)
0.03062
(0.187)
0.961
0.35176 *
(2.028)
0.98666 **
(6.367)
0.04826
(0.294)
-0.02961
(-1.070)
0.960
Latin America 0.35936 **
(3.925)
0.88167 **
(7.397)
0.07099
(0.582)
0.9264
0.23511
(1.714)
0.89777 **
(7.507)
0.06100
(0.500)
0.01963
(1.213)
0.9268
Middle East &
North Africa
0.45468 * *
(4.351)
0.76027 * *
(4.722)
0.19492
(1.163)
0.8866
-0.10391
(-0.297)
0.72089 * *
(4.498)
0.33244
(1.819)
0.09114
(1.702)
0.8914
Sub-Saharan
Africa
0.12065 * *
(2.571)
1.44300 * *
(10.805)
-0.41690 **
(-2.991)
0.9495
0.14566
(1.310)
1.45708 **
(10.731)
-0.43748 **
(-3.052)
-0.00399
(-0.252)
0.9484
INTERCEPT
PYR,.,
PYR,, UBSPI,, Adj. R:
All LDCs 0.27561 * *
(7.539)
1.00278 **
(14.370)
-0.01640 (-
0.229)
0.00540
(0.371)
0.9499
Asia 0.18955 *
(2.215)
1.00647 **
(6.589)
0.03539
(0.218)
-0.03245
(-1.126)
0.9601
Latin America 0.35764 **
(3.873)
0.88245 * *
(7.358)
0.07096
(0.578)
0.00528
(0.252)
0.9255
Middle East &
North Africa
0.47149 * *
(4.430)
0.75866 * *
(4.685)
0.19765
(1.176)
0.08465
(1.259)
0.8876
Sub-Saharan
Africa
0.10542 *
(2.058)
1.46642 **
(10.758)
-0.43040 **
(-3.031)
0.022597
(0.963)
0.9487
Notes: Dependent variable is average number of vears of schooling (primary
education), t-values are in parenthesis. LBSPI is lower-bound SPI, UBSPI is
upper-bound SPI.
* Statistically significant at the 5 percent level.
* * Statistically significant at the 1 percent level.
146
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 9a.
Does hum an capital Granger-cause SPI?
(Dependent variable is lower-bound SPI)
INTERCEPT LBSPI,, LBSPI,, PYR,,
Adj. R;
All LDCs 1.58357 **
(9.198)
0.64321 * *
(16.273)
0.04001
(1.018)
0.4303
1.69112 **
(6.968)
0.70278 * *
(13.983)
0.00530
(0.109)
-0.17595 * *
(-3.657)
0.5758
Asia 1.73720 **
(4.476)
0.57630 * *
(6.733)
0.07443
(0.877)
0.3927
1.95866 **
(3.650)
0.73240 * *
(6.616)
-0.12607
(-1.187)
-0.08573
(-0.913)
0.4813
Latin America 2.43436 **
(6.730)
0.53534 * *
(7.392)
-0.02712
(-0.378)
0.268
2.52813 **
(4.185)
0.64599 * *
(7.672)
-0.08807
(-1.045)
-0.16449
(-1.341)
0.4152
Middle East &
North Africa
0.81684 *
(2.099)
0.68933 * *
(8.075)
0.12635
(1.420)
0.5367
1.82432 **
(3.890)
0.70514 * *
(8.357)
0.02895
(0.338)
-0.42178 * *
(-4.331)
0.8157
Sub-Saharan
Africa
1.20510 **
(4.484)
0.77186 * *
(10.105)
0.00026
(0.003)
0.5404
1.38183 **
(2.869)
0.65615 * *
(5.711)
0.12363
(1.129)
-0.20455
(-1.573)
0.6456
Notes: LBSPI is lower-bound SP
parenthesis.
* Statistically significant at the 5
** Statistically significant at the 1
, UBSPI is upper-bound SPI, t-values are in
percent level,
percent level.
147
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 9b.
Does human capital Granger-cause SPI?
(Dependent variable is lower-bound SPI)
INTERCEPT UBSPI,, PYR„ Adj. R:
All LDCs 0.02215
(0.345)
0.35717 * *
(6.833)
0.1246
0.16894
(1.221)
0.37589 * *
(6.521)
-0.08558
(-1.420)
0.1506
Asia 0.16265
(0.984)
0.46766 * *
(3.137)
0.1555
0.10537
(0.253)
0.43485 * *
(2.731)
0.04818
(0.291)
0.1153
Latin America -0.10729
(-0.837)
0.44548 * *
(4.408)
0.1835
0.16995
(0.398)
0.44244 * *
(4.359)
-0.09984
(-0.681)
0.178
Middle East &
North Africa
-0.23781 *
(-2.340)
0.27791 * *
(3.772)
0.1883
-0.17285
(-0.825)
0.24838 * *
(2.875)
-0.01387
(-0.115)
0.1385
Sub-Saharan
Africa
0.16678
(1.518)
0.32497 * *
(3.470)
0.0777
0.56101 *
(2.624)
0.29985 * *
(2.376)
-0.38567 **
(-2.654)
0.1774
Notes: LBSPI is lower-bound SP
parenthesis.
* Statistically significant at the 5
* * Statistically significant at the 1
1 , UBSPI is upper-bound SPI, t-values are in
percent level,
percent level.
148
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 10.
Contemporaneous Correlations between
Institutional Development and Socio-Political Instability
LBSPI UBSPI
All LDCs -0.60560 -0.38594
(0.0001) (0.0001)
Asia -0.50336 -0.48379
(0.0001) (0.0001)
Latin America -0.73452 -0.43844
(0.0001) (0.0001)
Middle East & -0.58557 -0.29549
North Africa (0.0001) (0.0334)
Sub-Saharan -0.75311 -0.39960
Africa (0.0001) (0.0001)
Notes: LBSPI is lower-bound SPL UBSPI is upper-bound SPI, Institutional
Development proxied by legislative effectiveness (LEGEF), p-values in
parenthesis.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 11a.
Controlling for institutions,
does LBSPI Granger-causes per capita GDP Growth?
INTERCEPT GDP,. LBSPI,. LEGEF,, Adj. R:
All LDCs 2.481889 **
(4.187)
0.182177 * *
(4.604)
-0.303266 **
(-2.722)
0.0538
2.817248 **
(2.748)
0.165540 * *
(4.066)
-0.342443 *
(-2.345)
-0.039341
(-0.147)
0.0500
Asia 2.497377 *
(2.065)
0.223921 *
(2.118)
0.117060
(0.496)
0.0337
1.486695
(0.639)
0.202945
(1.831)
0.355446
(1.024)
0.026766
(0.054)
0.0340
Latin America 0.863536
(0.849)
0.131721
(1.480)
-0.030617
(-0.168)
0.0021
0.539312
(0.247)
0.137169
(1.498)
0.005935
(0.022)
0.079827
(0.158)
-0.0050
Middle East &
North Africa
3.603405
(1.849)
0.057922
(0.673)
-0.344913
(-0.932)
-0.0086
5.441265
(1.811)
-0.054478
(-0.593)
-0.440192
(-0.946)
-0.661735
(-0.757)
-0.0271
Sub-Saharan
Africa
2.823999 **
(3.406)
0.124626
(0.0503)
-0.572735 * *
(-3.596)
0.0851
5.710956 **
(3.410)
0.126943 *
(2.019)
-0.932196 * *
(-3.899)
-0.908113
(-1.745)
0.1150
Notes: Dependent variable is OLS per capita GDP Growth Rate (five-vear
averages, 1960-1995), t-values are in parenthesis. LBSPI is lower-bound SPI,
UBSPI is upper-bound SPI, LEGEF is legislative effectiveness.
* Statistically significant at the 5 percent level.
* * Statistically significant at the 1 percent level.
150
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table lib .
Controlling for institutions,
does UBSPI Granger-causes per capita GDP Growth?
INTERCEPT g d p m UBSPI,., LEGEF,., Adj. R ~
All LDCs 0.616018 **
(2.726)
0.269736 * *
(5.529)
0.231037
(1.338)
0.0695
-0.006516
(-0.016)
0.267881 * *
(5.198)
0.304534
(1.616)
0.536911 *
(2.008)
0.0716
Asia 2.998808 **
(5.321)
0.273311 *
(2.331)
-0.442822
(-1.552)
0.1182
4.384518 **
(4.068)
0.214899
(1.731)
-0.731774 *
(-2.168)
-0.61494
(-1.271)
0.1423
Latin America 0.341794
(0.829)
0.157888
(1.528)
0.218299
(0.728)
0.0068
0.048886
(0.055)
0.173103
(1.620)
0.256589
(0.762)
0.177319
(0.375)
-0.0006
Middle East &
North Africa
1.066666
(1.462)
0.136677
(1.066)
1.236916 *
(2.434)
0.0795
1.682986
(1.293)
-0.018281
(-0.107)
1.034443
(1.803)
0.217131
(0.245)
0.0079
Sub-Saharan
Africa
-0.044823
(-0.142)
0.203524 * *
(2.837)
-0.015498
(-0.058)
0.0363
-0.980318
(-1.699)
0.214076 * *
(2.986)
0.206556
(0.716)
0.891145 *
(1.992)
0.0626
Notes: Dependent variable is OLS per capita GDP G row th Rate (five-vear
averages, 1960-1995), t-values are in parenthesis. LBSPI is lower-bound SPI,
UBSPI is upper-bound SPI, LEGEF is legislative effectiveness.
* Statistically significant at the 5 percent level.
** Statistically significant at the 1 percent level.
151
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 12a.
Controlling for institutions,
does per capita GDP Growth Granger-causes SPI?
(Dependent variable is lower-bound SPI)
INTERCEPT LBSPI,., GDP,., LEGEF,., Adj.
R1
AH LDCs 1.573620 **
(10.279)
0.688096 **
(23.747)
-0.012300
(-1.195)
0.4896
2.337032 * *
(8.193)
0.607719 **
(14.957)
-0.021361
(-1.886)
-0.218112**
(-2.922)
0.5190
Asia 1.884949 * *
(4.618)
0.630364 **
(7.969)
-0.036358
(-1.076)
0.3865
2.692156 * *
(3.980)
0.579503 **
(5.744)
-0.050101
(-1.555)
-0.307926 *
(-2.121)
0.5408
Latin America 1.893056 * *
(5.866)
0.582704 **
(9.729)
0.050945
(1.809)
0.3688
3.011157 * *
(3.714)
0.461281 **
(4.547)
0.045374
(1.336)
-0.227342
(-1.214)
0.3233
Middle East &
North Africa
1.396820 **
(3.986)
0.702495 **
(10.512)
-0.003570
(-0.230)
0.5207
1.559593 *
(2.250)
0.679274 **
(6.328)
-0.005741
(-0.271)
-0.084662
(-0.419)
0.4959
Sub-Saharan
Africa
1.363772 **
(6.124)
0.763120 **
(18.044)
-0.030856
(-1.738)
0.5969
1.994697 * *
(4.645)
0.697865 **
(11.385)
-0.040253 *
(-2.497)
-0.203915
(-1.528)
0.6790
Notes: Dependent variable is five-vear averages. 1960-1995. t-values are in
parenthesis. LBSPI is lower-bound SPI, UBSPI is upper-bound SPI, LEGEF is
legislative effectiveness. GDP is per capita GDP Growth.
* Statistically significant at the 5 percent level.
* * Statistically significant at the 1 percent level.
152
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 12b.
Controlling for institutions,
does per capita GDP Growth Granger-causef
(Dependent variable is upper-bound SP
> SPI?
)
INTERCEPT UBSPI,, GDP., LEGEF,, Adj. R'
All LDCs 0.097856
(1.354)
0.366232 **
(7.230)
-0.040939 **
(-2.634)
0.1593
0.200719
(1.507)
0.332058 **
(5.660)
-0.036577 *
(-2.154)
-0.074700
(-0.914)
0.1449
Asia 0.589727 *
(2.080)
0.446500 **
(3.060)
-0.104643
(-1.832)
0.1958
1.037445
(1.850)
0.377776 *
(2.061)
-0.109638
(-1.724)
-0.249218
(-1.001)
0.2019
Latin America -0.143994
(-0.889)
0.452498 **
(4.379)
0.015514
(0.374)
0.1747
0.082156
(0.235)
0.397206 **
(3.126)
0.009341
(0.211)
-0.115378
(-0.692)
0.1593
Middle East &
North Africa
-0.215442
(-1.722)
0.272171 **
(3.832)
-0.010656
(-0.495)
0.1968
-0.089325
(-0.405)
0.166773
(1.868)
-0.023464
(-0.862)
-0.063236
(-0.439)
0.0640
Sub-Saharan
Africa
0.189177
(1.768)
0.351834 **
(4.030)
-0.059520 *
(-2.448)
0.1435
0.115384
(0.591)
0.368014 **
(3.854)
-0.053414 *
(-2.114)
0.060641
(0.403)
0.1314
Notes: Dependent variable is five-vear averages. 1960-1995. t-values are in
parenthesis. LBSPI is lower-bound SPI, UBSPI is upper-bound SPI, LEGEF is
legislative effectiveness. GDP is per capita GDP Growth.
* Statistically significant at the 5 percent level.
** Statistically significant at the 1 percent level.
153
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Bibliography
Abramovitz, Moses and Paul David, "Convergence and Deferred Catch-Up:
Productivity Leadership and the Waning of American Exceptionalism", in R.
Landau, T. Taylor and G. Wright (eds.), The Mosaic of Economic Growth, Palo
Alto, Stanford University Press, 1996.
Adelman, Irma and Cynthia Morris, Society. Politics and Economic
Development: A Quantitative Approach, Baltimore, The Johns Hopkins
University Press, 1967.
Adelman, Irma and Cynthia Morris, "Nineteenth Century Developm ent
Experience and Lessons for Today", World Development.! 7. pp. 1417-1432,1989.
Adelman, Irma and Nobuhiko Fuwa, "Income Inequality and Development: The
1970s and 1980s Compared", Economie Apliquee. Vol. XLVT, No. 1, pp. 7-29,
1994.
Aizenman, J. and N. Marion (1993). "Policy Uncertainty, Persistence, and
Growth," Review of International Economics 1,145-163.
Alesina, Alberto and Roberto Perotti, "The Political Economy of Growth: A
Critical Survey of the Recent Literature and Some New Results", Columbia
University, Economics Discussion Paper No. 626,1992.
Alesina, Alberto and Dani Rodrik, "Distribution, Political Conflict and Economic
Growth: A Simple Theory and Some Empirical Evidence", in Alex Cukierman,
Zvi Hercowitz and Leonardo Leiderman (Eds.), Political Economy. Growth, and
Business Cycles , Cambridge. The MIT Press, 1992.
Alesina, Alberto and Roberto Perotti, "Income Distribution, Political Instability,
and Investment", European Economic Review, Vol.40, No.6, pp.1203-1228, June
1996.
Alston, Lee, "Empirical Work in Institutional Economics: An Overview", in L.
Alston, T. Eggertsson and D. North (eds), Empirical Studies in Institutional
Change. Cambridge, Cambridge University Press, 1996.
154
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Arat, Zehra, "Democracy and Development: M odernization Theory Revisited",
Comparative Politics. Vol. 21, No. 1 (1988), pp. 21-36.
Armijo, Leslie, Thomas Biersteker and Abraham Lowenthal, "It's Fine in Practice,
But How Does It Work in Theory? Understanding the Relationship Between
Democratization and Market-Oriented Economic Reform", Mimeo, University of
Southern California, 1994.
Artus, Patrick, Theorie de la Croissance et des Fluctuations. Paris, Presses
Universitaires de France, 1993.
Azariadis, Costas and Allan Drazen, "Threshold Externalities in Economic
Development", Quarterly journal of Economics, Vol. 105, No. 2, pp. 501-526, May
1990.
Bahmani-Oskooee, M., H. Mohtadi and G. Shabsigh. (1991)."Exports, Growth
and Causality in LDCs: A Re-examination," Toumal of Development Economics,
36,405-415.
Banks, A., Political Handbook of the W orld. New York, McGraw-Hill, 1984.
Bardhan, Pranab, "The New Institutional Economics and Development Theory",
World Development, 17 (9): 1390-1394,1989.
Bardhan, Pranab, "The Contributions of Endogenous Growth Theory to the
Analysis of Development Problems: An Assessment", in Proceedings of the
Workshop on Endogenous Growth and Development. Siena , University of
Siena, July 1994.
Bardhan, Pranab, "The Nature of Institutional Impediments to Economic
Development", University of California at Berkeley, mimeo, January 1996.
Barro, Robert, "Economic Growth in a Cross Section of Countries", Quarterly
foumal of Economics, Vol. 106, No. 2 (1991), pp. 407-444.
Barro, Robert, "Democracy and Growth", W orking Paper No. 4909, Cambridge,
National Bureau of Economic Research, Inc., 1994.
Barro, Robert and Jong-Wha Lee, "International Comparisons of Educational
Attainment", Journal of Monetary Economics. Vol. 32, No. 3, pp. 363-394,
December 1993.
155
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Barro, Robert and Jong-Wha Lee, “Losers and Winners in Economic Growth", in
Proceedings of the W orld Bank Annual Conference on Development Economics
1993 , Washington D.C., The World Bank, 1994.
Barro, Robert and Xavier Sala-i-Martin, Economic Growth, McGraw-Hill, 1995.
Bates, Robert, "Social Dilemmas and Rational Individuals: An Assessment of the
New Institutionalism", in J. Hariss, J. Hunter and C. Lewis (eds.), The New
Institutional Economics and Third World Development, New York: Routledge,
1995.
Becker, Gary, "A Theory of Competition among Pressure Groups for Political
Influence", Quarterly loumal of Economics. Vol. 98, No. 3, pp. 371-400, August
1983.
Becker, Gary, Human Capital, Chicago: University of Chicago Press, 1993 (Third
edition).
Benhabib, Jess and Mark Spiegel, "The Role of Human Capital and Political
Instability in Economic Development", Rivista di Politica Economica, Vol. 82,
No. 11 (1992), pp. 55-94.
Benhabib, Jess and Mark Spiegel, "The Role of Human Capital in Economic
Development: Evidence from Aggregate Cross-Country Data", loumal of
Monetary Economics, Vol. 34, No.2, pp. 143-173, October 1994.
Berry, Albert, "The Income Distribution Threat in Latin America", University of
Toronto, mimeo, 1996.
Bhalla, Surjit, "Free Societies, Free Markets & Social Welfare", Mimeo, New
York, Goldman Sachs, 1994.
Bilson, John, "Civil Liberty - An Econometric Investigation", Kvklos. Vol.35,
pp.94-114,1982.
Blomstrom, M., Lipsey, R. and M. Zejan. (1996)."Is Fixed Investment the Key to
Economic Growth?", Quarterly loumal of Economics 111, 269-276.
Bollen, Kenneth, "Comparative Measurement of Political Democracy", American
Sociological Review. Vol. 45, No. 3 (1980), pp. 370-390.
156
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Bollen, Kenneth, "Liberal Democracy: Validity and Method Factors in Cross-
National Measures", American loum al of Political Science. Vol. 87, No. 4 (T993),
pp. 1207-1230.
Burkhart, Ross and Michael Lewis-Beck, "Comparative Democracy: The
Economic Development Thesis", American Political Science Review. Vol. 88, No.
4 (1994), pp. 903-910.
Boone, Peter, "Politics and the Effectiveness of Foreign Aid", NBER W orking
Paper 5308, October 1995.
Buchanan, James, The Limits of Liberty: Between Anarchy and Leviathan.
Chicago : University of Chicago Press, 1975.
Burkhart, Ross and Michael Lewis-Beck, "Comparative Democracy: The
Economic Development Thesis", American Political Science Review. Vol. 88, No.
4, pp. 903-910,1994.
Canova, Fabio and Graziella Bertocchi, "Did Colonization Matter for Growth?
An Empirical Exploration into the Historical Causes of Africa’ s
Underdevelopment", London, CEPR Discussion Paper 1444, September 1996.
Cardoso, Eliana and Ann Helwege, "Populism, Profligacy, and Redistribution",
in in R. Dombush and S. Edwards (eds.), The Macroeconomics of Populism in
Latin America. Chicago, University of Chicago Press, 1991.
Cardoso, Eliana and Albert Fishlow, "Latin American Economic Development:
1950-1980", loumal of Latin American Studies. Vol. 24 Suppl., pp.197-218,1992.
Chatterji, Monojit, Brian Gilmore, Kurt Strunk and Jon Vanasin, "Political
Economy, Growth and Convergence in Less-Developed Countries", World
Development. Vol. 21, No. 12 (1993), pp.2029-2038.
Clague, Christopher; Keefer, Philip; Knack, Steve and Mancur Olson, "Contract-
Intensive Money: Contract Enforcement, Property Rights, and Economic
Performance", IRIS Working Paper 151, February 1995.
Chen, B. and Y. Feng, "Some Political Determinants of Economic Growth: Theory
and Empirical Implications", European loum al of Political Economy. Vol. 12, pp.
609-627,1996.
Conte, M. and A. Darrat. (1988). "Economic Growth and The Expanding Public
Sector: A Reexamination," Review of Economics and Statistics 70, 322-330.
157
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Dahl, Robert, A Preface to Economic Democracy, Berkeley, University of
California Press, 1985.
De Gregorio, Jose, "Economic Growth in Latin America", loumal of
Development Economics. Vol.39, No.l, pp. 59-84,1992.
De Gregorio, Jose and Pablo Guidotti, "Financial Developm ent and Economic
Growth", World Development, Vol. 23, No. 3, pp. 433-448, March, 1995.
Deininger, Klaus and Lyn Squire, "Measuring Income Inequality: A New Data
Base", World Bank, mimeo, March 1996.
Dewatripont, Mathias and Gerard Roland, 1997. "Transition as a Process of
Large-Scale Institutional Change", in D. Kreps and F. Wallis (eds) Advances in
Economics and Econometrics: Theory and Applications (Volume 2), Cambridge,
Cambridge University Press.
Diamond, Larry, Juan Linz and Seymour Lipset (Eds.), Democracy in Developing
Countries. Four volumes, Boulder: Lynne Rienner, 1989.
Dick, William, "A uthoritarian versus Nonauthoritarian Approaches to Economic
Development", loum al of Political Economy. Vol. 82 (1974), pp. 817-827.
Dixit, A. K., The Theory of Equilibrium Growth. Oxford, Oxford University
Press, 1976.
Dhonte, P. and I. Kapur, "Towards a Market Economy: Structures of
Governance," IMF W orking Paper 97-11,1997.
Easterlin, Richard, Growth Triumphant: The Twenty-first Century in Historical
Perspective, Ann Arbor, University of Michigan Press, 1996.
Easterly, William and Sergio Rebelo, "Fiscal Policy and Economic Growth: An
Empirical Investigation", loum al of Monetary Economics, Vol. 32 (1993), pp. 417-
458.
Edwards, Sebastian, Crisis and Reform in Latin America: From Despair to Hope.
Oxford, Oxford University Press, 1995.
Eggertson, T., "The Economics of Institutions in Transition Economies", in S.
Schiavo-Campo (ed), Institutional Change and the Public Sector in Transitional
158
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Economies, World Bank Discussion Papers 241, W ashington, D.C., World Bank,
1994.
Engerman, Stanley and Kenneth Sokoloff, "Factor Endowments, Institutions, and
Differential Paths of Grow th among New World Economies", NBER Historical
Paper No. 66, December 1994.
Fajnzylber, Fernando, Unavoidable Industrial Restructuring in Latin America.
Duke University Press, 1990.
Feng, Yi, "Democracy, Political Instability, and Economic Growth", Presented at
the Public Choice Society Annual Conference, Long Beach, CA, 1995.
Fishlow, Albert, "Some Reflections on Comparative Latin American Economic
Performance and Policy", in T. Banuri (ed.), Economic Liberalization. No
Panacea: The Experiences of Latin America and East Asia, New York, Oxford
University Press, 1991.
Fishlow, Albert, "Inequality, Poverty, and Growth: Where Do We Stand?"
Washington, D.C., Annual World Bank Conference on Development Economics,
1995.
Gastil, Raymond, "The Comparative Survey of Freedom: Experiences and
Suggestions", in Alex Inkeles (Ed.), On Measuring Democracy: Its Consequences
and Concomitants. New Brunswick, Transaction Publishers, 1991.
Granger, C.W., "Causal Inference", The New Palgrave: Econometrics. New York,
W.W. Norton, 1987.
Greene, William, Econometric Analysis. Prentice-Hall, 1990.
Grier, Kevin and Gordon Tullock, "An Empirical Analysis of Cross-National
Economic Growth, 1951-1980", loumal of Monetary Economics. Vol. 24 (1989),
pp. 259-276.
Gupta, Dipak, The Economics of Political Violence. New York, Praeger, 1990.
Gurr, Ted, Polity II: Political Structures and Regime Change. 1800-1986. ICPSR,
1990.
Gwartney, James; Lawson, Robert and Walter Block, Economic Freedom of The
World: 1975-1995. The Fraser Institute, 1996.
159
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Haber, Stephen, "Capital Markets and Industrial Development: A Comparative
Study of Brazil, India, Mexico, and the United States, 1840-1930", IRIS Working
Paper 144, November 1994.
Haggard, Stephen and Steven Webb (Eds.), Voting for Reform: Democracy.
Political Liberalization, and Economic Adjustment, New York: Oxford University
P ress, 1994.
Helliwell, John, "Empirical Linkages Between Democracy and Economic
Growth", British loumal of Political Science, Vol. 24, pp. 225-248,1994a
Helliwell, John, "International Growth Linkages: Evidence from Asia and the
OECD", in T. Ito and A.Krueger (Eds.), Macroeconomic Linkage: Savings,
Exchange Rates, and Capital Flows. Chicago, University of Chicago Press, 1994b.
Hibbs, Douglas, Mass Political Violence: A Cross-National Causal Analysis. New
York, Wiley & Sons, 1973.
Hirschman, Albert, Exit, Voice, and Loyalty: Responses to Decline in Firms.
Organizations, and States, Cambridge, Harvard University Press, 1970.
Hirschman, Albert, "The Rise and Decline of Development Economics", in Essays
in Trespassing: Economics to Politics and Beyond, Cambridge, Cambridge
University Press, 1981.
Hirschman, Albert, Shifting Involvements: Private Interest and Public Action.
Blackwell, 1982.
Hirschman, Albert, "The On-and-Ot'f Connection Between Political and
Economic Progress". American Economic Review Papers and Proceedings. Vol.
84, No. 2 (1994), pp. 343-348.
Holtz-Eakin, D., W. Newey and H. Rosen. (1988). "Estimating Vector
Autoregressions with Panel Data," Econometrica 56,1371-1395.
Hsiao, Cheng, "Causality Tests in Econometrics", loum al of Economic Dynamics
and Control. Vol. 1, No. 3, pp. 321-346, August 1979.
Hsiao, Cheng, Analysis of Panel Data. Cambridge, Cambridge University Press,
1986.
Huntington, Samuel, Political O rder in Changing Societies, New Haven, Yale
University Press, 1968.
160
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IRIS, 1996. Governance and the Economy in Africa: Tools for Analysis and
Reform of Corruption. College Park, Md.: The IRIS Center (Institutional Reform
and the Informal Sector).
Jaegger, Keith and Ted Gurr, "Polity III" (available by anonymous ftp from
isere.colorado.edu, at directory pub/datasets/polity3), 1996.
Knack, Stephen and Philip Keefer, "Institutions and Economic Performance:
Cross-Country Tests Using Alternative Institutional Measures", Economics and
Politics. Vol.7, No.3, pp. 207-227,1995.
Knack, Stephen, "Institutions and the Convergence Hypothesis: The Cross-
National Evidence", W orking Paper No. 59, College Park, IRIS, University of
Maryland, 1994.
Kohli, Atul, "Democracy and Development", in John Lewis and Valeriana Kallab
(Eds.) Development Strategies Reconsidered. New Brunswick: Transactions
Books, 1986.
Kormendi, Roger and Philip Meguire "Macroeconomic Determinants of Growth:
Cross-Country Evidence", loum al of Monetary Economics. Vol. 16 (1985), pp.
141-163.
Krugman, Paul, "Toward a Counter-Counterrevolution in Development
Theory", in Proceedings of the World Bank Annual Conference on Development
Economics 1992. W ashington D.C., The World Bank, 1993.
Lai, Deepak, The Poverty of 'Development Economics'. Hobart Paperback No.
16, London, Institute of Economic Affairs, 1983.
Lau, Lawrence, "The Sources of Long-Term Economic Growth: Observations
from the Experience of Developed and Developing Countries", in R. Landau, T.
Taylor and G. Wright (eds.), The Mosaic of Economic Growth. Palo Alto,
Stanford University Press, 1996.
Leijonhufvud, Axel, "Life Among the Econ", W estern Economic loumal. v. 11,
no. 3, pp. 327-337,1973.
Levine, Ross and David Renelt, "Cross-Country Studies of Growth and Policy:
Methodological, Conceptual, and Statistical Problems", W orking Paper No. 608,
Washington, D.C., The W orld Bank, 1991.
161
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Levine, Ross and David Renelt, "A Sensitivity Analysis of Cross-Country
Growth Regressions", American Economic Review. Vol. 82, No. 4, pp. 942-963,
September 1992.
Levine, Ross and Sara Zervos, "Looking at the Facts: What We Know about
Policy and Growth from Cross-Country Studies", W orking Paper No. 1115,
Washington, D.C., The World Bank, 1993.
Lin, Justin and Jeffrey B. Nugent, "Institutions and Economic Development", in J.
Behrman and T.N.Srinivasan (eds.), Handbook of Development Economics:
Volume 3A. Amsterdam, North-Holland, 1995.
Lipset, Seymour, "Some Social Requirements of Democracy: Economic
Development and Political Legitimacy", American Political Science Review. Vol.
53 (1959), pp. 69-105.
Londono, Juan Luis, "Poverty, Inequality, Social Policy, and Democracy",
presented at World Bank Annual Conference on Development in Latin America
and the Caribbean, Rio de Janeiro, 1995.
Londregan, John and Keith Poole, "Poverty, The Coup Trap, and The Seizure of
Executive Power", W orld Politics, Vol. 42 (1990), pp. 151-183.
Mankiw, G.(1995). "The Growth of Nations", Brookings Papers on Economic
Activity 1, 275-310.
Mankiw, Gregory, David Romer and David Weil, "A Contribution to the
Empirics of Economic Growth", Quarterly loumal of Economics, Vol. 107, No. 2,
pp. 407-437, May 1992.
Marsh, Robert, "Does Democracy Hinder Economic Growth in the Latecomer
Developing Nations?", Comparative Social Research. Vol. 2, No. 2 (1979), pp.
215-248.
Marsh, Robert, "Sociological Explanations of Economic Growth", Studies in
Comparative International Development, Vol. 23, No. 4 (1988), pp. 41-76.
Masih, A. and R. Masih. (1995). "Temporal Causality and the Dynamic
Interactions among Macroeconomic Activity within a Multivariate Cointegrated
System: Evidence from Singapore and Korea", Weltwirtschaftliches Archiv 131,
265-285.
162
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Mauro, Paolo, "Corruption and Growth", Quarterly loumal of Economics,
Vol.110, No. 3, pp. 681-712, August 1995.
Mbaku, John, "The Political Economy of Development: An Empirical Analysis of
the Effects of the Institutional Framework on Economic Development", Studies
in Comparative International Development. Vol. 29, No. 2 (1994), pp. 3-22.
McMillan, John, Gordon Rausser and Stan Johnson, "Freedoms and Economic
Growth", Working Paper, Washington, D.C., Institute for Policy Reform,
November 1991.
McMillan, John, "M arkets in Transition", in D. Kreps and F. Wallis (eds)
Advances in Economics and Econometrics: Theory and Applications (Volume 2),
Cambridge, Cambridge University Press, 1997.
Menard, Claude, "M arkets as Institutions versus Organizations as Markets?
Disentangling Some Fundamental Concepts", loum al of Economic Behavior and
Organization. Vol. 28, No. 2, pp 161-182, October 1995.
Muller, Edward and Mitchell Seligson, "Civic Culture and Democracy: The
Question of Causal Relationships", American loum al of Political Science. Vol. 88,
No. 3 (1994), pp. 635-652.
Murphy, Kevin, Schleifer, Andrei and Robert Vishny, "The Allocation of Talent:
Implications for Grow th", Quarterly loum al of Economics. Vol. 106, pp. 503-530,
May 1991.
Morley, Samuel, Poverty and Inequality in Latin America. Johns Hopkins
University Press, 1995.
Naim, Moises, "Latin America's Journey to the Market: From Macroeconomic
Shocks to Institutional Therapy", International Center for Economic Growth,
Occasional Paper N o.62,1995.
North, Douglass, "Sources of Productivity Change in Ocean Shipping, 1600-
1850," loumal of Political Economy. Vol 76, No. 5, pp. 953-970,1968.
North, Douglass, Institutions. Institutional Change and Economic Performance.
Cambridge , Cambridge University Press, 1990.
North, Douglass, "The New Institutional Economics and Third W orld
Development", in J. Hariss, J. Hunter and C. Lewis (eds.), The New Institutional
Economics and Third W orld Development. New York, Routledge, 1995.
163
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
North, Douglass, "The Contribution of the New Institutional Economics to an
Understanding of the Transition Problem", paper presented at UNU/WIDER,
Helsinki, March 1997.
Nugent, Jeffrey, "The Demise of Economic Development in Latin America and Its
Implications for Other Developing Countries", loumal of Economic
Development. Vol.16, N o.l, pp. 7-35, June 1991.
Nugent, Jeffrey, "Institutions, Markets and Developmental Outcomes", paper
presented at the Conference on Evaluation and Development: The Institutional
Dimension, Washington, World Bank, 1997.
Nugent, Jeffrey B. and M ustapha K. Nabli, "Development of Financial Markets
and the Size Distribution of Manufacturing Establishments: International
Comparisons," World Development. Vol. 20, pp.1489-1500, October 1992.
Olson, M. (1963). "Rapid Growth as a Destabilizing Force," loumal of Economic
History 23, 529-552.
Olson, Mancur, "Big Bills Left on the Sidewalk: Why Some Nations Are Rich, and
Others Poor", loumal of Economic Perspectives. Vol 10, No. 2, pp. 3-24,1996.
Oxley, L. (1994). "Cointegration, Causality and W agner's Law: A Test for Britain,
1870-1913", Scottish loum al of Political Economy 41. 286-298.
Ozler, Sule and Dani Rodrik, "External Shocks, Politics and Private Investment:
Some Theory and Empirical Evidence", loumal of Development Economics. Vol.
39 (1992), pp. 141-162.
Pastor Jr., Manuel and Eric Hilt, "Private Investment and Democracy in Latin
America", World Development. Vol. 21, No. 4 (1993), pp. 489-507.
Perotti, Roberto, "Income Distribution, Politics, and Growth", American
Economic Review Papers and Proceedings. Vol. 82, No. 2 (1992), pp. 311-316.
Perotti, Roberto, "Income Distribution and Investment", European Economic
Review. Vol. 38, pp. 827-835, April 1994.
Persson, Thorsten and Guido Tabellini, "Growth, Distribution, and Politics", in
A. Cukierman, Z.Hercowitz, and L. Leiderman (eds.), Political Economy.
Growth, and Business Cycles. Cambridge, The MIT Press, 1992.
164
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Persson, Torsten and Guido Tabellini, “Is Inequality Harmful for Growth?",
American Economic Review, Vol. 84, No. 3 (1994), pp. 600-621.
Porter, Philip and Gerald Scully, "Institutional Technology and Economic
Growth", Public Choice. Vol. 82, pp. 17-36,1995.
Pourgerami, Abbas, "The Political Economy of Development: A Cross-National
Causality Test of Development-Democracy-Growth Hypothesis", Public Choice.
Vol. 58 (1988), pp. 123-141.
Pourgerami, Abbas, Development and Democracy in the Third W orld. Boulder:
Westview Press, 1992.
Pourgerami, Abbas and Djeto Assane, "Macroeconomic Determinants of Growth:
New Measurement and Evidence of the Effect of Political Freedom", Applied
Economics, Vol. 24 (1992), pp. 129-136.
Pritchett, Lant, "Where Has All the Education Gone?", WPS 1581, The World
Bank, March 1996.
Przeworski, Adam and Fernando Limongi, "Political Regimes and Economic
Growth", loumal of Economic Perspectives, Vol. 7, No. 3 (1993), pp. 51-69.
Porter, Philip and Gerald Scully, "Institutional Technology and Economic
Growth", Public Choice. Vol. 82, pp. 17-36,1995.
Psacharopoulos, George, Returns to Education: An International Comparison.
Jossey-Bass Publishers, 1973.
Psacharopoulos, George and Ying Ng, "Earnings and Education in Latin
America", Education Economics. Vol. 2, No. 2, pp. 187-207,1994.
Raczynski, Dagmar, "Programs, Institutions, and Resources: Chile", in Dagmar
Raczynski (ed.), Strategies to Combat Poverty in Latin America. Baltimore, The
Johns Hopkins University Press, 1995.
Remmer, Karen, "Democracy and Economic Crisis: The Latin American
Experience", World Politics. Vol. 42, No. 3 (1990), pp. 315-325.
Reynolds, Lloyd, "The Spread of Economic Growth to the Third W orld: 1850-
1980", loumal of Economic Literature. Vol. 21, pp. 940-980,1983.
165
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Robinson, James, "Theories of Bad Policy", University of Southern California,
mimeo, 1996.
Rodrik, Dani, "Political Economy and Development Policy", European Economic
Review. Vol.36, No. 2-3, pp. 329-336, April 1992.
Roubini, Nouriel and Xavier Sala-i-Martin, "Financial Repression and Economic
Growth", loum al of Development Economics, Vol. 39 No.l, pp. 5-30,1992.
Rutherford, Malcom, 1994. Institutions in Economics: The Old and the New
Institutionalism ., Cambridge, Cambridge University Press, 1994.
Sachs, Jeffrey and Andrew Warner, "Economic Convergence and Economic
Policies", NBER Working Paper No. 5039, February 1995.
Saint-Paul, Gilles and Thierry Verdier, "Education, Democracy and Growth",
loum al of Development Economics. Vol. 42 (1993), pp. 399-407.
Sala-i-Martin, Xavier, "Cross-Sectional Regressions and the Empirics of
Economic Grow th", Economics W orking Paper No. 79 (Barcelona: Universitat
Pompeu Fabra, 1994).
Sanders, David, Patterns of Political Instability, New York, St. M artin's Press,
1981.
Scully, Gerald, "The Choice of Law and the Extent of Liberty", loum al of
Institutional and Theoretical Economics. Vol. 143 (1987), pp. 595-615.
Scully, Gerald, "The Institutional Framework and Economic Developm ent",
loum al of Political Economy. Vol. 86 No. 3, pp. 652-662,1988.
Sen, Am artya, "Economic Regress: Concepts and Features", in Proceedings of
The W orld Bank Annual Conference on Development Economics 1993.
W ashington D.C., The World Bank, 1994.
Shleifer, Andrei, "Establishing Property Rights", Proceedings of 1994 The World
Bank Annual Conference on Development Economics. March 1995.
Sloan, John and Kent Tedin, "The Consequences of Regime Type for Public-
Policy O utputs", Comparative Political Studies, Vol. 20, No. 1 (1987), pp. 98-124.
Solow, Robert, 1995. "But Verify: Review of Fukuyama's Trust", The New
Republic, Septem ber 11,1995.
166
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Solow, Robert, 1997. Learning from 'Learning by Doing': Lessons for Economic
Growth. Palo Alto, Stanford University Press.
Srinivasan, T.N., "Long-Run Growth Theories and Empirics: Anything New?", in
Proceedings of the University of Siena Workshop on Endogenous Growth and
Development, July, 1994.
Streeten, Paul, 1987. "Structural Adjustment: A Survey of Issues and Options",
World Development. 15 (12), pp. 1469-1482.
Summers, Robert and Alan Heston, "Penn World Tables, Version 5.6" (available
by anonymous ftp from www.nber.org, at pub/pwt56), 1994.
Taylor, C. and D. Jodice, W orld Handbook of Social and Political Indicators. New
Haven, Yale University Press, 1983.
Thomas, Vinod and Yan Wang, "Distortions, Interventions, and Productivity
Growth: Is East Asia Different?", Economic Development and Cultural Change.
Vol. 44, No. 2, pp. 265-288,1996.
Trebilcock, Michael, "W hat Makes Poor Countries Poor? The Role of Institutional
Capital in Economic Development", Univ. of Toronto, mimeo, November, 1995.
Venieris, Yiannis and Dipak Gupta, "Income Distribution and Sociopolitical
Instability as Determinants of Savings: A Cross-sectional Model", loumal of
Political Economy. Vol. 94, No. 4 (1986), pp.873-883.
Vorhies, Frank and Fred Glahe, "Political Liberty and Social Development: An
Empirical Investigation", Public Choice, Vol. 58 (1988), pp. 45-71.
Weede, Erich, "The Impact of Democracy on Economic Growth: Some Evidence
from Cross-National Analysis", Kvklos. Vol. 36, No. 1 (1983), pp. 21-39.
Wiesner, Eduardo, "Fiscal Decentralization and Social Spending in Latin
America: The Search for Efficiency and Equity", IDB W orking Paper No. 199,
1994.
Williamson, John (Ed.), The Political Economy of Policy Reform. Washington,
D.C., Institute for International Economics, 1993.
Wolf, Holger, "Growth Convergence Reconsidered", Weltwirtschaftliches
Archiv. Vol. 130, No. 4 (1994), pp. 747-759.
167
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
World Bank, The East Asian Miracle: Economic Growth and Public Policy.
Washington, D.C., The World Bank, 1993.
World Bank, Governance: The World Bank's Experience. Washington, D.C., The
World Bank, 1994.
Zellner, Arnold, "Causality and Causal Laws in Economics", loumal of
Econometrics. Vol. 39, No. 1/2, pp. 7-22, September/October, 1988.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IMAGE EVALUATION
TEST TARGET (Q A -3 )
1 . 0
l.l
g IS 1 M
•- I n 12.2
L ix
It
2.0
1 .8
1 .2 5
1.4
1 . 6
150mm
V
6 >
% s»
>
/ /
/
’ / /
A P P L I E D A IIWIGE . Inc
jg g g 1653 East Main Street
Rochester. NY 14609 USA
~ — Phone: 716/482-0300
Fax: 716/288-5989
O 1993. Applied Image. Inc.. All Rights Reserved
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

A theoretical and empirical examination of infant mortality decline in Brazil's northeast region, 1986--1996

PDF

Before and after reform: The effects of structural adjustment on women's employment in Latin America

PDF

Globalization and the decline of the welfare state in less developed countries

PDF

An analysis of the developmental state: The case of the Vietnamese textile and garment industries

PDF

Analysis of the role of government in Taiwan's industrialization and economic development

PDF

Compatibility standards, market power, and antitrust policy: Proprietary interfaces and their alternatives

PDF

The political economy of the state and development strategies: A case study of Turkey, 1930-1980

PDF

Dependence on foreign labor, quality of education and unemployment in the GCC countries: In search of solutions

PDF

The Role Of The State In The Development Of The Indonesian Textile And Garment Industries

PDF

Dialectical feminism and contradictory economic systems: Campesina daily life from Somoza to Aleman

PDF

An empirical analysis of foreign aid and the provision of international public goods

PDF

Key factors in appraising development project in Egypt

PDF

Economic transition, micro- and small -scale enterprise in Zimbabwe, and the significance of participatory development to institutional economics

PDF

An economic analysis of cooperative group behavior: Theory and evidence

PDF

The drivers of economic change: Technology, globalization, and demography

PDF

A multidimensional analysis of regional integration and cooperation: The case of the Economic Community of West African States

PDF

Changing role of the household as an income equalizer: An empirical study of Kuznets hypothesis on the case of Taiwan

PDF

Essays in environmental policy

PDF

Economic sanctions, domestic audiences, and international conflict

PDF

Characteristics of exemplary HUD subsidized housing facilities for low-income elders

Asset Metadata

Creator Campos, Nauro Ferreira (author)

Core Title Context is everything: An empirical study of institutions, instability and economic growth

Degree Doctor of Philosophy

Degree Program Political Economy and Public Policy

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

Tag Economics, General,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Nugent, Jeffrey (committee chair), [illegible] (committee member), Easterlin, Richard (committee member), Lopez-Lee, David (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c17-324269

Unique identifier UC11353752

Identifier 9835119.pdf (filename),usctheses-c17-324269 (legacy record id)

Legacy Identifier 9835119.pdf

Dmrecord 324269

Document Type Dissertation

Rights Campos, Nauro Ferreira

Type texts

Source 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 au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA