University of Southern California Dissertations and Theses

Innovation: financial and economics considerations

(USC Thesis Other)

Innovation: financial and economics considerations

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content i

INNOVATION: FINANCIAL AND ECONOMICS CONSIDERATIONS

By

Weiran Deng

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
ECONOMICS

August 2020

Copyright 2020 Weiran Deng
ii
Acknowledgements

I would like to express my special thanks of gratitude to my advisor, Prof. Fernando Zapatero,
for providing guidance and feedback throughout this project. I would also like to thank
professors in my dissertation committee, including Prof. Joshua Aizenman, Prof. Arthur
Korteweg, and Prof. Romain Ranciere, for their extremely valuable comments.

I am honored to have great coauthors. They are Dr. Andrea Belz, Prof. Yongxiang Wang, Prof.
Simon Wilkie, and Yinqi Zhang. Thank you for all the inspirations and support.

Last but not least, I would like to thank my family for constantly encouraging me during my
Ph.D. years. I could not realize my dream of being an economist without their continuous
support.

iii
TABLE OF CONTENTS
Acknowledgements ......................................................................................................................... ii
List of Tables .................................................................................................................................. v
List of Figures ................................................................................................................................ vi
Abstract ......................................................................................................................................... vii
Chapter 1: Evaluating Regulation on Vertical Integrations in the U.S. Television Industry .......... 1
Introduction ................................................................................................................................. 2
The Research Question ............................................................................................................... 5
Sample and Data Source ............................................................................................................. 7
Empirical Strategy ...................................................................................................................... 9
Results ....................................................................................................................................... 12
Discussion ................................................................................................................................. 14
Limitation and Future Research Directions .............................................................................. 15
Conclusion ................................................................................................................................ 17
Chapter 2: Founding Team’s Human Capital and Financial Resources— ................................... 26
Evidence from Nascent High-tech Ventures ................................................................................. 26
Introduction ............................................................................................................................... 27
Literature and Hypotheses Development .................................................................................. 28
Data and Summary Statistics .................................................................................................... 30
Empirical Analysis .................................................................................................................... 36
Robustness Checks ................................................................................................................... 38
Discussion ................................................................................................................................. 39
Limitations ................................................................................................................................ 40
Conclusion ................................................................................................................................ 41
Chapter 3: Do M&A Affect Patent Inventors’ Productivity? ....................................................... 54
Introduction ............................................................................................................................... 55
Hypotheses Development ......................................................................................................... 58
Data ........................................................................................................................................... 60
iv
Empirical Strategy .................................................................................................................... 64
Results ....................................................................................................................................... 65
Discussion ................................................................................................................................. 68
Conclusion ................................................................................................................................ 69
Chapter 4: How Investment in Private Equity Affect Economic Inequality in the U.S.? ............. 82
Introduction ............................................................................................................................... 83
Literature Review ..................................................................................................................... 84
Reconsidering the Amendment of Accredited Investor Standard ............................................ 96
The Mechanism that Investment in Private Equity Affects Inequality ................................... 100
Discussion ............................................................................................................................... 104
Conclusion .............................................................................................................................. 106
References ................................................................................................................................... 137
Appendices .................................................................................................................................. 154
Appendices for Chapter 1 ....................................................................................................... 154
Appendices for Chapter 2 ....................................................................................................... 173
Appendices for Chapter 3 ....................................................................................................... 175
Appendices for Chapter 4 ....................................................................................................... 179

v
List of Tables
Table 1.1. Structure of U.S. Television Industry ............................................................................ 6
Table 1.2. Qualified Vertical Integrations from 1993 to 2012 ....................................................... 9
Table 1.3. Two Sets of Synthetic Control ..................................................................................... 10
Table 1.4. Description of Information for Synthetic Control Method .......................................... 12
Table 2.1. Data Processing ........................................................................................................... 42
Table 2.2. Distribution of Funding Sources (Not Mutually Exclusive) ....................................... 43
Table 2.3. Descriptive Statistics .................................................................................................... 44
Table 2.4. Correlation Matrix ....................................................................................................... 45
Table 2.5. Team Size and Self-financing—Logit Model .............................................................. 47
Table 2.6. Covariates Balance of Entropy Balancing ................................................................... 48
Table 2.7. Self-financing and Team Size—Results of Entropy Balancing ................................... 49
Table 2.8. Subsample Ventures—without Mature Outside Financing ......................................... 50
Table 2.9. Subsample Ventures—Exclude Ventures with Zero Self-reported Founder ............... 51
Table 2.10. Different Thresholds of Large Team ......................................................................... 52
Table 3.1. Data Processing ............................................................................................................ 71
Table 3.2. Summary Statistics ...................................................................................................... 72
Table 3.3. Productivity of Acquirer Inventor ................................................................................ 74
Table 3.4. Productivity of Acquirer Inventor in Horizontal Mergers ........................................... 76
Table 3.5. Productivity of Target Inventor ................................................................................... 78
Table 3.6. Productivity of Target Inventor in Horizontal Mergers ............................................... 80
Table 4.1. Threshold and Average Incomes in Top Income Groups in the United States in 2010
..................................................................................................................................................... 111
Table 4.2. Threshold and Average Wealth in Top Wealth Groups in the United States in 2012 115
Table 4.3. Percent of Families that Hold Assets, 1989–2016 ..................................................... 116
Table 4.4. Mean Asset Holdings, 1989–2016 (thousands of 2016 dollars) ................................ 118
Table 4.5. Average Assets and Debts of Asset Groups, 1989–2016 (in thousands of 2016 dollars)
..................................................................................................................................................... 120
Table 4.6. Households Qualifying under Existing Accredited Investor Criteria ........................ 126
Table 4.7. Type and Frequency of Limited Partner .................................................................... 127
Table 4.8. Average Private Equity Pay in the U.S. ..................................................................... 128

vi
List of Figures
Figure 1.1. Percentage Changes in Affiliate Fees from 2011 to 2017 .......................................... 19
Figure 1.2. Counterfactual of “The Travel Channel” ................................................................... 20
Figure 1.3. Counterfactual of “ROOT Sports Northwest” ............................................................ 21
Figure 1.4. Counterfactual of “AT&T Sports Pittsburgh” ............................................................ 22
Figure 1.5. Average Treatment Effects on the Treated of “The Travel Channel” ........................ 23
Figure 1.6. Average Treatment Effect on the Treated of “ROOT Sports Northwest” .................. 24
Figure 1.7. Average Treatment Effect on the Treated of “AT&T Sports Pittsburgh” .................. 25
Figure 4.1. Top Income Shares in the United States, 1917-2010 ............................................... 109
Figure 4.2. Top Wealth Shares in the United States, 1917-2012 ................................................ 112
Figure 4.3. Composition in Shares of Assets of the Wealthiest 1 Percent, 1989-2016 .............. 121
Figure 4.4. Private Equity Fundraising Activity in the United States, 2006-2019 ..................... 122
Figure 4.5. Private Equity Deal Activity in the United States, 2006-2019 ................................. 124
Figure 4.6. Income Composition of Top Groups within the Top Decile in 1929 and 2007 ....... 129
Figure 4.7. Number of U.S. Listings, 1980-2018 ....................................................................... 131
Figure 4.8. U.S. Unicorn Count and Aggregated Post-valuation ($B), 2006-2018 .................... 132
Figure 4.9. U.S. VC Pre-money Valuation ($M) by stage, 2006-2019 ...................................... 133
Figure 4.10. Quartile Breakdown for Valuation at Exit for U.S. VC, 2006-2019 ...................... 135

vii
Abstract
“Changes call for innovation, and innovation leads to progress.” –Li Keqiang (Premier of the
People’s Republic of China)
Innovation, which shapes the future of the world, is the key to growth and prosperity.
This dissertation explores topics around innovation from multiple perspectives. In terms of the
company-lifecycle while doing innovation, the dissertation covers both high-tech startups and
dominant companies. Regarding people involved in innovation activities, this dissertation pays
attention to startup founders, patent inventors, and private company investors. From the point of
innovation policy, this dissertation provides insights on both the regulation on dominant
companies and the protection to average investors.
Chapter 1 investigates the effect of the Program Access Rule (PAR), an antitrust law on
input prices in the U.S. television industry, through applying the Generalized Synthetic Control
Method. Results show that vertical integration itself increases input prices, while combining it
with the PAR can hold prices almost unchanged. The empirical strategy we propose has the
potential to be implemented in a broader context, especially in the fast-growing high-tech
industry where vertical integration raised the most concern.
Chapter 2 explores nascent high-tech ventures prior to incorporation. Self-financing is the
most common financing strategy for ventures in their earliest years. We find that entrepreneurial
teams are more likely to invest in their own businesses if their founding teams are smaller. Our
finding suggests that the founding team’s human capital can potentially substitute for its
financial commitment. From a policy perspective, it becomes important to determine how to
provide both human and financial resources to nascent high-tech ventures. In the strategy
viii
domain, this has important implications for early decision-making since the lack of funding may
have subsequent consequences.
Chapter 3 studies whether and how mergers and acquisitions (M&A) affect the
productivity of patent inventors. Career paths of inventors are identified by their patent records.
Completed and withdrawn M&A biddings are exploited as a quasi-experiment. Results show that
for both acquirer inventors and target inventors, the number of patents will drop after
experiencing M&A. When the acquirer and target are in the same industry, the quality of patent
also drops for both acquirer- and target inventors. The results support the hypothesis of cultural
conflict and competition for funding among inventors. We did not see evidence supporting
collaboration among inventors or increased productivity due to increased pressure on job security
concerns.
Chapter 4 sheds light on a current debate on government policy and regulation. In the
past few years, income and wealth concentration in the United States have drawn tremendous
concern. As private equity attracts more and more attention, some people have connected
inequality with the exclusive private equity investment opportunity to accredited investors. I
argue that (1) marginal accredited investors do not benefit from the opportunity of investing in
private equity; (2) general partners of startup companies, including general partners in private
equity firms and angel investors, gain excess returns from private equity investment, which is
one of the factors driving the increasing income and wealth concentration.

1

Chapter 1: Evaluating Regulation on Vertical Integrations in the U.S.
Television Industry
Weiran Deng, Simon Wilkie
1
, Yinqi Zhang
2

Abstract
This article investigates the effect of the Program Access Rule (PAR), an antitrust law on input
prices in the U.S. television industry, through applying the Generalized Synthetic Control
Method. During the period between 1993 to 2012, the PAR was implemented to prevent cable
television distributors, who owned networks through vertical integration, from (1) making
exclusive contracts and (2) conducting price discrimination. Since the PAR and vertical
integration can generate economic impact at the same time, the price effects from these two
events are not differentiated in existing literature using a difference-in-differences strategy. In
this paper, we will estimate the price effect of the PAR by exploiting an unexplored feature of
the program, specifying that the rule is only imposed on cable distributors and not on satellite
distributors, which enables comparison between results from two sets of generalized synthetic
control method. Current results show that vertical integration itself increases input prices, while
combining it with the PAR can hold prices almost unchanged. It shows that the PAR can restrict
the increases in input prices that are caused by vertical integration.

Keywords: Vertical Integration, Regulation, Synthetic Control Method
JEL Codes: K21, K23, L82

1
Monash Business School, Monash University
2
Department of Economics, Dornsife College of Letters, Arts, and Sciences, University of Southern California

2
Introduction
Vertical integration is a business strategy that prevails in many upstream-downstream
industries. It has pro-competitive effects on the market through the elimination of double
marginalization and the alignment of investment incentives (Grossman and Hart 1986; Spengler
1950). However, when companies gained significant market power, their vertical integration
proposals often raised concerns about potential incentives to foreclose rivals (Hart and Tirole
1990; Ordover, Saloner, and Salop 1990). The most recent example is the merger between
AT&T and Time Warner. The Department of Justice officially challenged the merger “partially
because of the fear that the merged entity might demand higher prices and more favorable terms
from its rivals and seeks to enjoin the transaction” (United States District Court for The District
of Columbia 2017). Back in 2009, in their review of the Comcast-NBCUniversal merger, the
FCC was clear in its beliefs about the impacts of vertical integration of a multichannel video
programming distributors into the programming market: “Comcast could use exclusionary
program access strategies to reduce competition from all significant current and potential rivals
participating in those markets.” (Baker 2011).
The two key players in the U.S. television industry are multichannel video programming
distributors (“MVPDs”), such as Comcast, and content producers (networks/ channels), such as
CNN. Downstream MVPDs negotiate with content producers over a monthly per subscriber
“affiliate fee” which is an input price as to the final price that MVPDs charge consumers. There
are usually three downstream MVPDs as choices: a local cable company (e.g., Comcast or Time
Warner Cable) or one of two nationwide satellite companies (DirecTV and Dish Network)
3
.

3
This eliminates the potential impact from new entrants like Google, and now over-the-top provides like Netflix,
both of which appeared after the sample period of this study.
3
Though satellite distributors are now considered as equally powerful as cable distributors,
they were new entrants to the industry during the early 1990s. To protect these new entrants,
“Cable Television Consumer Protection and Competition Act” was implemented in 1992
(Federal Communications Commission 1992). Among the provisions in this Act, the Program
Access Rule (PAR) is the most influential one, preventing cable distributors who own networks
through vertical integration to (1) make exclusive contracts or (2) make discriminative pricing to
their non-competitors and competitors. In 2012, PAR was allowed to lapse. This paper focuses
specifically on the regulation period from 1993 to 2012.
Difference-in-differences (“DiD”) regression is the “standard methodology for estimating
the effects of mergers” (Kwoka 2015). This strand of literature includes Hosken and Taylor
( 2007), which applies the methodology in the U.S. petroleum industry, Kwoka and Shumilkina
( 2010), which analyzes an airline merger, Farrell et al. ( 2009) and Ashenfelter et al. ( 2011),
which both focus on hospital mergers. Previous empirical work studying the price effect from
mergers, targeting the same industry same time, is Ford (2018). That paper applies the DiD
methodology to quantify the effects of the Comcast-NBCUniversal merger and shows that the
input price is not increased after the merger. However, this study cannot separate the effect of the
merger itself from the effect of the regulation along with the merger. As a result, it is unclear
whether the regulation is effective or not on regulating input prices. Crawford et al. (2018) builds
a structural model to evaluate the vertical merger’s impact on consumer welfare systematically.
Their model shows that the PAR is effective in reducing integrated firms’ incentives to increase
rivals’ input costs. Our empirical analysis finds similar results by using a more intuitive reduced-
form method, which is the generalized synthetic control method.
4
This paper exploits an unexplored feature of the regulation: only cable distributors are
subject to the PAR, yet vertical integration happened to both cable and satellite distributors. This
feature offers an opportunity to see two different types of effects: effects from vertical
integration and PAR by evaluating how the affiliate fee (input price) changes before and after a
cable distributor bought a network, and effects purely from vertical integration by evaluating
how affiliate fee changes before and after a satellite distributor bought a network. The effect of
the PAR is then identified by comparing these two effects.
Due to the fact that only very few networks strictly fit the situation required for this
analysis—independently operated for a long period, then acquired by either a cable or satellite
distributor during the period of 1993 to 2012, synthetic control method naturally become the best
fitted empirical strategy. Synthetic control is a method designed for case studies, as each
treatment is regarded as one particular case. It is intuitive for industry practitioners and has been
developed quickly in recent years. Specifically, we apply Generalized Synthetic Control, which
is an extended version by Xu (2017), for exploiting its feature of allowing negative weights
while constructing synthetic units.
We show that vertical integration itself leads to an increase in affiliate fees, i.e., the
vertically integrated MVPD increases the input cost of its competitors. However, when the PAR
works together with vertical integration, affiliate fee is unchanged, which implies that the
regulation effectively protects other competitors in the market from the negative impact of a
vertical integration.
This paper is related to several strands of literature. First of all, it contributes to the
evaluation of antitrust law. Secondly, as comparative case studies, the cases we analyze in this
paper contribute to the understanding of influential mergers and acquisitions in the U.S.
5
Television industry. Last but not least, the paper shows an effective empirical strategy to analyze
antitrust law where regulation and events often coincide.
The paper is structured as follows: the first section introduces the industry and the
Program Access Rule. We formalize our research question in the following “the research
question” section. The third section describes the sample and data source. The empirical strategy
is explained in the fourth section. In the fifth section, we show results from the generalized
synthetic control method. Section six explains some points that have not been fully explained.
Section seven discusses the limitations of this study and future research directions. Section eight
concludes.
The Research Question
The underlying difference between cable and satellite distributors relies on how they
transmit video signals to households. Cable distributors transmit their video signals through a
physical wire, whereas satellite distributors transmit the signals wirelessly. The Cable Television
Consumer Protection and Competition Act of 1992 requires “effective competition” in cable
industry which was defined by “fewer than 30 percent of the households in the franchise area
subscribe to the cable service of a cable system”
4
unless a multichannel video programming
distributor is operated by the franchising authority (Federal Communications Commission 1992),
making cable distributors to share the market rather than compete with each other. As a result,
during the period that we study, U.S. households were able to subscribe to a multichannel
television bundle from one of the three downstream MVPDs: a local cable company (e.g.,

4
According to the Cable Television Consumer Protection and Competition Act of 1992, the franchise area is “(i)
served by at least two unaffiliated multichannel video programming distributors each of which offers comparable
video programming to at least 50 percent of the households in the franchise area; and (ii) the number of households
subscribing to programming services offered by multichannel video programming distributor exceeds 15 percent of
the households in the franchise area.”
6
Comcast, or Time Warner Cable) or two nationwide satellite companies (DirecTV and Dish
Network). The market structure of the U.S. television industry is shown in table 1.1.
Competitors
Non-competitors Satellite 1 Satellite 2
Cable 1 Cable 2 Cable 3

Table 1.1. Structure of U.S. Television Industry
Suppose Cable 1 merges with one of the content providers. Then Cable 1 will try to
increase the affiliate fees for its competitors (i.e., Satellite 1 & 2)
5
. The PAR program tries to
limit Cable 1’s ability to raise affiliate fees because it requires Cable 1 to also increase the
affiliate fees for Cable 2 & 3, which will result in a decrease in quantity sold that is unwilling to
see. Since Cable 1 does not have any incentives in raising the input costs of Cable 2 & 3, the
PAR is able to reduce the negative impact of Cable 1’s vertical integration on Satellite 1 & 2.
Figure 1.1 is motivating statistics, illustrating the potential effects of the PAR. There are
14 networks in total that have unchanged ownership by distributors from 2011 to 2017. Among
them, 12 are owned by cable distributors (blue bars in figure 1.1), and the other two are owned
by satellite distributors (red bars in figure 1.1). We can see that when PAR was no longer
effective after 2012, there are significant increases in affiliate fees for networks owned by cable
companies, while the increases of that for networks owned by satellite distributors are much
smaller.
Therefore, we try to test the following hypothesis in this study:

5
The PAR prevents cable distributors to make exclusive contracts.
7
The existence of other nonrival cable distributors can make the PAR, which only applies
to cable distributors, effective in controlling the increase of affiliate fees due to vertical
integration.
Sample and Data Source
The Sample
This paper takes all 344 TV networks operated in the U.S. into consideration, including
types of Basic Cable, Broadcast, Home Shopping, Premium, and Regional Sports Networks. The
sample period is 1993 to 2012, during which PAR is effective.
Operational Information of Networks
All network information is from Market Intelligence database of S&P Global, which is
the best in class for cable network economics. For each network, the following information is
available: subscribers, average subscribers, inner market subscribers, outer market subscribers,
network satellite subscribers, affiliate revenue, affiliate revenue per average subscriber per
month, gross advertising revenue, net advertising revenue, other operating revenue, operating
revenue net, programming expenses, operating SG&A expenses, total operating expenses, cash
flow, and cash flow margin. Definition of these variables are in Appendix 1-A.
In addition, we calculate two extra variables, profit and bargaining power, which are
more directly related to our analysis. Their definitions are also summarized in Appendix 1-A.
The database covers data from 1989 to 2021 (prediction data if exceed current year).
However, since launch dates can vary among networks and many of them had already defunct,
the available years could vary for each network.
Information about Vertical Integration
8
Yearly ownership information and all Mergers and Acquisitions information that induces
ownership changes of networks are recorded in a yearbook published by SNL, called
“Economics of Basic Cable Networks”. 2012, 2011, 2010, and 2009 version of this yearbook is
available online through a subscription of the SNL database. By request, we got other historically
yearbooks from SNL’s agent, including the version of 2007, 2006, 2004, 1998, 1996, 1994, and
1993.
Treatment Units
The purpose of the empirical work in this paper is to see how the affiliate fee changes
after a network is acquired by a distributor. Thus, the very first thing is to find out which
networks were acquired by distributors between 1992 and 2012, with long enough time keeping
ownership unchanged. We require the ownership status to remain unchanged for at least three
years both before and after the vertical integration.
To find out which networks are transacted from independent owners to distributors, we
first pay attention to all the M&A transactions of networks, selecting those that distributors are
involved in. For these transactions, we double-check with the yearly ownership information to
make sure that the ownership neither changes throughout the years before the transaction nor
changes throughout the years after the transaction. In addition, we require the networks to have
at least ten years of history before the vertical integration and have no idiosyncratic material
change during the sample period
6
.
After the whole procedure, we found that though there were many vertical integrations in
the industry, there are only very few networks that have clean histories of ownership changes as

6
We have one case, which is “AT&T Rocky Mountain”, that meets all other requirements but had idiosyncratic
material change during the sample period. The “AT&T Rocky Mountain” lost the broadcast rights of its main
contents (NBA’s Denver Nuggets and NHL’s Colorado Avalanche) in 2004 when the owner of teams, Stan
Kroenke, launched the competing regional sports network Altitude Sports and Entertainment.
9
we required. Network names and associated transactions that are qualified for our analyses are
listed in Table 1.2.

Date Seller Buyer Price ($mil.) Terms (%) Network
March 2007 Discovery
Comm.
Cox
Communications
(cable)
1,018.2 75 The Travel
Channel
May 2009 Liberty Media Direct TV
(satellite)
400.5 65 GSN
May 2009 Liberty Media Direct TV
(satellite)
141.2 100 ROOT Sports
Northwest
May 2009 Liberty Media Direct TV
(satellite)
129.9 100 AT&T Sports
Pittsburgh

Table 1.2. Qualified Vertical Integrations from 1993 to 2012

Empirical Strategy
Since only very few networks are qualified for treatment units, the Synthetic Control
Method is naturally the most fitting empirical strategy for our study as it is designed for case
studies.
Introduction of Synthetic Control Method
The Synthetic Control Method was first introduced in the paper by Abadie et al. (2010).
It generalizes the usual difference-in-differences, a fixed-effects estimator, by allowing
unobserved confounding factors to vary over time. For a given treated unit, Synthetic Control
Method uses a data-driven algorithm to compute an optimal control (synthetic unit) from a
weighted average of potential candidates not exposed to the treatment (donor pool). The
synthetic unit presents what would have happened to the treatment unit if it had not received
treatment. The weights are chosen to best approximate the characteristics of the treated unit
during the pre-treatment period. The treatment effect is the difference between the value of the
10
outcome variable for the treatment unit and the synthetic unit post-treatment. The method fits
case studies the most since the method is applied to each treatment unit.
In this paper, we use the Generalized Synthetic Control (GSC) method, developed by Xu
( 2017), which allows negative weights to be assigned to units in the donor pool. When partial
correlations of pre-treatment covariates are not all non-negative, this method can improve the
out-of-sample prediction of the original synthetic control method. Compared to Doudchenko and
Imbens (2017), which is another extension of the original synthetic control method that relaxes
the constraint on non-negative weights, the GSC method considers interactive fixed effect to deal
with unobserved time-varying confounders, which is very likely to happen in our case.
As Table 1.3 shows, two sets of synthetic control are constructed in this paper. One for
the networks vertically integrated with cable distributors, the other is for the networks vertically
integrated with satellite distributors. The former is under the PAR regulation, while the latter is
not under the PAR regulation. The effect from PAR is identified by comparing the “Vertical
Integration + PAR” effect, which is from the first synthetic control, and the “pure VI” effect,
which is from the second synthetic control
7
.
Synthetic Control Treatment unit Controls
The first set A vertically integrated and
regulated network
Not vertically integrated networks
The second set A vertically integrated but not
regulated network
Not vertically integrated networks

Table 1.3. Two Sets of Synthetic Control
“The Travel Channel” is a treatment unit that belonged to the first set of synthetic control,
and “GSN”, “ROOT Sports Northwest”, “AT&T Sports Pittsburgh” are treatment units that
belonged to the second set of synthetic control. However, “GSN” is too special to be regarded as

7
More details on how the comparison can be made is in Appendix 1-B.
11
a valid treatment, as it has the lowest affiliate fee throughout time among all the networks. Thus,
we exclude this network in the final analysis
8
.
To run the synthetic control algorithm, we need to clarify the treatment unit, the treatment
period, predictors, and the donor pool. We will discuss predictors and donor pool then
summarize all other information for each treatment.
The Predictors
The outcome variable is affiliate fee per average subscriber per month. Predictors are
variables that can explain the outcome variable. After controlling predictors, all changes after
treatment can then be attributed to the effect from the treatment. Here we use two types of
predictors for outcome variable. First, we use value of outcome variables in prior years. For
example, “The Travel Channel” is treated in 2007, then its affiliate fee per average subscriber per
month in any years between 1993 to 2007 can be used as predictors. These are to capture
unobserved idiosyncratic characteristics that may affect the outcome variable. Second, we use
variables that are the best proxies for factors that determine the demand and the supply of
networks. The quality of networks affects the demand. We use net advertising revenue per
average subscriber per month, cash flow per average subscriber per month, and cash flow
margin as proxies for the quality of networks. The cost of networks affects the supply. We use
programming Expenses divided by total revenue, which also captures bargaining power, to
measure to what extent programming costs are influential for content providers.
In a sum, predictors for each treatment unit include: Net Advertising Revenue per
Average Subscriber per Month, Cash Flow per Average Subscriber per Month, Cash Flow

8
This point is illustrated in Figure 1-D-4 of Appendix 1-D.
12
Margin, Bargaining Power, and Affiliate Fee per Average Subscriber per Month in years prior to
treatment period.
Donor Pool
Donor Pool consists of networks that were never vertically integrated during the sample
period. The donor pool could be different for each treatment unit because: (1) the launch date for
each treatment unit could be different so that the pre-treatment period is different, (2) predictors
could be missing for some potential control units during the pre-treatment period of each
treatment unit.
Treatment unit, treatment period, pre-treatment period, and associated number of
networks in the donor pool for each treatment unit are summarized in Table 1.4.

Treatment unit Treatment period Pre-treatment period Number of networks in
donor pool
The Travel Channel 2007 1997-2006 48
ROOT Sports Northwest 2009 1993-2008 25
AT&T Sports Pittsburgh 2009 1993-2008 25

Table 1.4. Description of Information for Synthetic Control Method
Results
The causal inference from the Generalized Synthetic Control Method mainly predicts
potential outcomes in posttreatment periods. Along the way, results include selected synthetic
units with their assigned weights.
Synthetic Units
The Generalized Synthetic Control Method gives us unit weights for constructing
synthetic units and predictor balance between treatment units and synthetic units. Those for each
network are summarized in Appendix 1-C.
13
There are 46 networks assigned non-zero weights while constructing the synthetic unit
for “The Travel Channel”. Predictors are very close to each other between “The Travel Channel”
and its synthetic unit. As to “ROOT Sports Northwest”, there are 26 networks assigned non-zero
weights while constructing the synthetic unit for it. Predictors become close to each other
between this network and its synthetic unit except “cash flow per average subscriber per month”
and “net advertising revenue of per average subscriber per month”. Mentioning “AT&T Sports
Pittsburgh”, 24 networks are assigned non-zero weights while constructing its synthetic unit. All
predictors, except “net advertising revenue of per average subscriber per month”, reach balance.
The Estimated Average Treatment Effects on the Treated
The actual outcomes of treatment units and their estimated counterfactuals by period are
shown in Figure 1.2 to Figure 1.4. The uncertainty estimates of the GSC estimator relies on the
parametric bootstrap procedure. In Figure 1.5 to Figure 1.7, we show the estimated average
treatment effect on the treated units by period.
Remind that the only case showing the effect of “Vertical Integration (VI) plus PAR” is
“The Travel Channel”. The average treatment effect on the affiliate fee per average subscriber
per month is zero, as figure 1.5 shows. The cases showing effect purely from vertical integration
are illustrated through the analyses on the other two networks. The average treatment effects on
the affiliate fee per average subscriber per month of the two networks are both positive and get
larger throughout time, as shown in figure 1.6 and figure 1.7, though the increases are not
significantly different from zero.
These results suggest that satellite distributors have incentives to increase input prices in
order to reduce competition from other cable companies, while regulation on vertical integrations
14
in the U.S. television industry can control input prices. The results are consistent with our
hypothesis.
Discussion
Contract Length
In general, the length of contracts between MVPDs and networks is five years. For this
reason, the treatment effect for a sample with many treatment units can be interpreted as twenty
percent of the actual average treatment effect with other assumptions. In our scenario, since the
synthetic control method is applied to only one treatment unit in each case, the treatment effect
we observe is the actual treatment effect. Fixed contract length explains why the treatment effect
might not show up right after the vertical integration.
The Importance of Separating the Effect from Regulation
Study that is most close to the current one is Ford (2017), in which all vertical integration
cases are used in constructing the sample in order to apply the difference-in-differences method.
The results of that study show no increase in affiliate fees after vertical integration when the
PAR exists, which is the same as what we show in this paper when vertical integration happens
to cable distributors.
However, Ford (2017) offers no information on whether the PAR is effective. The failure
to do such an analysis left difficulties for economists and antitrust practitioners to judge whether
no increase in affiliate fee after the merger is because the merger fundamentally does not create
incentives for increasing the input price, or is because the PAR effectively restraints the MVPDs
to do so. This is especially important when it comes to proposing new regulations in other
industries, as we must first know the effect of regulation itself. For example, in the fast-growing
high-tech industry, there are so-called “superstar” firms that have gained unparalleled market
15
power throughout time. Those firms are acquiring startups more frequently, which has raised
concerns about suppressing entrepreneurship and innovation. The debate on how to properly
regulate “superstar” firms has never been stopped. The method proposed in our paper offers a
helpful angle on evaluating antitrust law.
The Advantage of Reduced Form Merger Analysis
Another method that can answer the same question as we did is in Crawford et al. (2018).
That paper uses a structural demand estimation, which is known as the “BLP Method” of
econometric estimation
9
. The “BLP Method” is able to separate the effect of vertical integration
from the effect of the PAR. However, the method also has its drawback, which is being unable to
account for any changes associated with the treatment as the demand curves are assumed to be
the same in both pre-treatment and post-treatment period (Peters 2006).
In the contrary, the synthetic control method we used in this study has already considered
changes along the time. It is a much more intuitive way to compare the actual outcomes of the
treatment unit and the potential outcome of its counterfactual during the post-merger period. In
general, reduced-form estimation is not good at disentangling from all the mechanisms that
induce the effect on the outcome. This paper, however, propose to compare results from two sets
of reduced form estimation and gain insights on mechanisms.
Limitation and Future Research Directions
Endogeneity
To address causal effect on the affiliate fee, one crucial assumption is that “error term of
any unit at any time is independent of treatment assignment, observed covariates, and
unobserved cross-sectional and temporal heterogeneities of all units at all times” (Xu 2017). We

9
The method is named after Berry, Levinsohn & Pakes (1995).
16
admit that the endogeneity might be an issue for our study if the primary purpose is to get a
causal effect on the affiliate fee from vertical integration (and the PAR). However, our goal is to
evaluate the PAR by comparing two sets of effect from the exact same method. If there is indeed
endogeneity, the issue shall exist in both the effect from the vertical integration of cable
distributor and the effect from the vertical integration of satellite distributor, which are supposed
to be differenced out by comparison.
The Validity of Comparison
Two points might make our comparison between the two sets of synthetic control results
invalid. First, the comparison is made between different types of networks. “The Travel
Channel” is a basic cable channel, while “ROOT Sports Northwest” and “AT&T Sports
Pittsburgh” are both sport channels that have more considerable bargaining power when the
contract is made. The contract between the distributor and content provider is often made as a
bundle that includes a number of networks. Yet the affiliate fee per average subscriber per month
(outcome) is set for each single network. The network that has more bargaining power will get a
better outcome. Though we do not control network types, we do have proxies for bargaining
power and we control values of affiliate fee in pretreatment periods to capture unobserved
idiosyncratic characteristics, so we believe the comparison remains valid.
Secondly, the comparison is made between treatments that happened in different years.
The difference between two treatment effects may also capture effects from other factors that
occurred during the different years. However, this is the best we can do to get insights into the
effect purely from the PAR.
Cross validation and future research direction
17
The synthetic control method has its limitation. Its case study nature makes the
generalization of results from the synthetic control method to be hard. If we could have several
more treatment units, cross-validation would be helpful in gaining confidence in generalizing the
results to other scenarios. If we could have even more treatment units in both cable vertical
integration case and satellite vertical integration case, we can compare results from two sets of
difference-in-differences results. This could also help in verifying our current results from the
generalized synthetic control method with an imperfect fit.
Enlarge the number of qualified treatment units and then compare two sets of “DID”
results is a feasible future research direction. In this study, we require the treatment unit to have
at least a ten-year pre-treatment period to meet the requirements of applying the synthetic control
methods. If we could relax this constraint, more vertical integration cases are likely to be
qualified treatments, which will allow us to verify the results by other methods.
Conclusion
By investigating into an unexplored feature of the regulation on vertical integration and
applying generalized synthetic control method, we find that Program Access Rule (PAR)
implemented in the U.S. television industry during 1993-2012 prevent cable distributors from the
increasing price of networks they owned after vertical integration. The novelty of our method is
on separating the effect of the PAR from the effect of vertical integration.
The paper has its limitations. The price effect we investigated in the two sets of synthetic
control models maybe not enough to address potential endogeneity issues. The comparison we
made between the two sets of synthetic control results are based on different channel type and
treatment in different years. Despite its limitation, this study provides the most intuitive and
practical empirical strategy for distinguishing the price effect from vertical integration itself and
18
regulation on the vertical integration. Regulations on vertical integration become increasingly
important as more and more large technology companies are accused of abusing their market
power. The empirical strategy we propose here has the potential to be implemented in a broader
context.

19
Figure 1.1. Percentage Changes in Affiliate Fees from 2011 to 2017

0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
80.00%
90.00%
100.00%
Bravo
Chiller
Cloo
CNBC
CNBC World
E!
Golf
MSNBC
Oxygen
SyFy
Universal HD
USA
GSN
MLB Network
20
Figure 1.2. Counterfactual of “The Travel Channel”

21
Figure 1.3. Counterfactual of “ROOT Sports Northwest”

22
Figure 1.4. Counterfactual of “AT&T Sports Pittsburgh”

23
Figure 1.5. Average Treatment Effects on the Treated of “The Travel Channel”

24
Figure 1.6. Average Treatment Effect on the Treated of “ROOT Sports Northwest”

25
Figure 1.7. Average Treatment Effect on the Treated of “AT&T Sports Pittsburgh”

26

Chapter 2: Founding Team’s Human Capital and Financial Resources—
Evidence from Nascent High-tech Ventures

Weiran Deng and Andrea Belz
10

Abstract
In this paper, we explore nascent high-tech ventures prior to incorporation. Self-financing is the
most common financing strategy for ventures in their earliest years. Theories from human capital
management and startup strategy predict opposing effects for the relationship between team size
and self-financing. We find that entrepreneurial teams are more likely to invest in their own
businesses if their founding teams are smaller. Our finding addresses the importance of team size
and suggests that the founding team’s human capital can potentially substitute for its financial
commitment.

Keywords: Entrepreneurial Finance, Human Resources, Innovation
JEL Codes: L26, J24, O30

10
Viterbi School of Engineering, University of Southern California
27
Introduction
Startups contribute disproportionately to economic growth (Haltiwanger, Jarmin, and
Miranda 2013) with those in high-tech industries potentially having an outsized impact (Aghion
and Howitt 1994; Solow 1956, 1957). However, little is known about the earliest stages of
venture launch. Many data sources collect ventures’ information since they are registered as
firms. For instance, by using data from Kauffman Firm Survey, Robb and Robinson find that
entrepreneurs hold “levered equity claims” in their own startups. The reason is that entrepreneurs
seek debt sources from personal balance to fund the startups (Robb and Robinson 2014). By
analyzing the Longitudinal Business Database for the US private, nonfarm sector, Haltiwanger
shows that more than 90% of startups have less than 20 employees, making up half of all the
employment from startups (Haltiwanger 2015). Studies of even the earliest firms already
experience a form of selection bias, as the teams have already crystallized to the point of
incorporation. Although many activities predate this legal action, sparse data on coalescing firms
have limited the understanding of the first steps toward venture launch.
The resource-based view of the firm suggests that resources generate competitive
advantage (Jay Barney 1991; Wernerfelt 1984), and that these comprise human, social, financial,
physical technology and organizational resources (Dunne et al. 2019). For nascent ventures,
financial resources create positive cash flow to avoid early failure (Liao, Welsch, and Moutray
2008; Thornhill and Amit 2003), prior to capturing market opportunities (Auken and Neeley
1996).
On the other hand, the founding team plays a critical role during venture’s earliest stage
and may represent the firm's first asset (e.g., Nelson 2003). Founders may bring their talents,
skills, and networks into the venture (Choi et al. 2019; Muñoz-Bullon, Sanchez-Bueno, and Vos-
28
Saz 2015; Tzabbar and Margolis 2017). However, literature on the impact of founding team on
venture financing is sparse and shows mixed results where it exists.
In this study, we leverage a novel survey data describing high-tech ventures, of which
only 34 percent have been legally incorporated, and explore how the new venture’s founding
team affect its subsequent financing outcome. We explore the interplay between financial and
human resources by examining how the founding team size affects the probability of self-
financing. Our finding will extend our knowledge of how new venture team affects early-stage
high-tech ventures’ financing activities and provide insights on strategies that new venture teams
could apply.
Literature and Hypotheses Development
Financial resources of nascent ventures
Entrepreneurial firms face more severe agency problems and information asymmetry
problems than publicly listed companies, influencing their ability to attract funding (e.g. (Scherr,
Sugrue, and Ward 1993)). As a result, financing them requires “separate contractual solutions”
from the traditional ones designed for public companies (Mustapha and Tlaty 2018). Generally,
entrepreneurial firms can attract capital from angel investors, venture capital funds, and
corporate investors (Denis 2004), although newer sources of capital include crowdfunding
platforms, university-based funds, accelerators (incubators), and family offices (Bellavitis et al.
2017; Block et al. 2018).
However, external financial resources may not be available, especially for nascent
ventures with more significant information asymmetry and agency problems. Most startups fund
themselves in the initial years of their businesses (Cole 2009; Oranburg 2016), possibly through
29
external bank debt (Robb and Robinson 2014), or in combination with so-called "friends and
family" money, and business alliances (Markova and Petkovska-Mircevska 2009).
In addition to offering much-needed capital, self-financing operates as a signaling
mechanism, conveying information and therefore lowering costs for external investors, in part by
showing confidence in the business from current stakeholders. The entrepreneur's capital
commitment has been linked to a higher probability of leveraging external financing
(Huyghebaert and Van De Gucht 2007; Minola and Giorgino 2008). Signal through capital
commitments from ventures with multiple founders is more effective than ventures with a single
founder (Giga 2019).
We can then posit that a larger team is more likely to finance its own venture because this
provides much-needed resources, as well as conferring stronger signaling benefits. In other
words:
H1. Team size positively affects the probability of founding team self-financing.
Human Capital
Human capital is defined as the stock of knowledge and skills belonging to individuals
(Becker 1964) and is linked to both the tendency toward entrepreneurship and venture outcomes.
Formal education and industry experience are categorized respectively as general and specific
human capital (Baptista, Karaöz, and Mendonça 2014; Ganotakis 2012). Formal education and
previous startup experience robustly predict nascent entrepreneurs (Davidsson and Honig 2003).
For entrepreneurs motivated by opportunities rather than unemployment, both general and
specific human capital skills predict a higher probability of early survival (Baptista et al. 2014).
In addition, demographic characteristics (i.e., gender, race, and others) may affect a venture’s
survival and growth by impacting how the entrepreneur developed the relevant knowledge and
30
skills (Cooper, Gimeno-Gascon, and Woo 1994). More importantly, specific human capital has
been found to be more critical for the performance of nascent high-tech firms (Ganotakis 2012).
New ventures, especially the fast-growing ones, are more commonly founded by teams
rather than single entrepreneurs (Cooney 2005; Klotz et al. 2014). The founding team can affect
firm behavior (Beckman 2006). The structure of the founding team is an important strategic
decision to make during the earliest stage of a firm (Jin et al. 2017). A larger team may affect
firm performance positively through an advantage in processing information (Haleblian and
Finkelstein 1993). It is reasonable to presume that a larger team has more human capital
conditional on entrepreneurs’ characteristics such as experience and education.
The human capital of a nascent venture is vital in attracting external investors; its
importance relative to the business model is often expressed as a conundrum in “backing the
horse or the jockey”. It is generally accepted that entrepreneurs' characteristics have a great
impact on external investors’ decision-making (Harrison and Mason 2017; Kaplan, Sensoy, and
StrÖmberg 2009). Furthermore, entrepreneurs who see themselves as limited in their abilities
prefer self-financing (Carter and Auken 2005). In other words, ventures with more human capital
may have a lower need to self-finance.
Therefore, human capital has an important role in venture launch and success, and larger
teams offer greater human resources, potentially affecting the preference for self-financing. In
other words, the founding team’s human resources can substitute for financial ones. This stream
of literature leads to the second hypothesis, operating counter to our first one:
H2. Team size negatively affects the probability of founding team self-financing.
Data and Summary Statistics
Sample
31
Our sample is drawn from entrepreneurs participating in the National Science Foundation
(NSF) Innovation Corps (“I-Corps”) program. The entrepreneurs are self-identified, they were in
the regional I-Corps program at the University of Southern California (USC) and took
Accelerating Commercialization of Collegiate Engineering and Science Survey (ACCESS)
study. The NSF I-Corps program aims to “foster entrepreneurship and accelerate the translation
of knowledge derived from research into emerging products and services”, the ultimate goal of
which is to attract subsequent third-party funding (NSF 2020). As that study introduced, the
program was advertised to undergraduates, graduate students, and postdoctoral fellows through
university channels. To be eligible, a team must be fully formed, though a legalized firm is not
required, and a team must have a “deep technology” (Belz and Zapatero 2019). The eligible
teams are invited to enroll in a free two-week course on identifying marketplace needs (Denoo,
Van Boxstael, and Belz 2019).
The ACCESS contains questions on the nascent ventures’ startup progress, financing
status, business model development, technology maturity, human capital, and characteristics of
entrepreneurs. There are two rounds of the survey. The first survey is administered to all
participants of the program from December 2014 through November 2017 by allocating time
during the second class of the training. In the first class, the definition of terms used in the survey
was defined in order to ensure that every participant has the same level of knowledge on those
terms. The response rate is close to 100% (404 surveys). After removing observations with
incomplete or inaccurate measurements (such as those where the percentage of funds raised
exceeds 100%), 254 observations remained, including multiple responses from the same venture.
In the case that multiple recipients are in the same team, we end up identifying lead founders and
using only answers from them as they have the greatest impact on the venture (de Jong, Song,
32
and Song 2013). The second round of the survey was taken a year after each team’s first survey.
Among all 208 high-tech ventures who have completed answers in the first round, 201 of them
answered the second round, which represent the final study sample (Table 2.1).
We follow the analysis framework of Denoo et al. (2019). Though the sample and most
of the variables we use here are the same as those in Denoo et al. (2019), we have our advantage
in data. The difference is on the number of observations. Since we conduct this study later, more
second-round surveys got responded and thus resulted in a larger sample size.
Variables
Variable descriptions are summarized in Table 2.3.
Dependent variable:
Self-financing: A binary variable that takes a value of 1 if the venture is funded by
founder or cofounders, and takes a value of 0 otherwise. In our sample, 35% of ventures in our
sample are self-financed.
Independent variable:
The team size is specified in two ways: 1) As a continuous variable counting team
members; and 2) As a binary variable defining a large team with a threshold. Details follow:
Team size: Counts the number of people who founded the venture together. The average
team size in our sample is 2, with a maximum of 6 people.
Large team: We define a team as "large" if the number of founders is equal or greater
than 3, with 39% of ventures categorized this way. The threshold is varied for our robustness
checks.
Control variables:
The venture's characteristics used as control variables include:
33
Funded by outsiders: Founders whose ventures are financed through other methods may
have less incentive to invest their own money. In our sample, 28% of ventures have grants
financing, 6% from family and friends, 2% have angel investment, 3% have venture capital, and
1% have bank financing, methods recognized as sources of capital (Denis 2004; Islam, Fremeth,
and Marcus 2018; Markova and Petkovska-Mircevska 2009). Because these funding sources are
not mutually exclusive, 56% of all ventures were funded by at least one of these sources. This
variable is a dummy set to 1 if the venture is funded by investors other than the founding team.
Technology maturity: The extent to which the venture's technology has advanced, based
on the work of Denoo et al. (2019) using this as a proxy for venture emergence (Denoo et al.
2019). The ACCESS study includes a question asking the entrepreneur to identify the
technology's readiness at the time of the survey (see the Appendix 2-A). The scale is derived
from the nine-level "Technology Readiness Level”, widely used in the aerospace and defense
industry (Mankins 1995, 2009). Young, science-based firms typically do not operate at the
highest technology readiness stages (De Silva, Howells, and Meyer 2018) and the technology
readiness may affect whether or not a venture can move through the process of screening (Brush,
Edelman, and Manolova 2012). Small businesses in the aerospace industry may advance in
technology maturity as a result of government funding (Belz et al. 2019).
Business model: The extent to which all the firm’s business model has advanced. The
survey includes 21 questions describing business model development as a measure of
commercial maturity (Appendix 2-B), each of which is evaluated on a scale of 1 to 7, where a
higher number indicates that this aspect of business model is more developed. The measurement
of business model is an average value of the answers to the 21 questions. Business model
34
development is believed to have an influence on ventures’ technology maturity (Denoo et al.
2019; Helslop, McGregor, and Griffith 2001). The average score of this in our sample is 4.
Legal form: Incorporation has been shown to positively affect a venture’s bank financing
(Cassar 2004) and is a pre-requisite for funding. This dummy variable takes a value of 1 if the
venture has been incorporated, which is true for 34% of the ventures in our sample.
Patent: Patents can serve as signals for startup financing (Conti, Thursby, and Thursby
2013), and a patent is associated with substantial firm annual growth (Helmers, Rogers, and
Building 2009). This is a binary variable taking the value of 1 if the firm has been issued patents;
14% of ventures have patents.
Venture age: We control for the elapsed time since venture launch by subtracting the
venture's founding date from that of the first-round survey, because of evidence suggesting that
age benefits firm performance (Rossi 2016). The average age for ventures in our sample is 32
months.
Industry: Because some investors have been known to focus on one or two industries
(Morrissette 2007), we control for the industry. Each team reported if its company was in
hardware (41%), software (21%), or life sciences (13%). The rest are categorized as "other"
industries.
Number of advisors: Each respondent reports the number of people serving as the
venture's advisors. Because several studies indicate the importance of mentors in venture
progress (Denoo et al. 2019; Klotz et al. 2014; Ozgen and Baron 2007), mentoring is prevalent
and increasingly encouraged by organizations such as incubators, accelerators and public policy
program (NSF 2020; Ozgen and Baron 2007). The average number of advisors is 3 in our
sample.
35
The entrepreneur's characteristics used as control variables include
11
:
Industry experience: The number of years that the founder has worked in the new
venture's industry (Dimov 2010). The average industry experience for founders in our sample is
five years.
Research experience: The number of years that the founder has conducted research in the
new venture's technology fields. Following Denoo et al. (2019), we regard this as an appropriate
measure for the founder’s technical work experience, shown to determine venture growth
(Colombo and Grilli 2005). The average research experience for founders in our sample is seven
years.
Entrepreneurial experience: The number of years that the founder has worked in startups
prior to the current one. Research has shown that startup veterans exhibit a positive relationship
between entrepreneurial experience and performance (Toft-Kehler, Wennberg, and Kim 2014).
The average entrepreneurial experience for founders in our sample is three years.
Education: The founder’s highest level of education is measured as an ordinal variable
with seven categories ranging from 1= “High school diploma” to 10= “Law, MD, PhD or other
doctorate degree”. About half of the founders in our sample have the highest level. The average
score is 6. Papers have suggested that the founders' education level helps startups using
bootstrapping finance to achieve revenue growth (Ye 2017).
Age: The founder’s age at the time of the survey. The entrepreneur’s gender and age
serve as commonly used controls while investigating the impact on venture performance (Roper
1998). The average age in our sample is 35.

11
We use characteristics of respondents. When there are multiple respondents, we take the highest value among all
respondents. As long as one respondent is female, gender takes value of 1.
36
Gender: A dummy variable indicates whether the entrepreneur is female (1) or male (0).
24% of the founders in our sample are female. The sales and income of women-owned
businesses lag behind their male-owned counterparts, possibly because of a lack of resources
(Carter, Williams, and Reynolds 1997).
Empirical Analysis
We use a logistic (logit) model to estimate the probability of self-financing in Table 2.5.
Model 1 is the baseline, which indicates that the legal form, funding from outsiders, business
model development level, founder’s experience, and gender are linked to the probability that the
team funds itself. Model 2 suggests that the team size has a significant negative impact on the
probability of self-financing, with an average marginal effect (AME) of -9.4% (standard
error=0.022, p-value of 0.000) Model 3 is related but uses a binary variable, large team (defined
as a team of > 2 members), as the predictor; this model is created to align with our method of
addressing the selection concerns and gives consistent results with an AME of 19.1% (standard
error=0.059, p-value of 0.001).
Endogeneity – that the selection process is correlated with the measured outcomes – is a
concern. One source of endogeneity is reverse causality, in which the dependent variable affects
the independent variable and the measured coefficient does not represent a causal relationship. In
our research setting, it is unlikely that the team coalesces after investment, as the initial
ownership allocations might impact the decision to invest. A second type of endogeneity results
from confounding, in which both the dependent variable and the independent variable are
affected by covariates such that the independent variable or treatment is no longer randomly
assigned and the coefficient does not represent a causal effect. In other words, characteristics
determining the individual's propensity toward entrepreneurship also affect a self-financing
37
decision, or the sample is subject to “self-selection”. This is the endogeneity concern that we
want to address here. Education, experience, and gender have been linked to entrepreneurship
generation (Poschke 2013; Rocha, Carneiro, and Amorim Varum 2015; Scherer, Brodzinski, and
Wiebe 1990). Furthermore, opportunity recognition, which can be represented by the venture’s
characteristics, may also affect individual’s occupational choice (Dyer, Gregersen, and
Christensen 2008; Nicolaou et al. 2009). Thus, addressing potential confounding or self-selection
is necessary for the study.
Self-selection can be addressed in various ways; for instance, matching methodologies
are popular ways of assigning untreated samples into a control group (weight of 1) or discarded
entirely (weight of 0), based on absolute distance, a propensity score, or other measures of
similarity between untreated and treated samples. In the context of eastern Germany, propensity
score matching helps to show that public innovation subsidies stimulate firm R&D activities by
4%(Almus and Czarnitzki 2003). As to venture financing, GMM estimators have helped to
separate the effects from “treatment” and from “selection” and show that venture financing
stimulate the growth of new technology-based firms’ employment, sales, and productivity
(Bertoni, Colombo, and Grilli 2011; Croce, Martí, and Murtinu 2013). Alternatively, other
methods analyze covariate moments to assign weights at the observation level. Entropy
balancing (Hainmueller 2012) is such a method, assigning a real number as a weight to each
observation, allowing the researcher to keep data without imposing rigid constraints, calipers, or
an analytical form for the treatment model. In addition, it is particularly robust because even if
the treatment or outcome model is misspecified, the treatment effect result is still unbiased.
We thus apply entropy balancing on our study to estimate the causal impact of team size
on the probability of self-financing. We use the control variables of the logit model as the
38
covariates. We use an entropy balancing algorithm embedded directly in statistical software to
assign weights for the untreated observations in order to have balanced first and second moments
of covariates in treatment and control groups. Table 2.6 indicates the results of this algorithm and
shows that the standard differences are significantly reduced for all the covariates by this entropy
balancing process.
The weights derived from the entropy balancing are used to evaluate a simple logit model
(Table 2.7). The estimated AME is -18% (standard error=0.088, p-value of 0.046), consistent
with the initial estimate (model 3 of Table 2.5). This suggests that there is a slight confounding
issue in the relationship between team size and probability of self-financing. In other words, we
estimate that a large team is 15-20% less likely to self-finance.
Robustness Checks
In the original model (Table 2.5), we use all observations; however, among the 201
candidate ventures, ten (5%) received funding from angel investors, venture capitalists, or banks
at the time of the survey. These ten may be more mature than the rest and thus could be outliers
driving the estimation. Therefore, we apply a logit model on the sample, excluding these ten
ventures as a robustness check (Table 2.8) and find an AME of 10% (standard error=0.023, p-
value of 0.000), in agreement with our previous estimate.
Furthermore, six ventures did not identify a founder. This is most likely to happen in the
case that team has not yet determined the roles of each participant. We did not exclude these data
points because this may reflect the venture's early stage. To ensure that these observations do not
drive the result, we conduct a robustness check by excluding them (Table 2.9) and find effects
consistent with our earlier results. The AME excluding the six ventures is 11% (standard
error=0.024, p-value of 0.000).
39
Most importantly, we vary the threshold defining a “large team” with models that adjust
the minimum team size, creating four models indicating two, three, four, or five founders (Table
2.10)
12
. We find that the team size consistently shows a significant negative effect on the
probability of self-financing, with the AME increasing monotonically from -16% (standard
error=0.075, p-value of 0.033) for a team comprising two or more members, to -29% for a team
comprising five or more members (standard error=0.061, p-value of 0.000).
Discussion
In this paper, we determine that the team size negatively affects a founder's decision to
self-finance. In contrast to the idea that a larger team contributes to provide resources and confer
signaling benefits, we find that larger teams instead are less likely to finance. Each additional
person reduces the probability that the team finances itself by roughly 10%, demonstrated in a
model addressing potential selection effects, team threshold sizes, and model specifications.
In contrast to the idea that a team finances the company for both resources and signaling
benefits (H1), we find that human resources can substitute for financial ones (H2). This idea has
previously been explored (Chandler and Hanks 1998) but not in the context of a fuller team.
This has important implications for many reasons and contributes to scholarship in
strategy, organizational behavior, and entrepreneurial finance. In short, we can ask the question:
Does the company need the funding?
We find that larger teams, the business model becomes a predictor of increasing
importance. This finding thus contributes to scholarship on strategy because venture financing is
the immediate objective of nascent firms. Entrepreneurship scholarship typically focuses on
more substantive outcomes, such as raising venture capital, hiring, and generating ventures;

12
There are 16 ventures with five founders or more. 33 ventures with four founders or more.
40
however, in the earliest stages, financing represents not the strategic journey but the destination
itself. It is, therefore, counter-intuitive for a large team not to finance itself.
One reason that a larger team may not finance its own venture is that internal conflict
may inhibit consensus on strategy. Conflict may represent disagreements among team members
on the way to accomplish objectives (cognitive conflict), or those due to interpersonal
differences (affective conflict) (Jehn 1997), the former of which positively affects new venture’s
profit, sales, and growth, but the latter of which is negatively associated with these three aspects
(Ensley and Pearce 2001; Klotz et al. 2014; Mathieu et al. 2008). In this context, cognitive
conflict could prevent the team from converging on a strategy, thus inhibiting team members
from contributing sorely needed capital to the venture.
On the other hand, if team members do agree on the strategy and pursue it in a unified
fashion, then perhaps both general (formal education) and specific (industry experience) do
contribute to the venture's resources (Becker 1964).
Limitations
Several factors affect our study. These ventures are nascent and more than half have yet
to incorporate, which may affect the readiness of the venture to accept investment. It may be that
the venture has yet to reach an inflection point at which the team contributes meaningfully for
both resources and signaling to external investors (Giga 2019).
As a related concern, we do not have information on venture outcomes in a longitudinal
study. Financing is an intermediate step to deeper impacts, such as revenues, job creation, and
other stronger economic impacts.
The entrepreneur’s personal wealth can affect the decision to finance one's own venture
(Kerr and Nanda 2009). If an entrepreneur is financially constrained, he/she can contribute very
41
little despite knowing that committing financial resources is an optimal decision. Our survey did
not collect information on personal wealth.
It is also important to understand if the initial resources come from a single team member
or are broadly distributed. For instance, a single member may choose to invest early and exploit
the tax benefits of the 83(b) election associated with founders' equity. On the other hand,
contributions from multiple team members could reflect the same cross-validation and signaling
role that self-financing plays for external investors.
Conclusion
In this paper, we explore a novel sample of nascent high-tech ventures in which only 34%
have incorporated. They commonly self-finance, but we find that each additional team member
reduces the probability that the team finances itself by roughly 10%. Our model addresses
potential selection effects, team threshold sizes, and model specifications. According to the
resource-based view of the firm, in the earliest stages, funding is the pivotal resource to stimulate
the venture’s competitive advantage, but this finding suggests that the founding team’s human
capital can be a substitute to its financial commitment.
From a policy perspective, it becomes important to determine how to provide both human
and financial resources. In the strategy domain, this has important implications for early
decision-making since the lack of funding may have subsequent consequences.

42
Table 2.1. Data Processing

Number of respondents 404
Incomplete responses -103
Inconsistent measurements -43
Number of responses removed for having missing values -11
Final number of respondents 247
Final number of ventures 201

43
Table 2.2. Distribution of Funding Sources (Not Mutually Exclusive)

Funding source Number Percentage
Team 71 35.32
Grants 56 27.86
Family & friends 12 5.97
VC 6 2.99
Angels 4 1.99
Bank 2 1.00
None 88 43.78

44
Table 2.3. Descriptive Statistics
Ventures (N = 201)
Variables Mean Std.
Dev.
Min Max
Self-financing 0.35 0.48 0 1
Team size 2.42 1.24 0 6
Large team 0.39 0.49 0 1
Funded by
Outsiders
0.30 0.46 0 1
Technology
maturity
3.06 1.56 1 6
Business model 4.39 1.30 1 7
Legal form 0.34 0.47 0 1
Patent 0.14 0.35 0 1
Venture age
(months)
32.03 44.02 0 329
Industry—
Hardware
0.41 0.49 0 1
Industry—Software 0.21 0.41 0 1
Industry—Life
Science
0.13 0.34 0 1
Number of advisors 2.94 7.51 0 100
Industry experience 5.17 7.70 0 35
Research
experience
6.57 7.19 0 35
Entrepreneurial
experience
3.45 5.81 0 35
Education 6.02 1.33 1 7
Age 35.26 11.57 19 72
Gender (female) 0.24 0.43 0 1

45
Table 2.4. Correlation Matrix
1 2 3 4 5 6 7
1 Funded by
Team
1.00
2 Team size -0.26* 1.00
3 Funded by
Outsiders
-0.04 -0.01 1.00
4 Venture age 0.16* -0.01 0.10 1.00
5 Legal form 0.40* -0.02 0.19* 0.20* 1.00
6 Patent 0.17* 0.09 0.16* 0.26* 0.25* 1.00
7 Industry—
Hardware
-0.01 0.00 0.02 0.08 -0.00 0.06 1.00
8 Industry—
Software
0.00 -0.03 -0.10 -0.11 -0.03 -0.04 -0.43*
9 Industry—Life
Science
-0.07 0.07 0.07 0.02 0.10 -0.03 -0.04
10 Technology
Maturity
0.27* 0.01 0.15* 0.07 0.38* 0.16* -0.07
11 Business
Model
0.27* -0.08 0.18* 0.08 0.26* 0.09 0.01
12 Number of
advisors
-0.04 0.15* 0.05 -0.01 0.20* 0.04 -0.08
13 Industry
experience
0.09 -0.04 0.06 0.25* 0.15* 0.16* -0.09
14 Research
experience
0.09 0.04 -0.01 0.33* 0.11 0.17* -0.01
15 Entrepreneurial
experience
0.27* 0.04 0.09 0.27* 0.26* 0.24* -0.03
16 Education -0.06 0.06 -0.11 0.13 -0.09 0.09 0.07
17 Age 0.24* 0.00 0.02 0.43* 0.25* 0.24* 0.01
18 Gender -0.13 0.10 0.10 0.01 0.08 -0.00 -0.05

8 9 10 11 12 13 14
8 Industry—
Software
1.00
9 Industry—Life
Science
-0.20* 1.00
10 Technology
Maturity
0.09 -0.03 1.00
11 Business
Model
0.01 -0.06 0.24* 1.00
12 Number of
advisors
-0.03 0.02 0.11 0.11 1.00
46
13 Industry
experience
0.05 0.01 0.11 0.14* 0.00 1.00
14 Research
experience
0.03 0.00 0.13 0.01 0.03 0.46* 1.00
15 Entrepreneurial
experience
0.03 -0.05 0.05 0.17* 0.11 0.48* 0.26*
16 Education -0.02 0.04 -0.14 -0.03 -0.04 0.12 0.34*
17 Age -0.00 -0.02 0.09 0.14 -0.06 0.56* 0.62*
18 Gender -0.01 -0.05 -0.12 -0.07 0.16* -0.07 0.04

15 16 17 18
15 Entrepreneurial
experience
1.00
16 Education -0.08 1.00
17 Age 0.50* 0.35* 1.00
18 Gender 0.09 -0.04 -0.02 1.00

Note: Standard errors in parentheses. * p<0.05.

47
Table 2.5. Team Size and Self-financing—Logit Model
Probability of Self-financing
(1) (2) (3)
Team size −0.676∗∗∗
(0.181)

Large team −1.301∗∗
(0.428)
Funded by
Outsiders
-1.023*
(0.460)
-1.226*
(0.492)
-1.069*
(0.472)
Tech Maturity 0.197
(0.130)
0.275
(0.141)
0.226
(0.138)
Business Model 0.386*
(0.157)
0.356*
(0.161)
0.350*
(0.158)
Legal form 1.708***
(0.441)
1.713***
(0.470)
1.628***
(0.460)
Patent 0.402
(0.535)
0.738
(0.553)
0.589
(0.541)
Venture age 0.003
(0.005)
0.003
(0.005)
0.002
(0.005)
Industry—
hardware
-0.533
(0.471)
-0.600
(0.503)
-0.561
(0.494)
Industry—
software
-0.464
(0.528)
-0.531
(0.555)
-0.391
(0.540)
Industry—life
science
-1.043
(0.682)
-0.769
(0.704)
-1.008
(0.693)
Number of
advisors
-0.056
(0.047)
-0.035
(0.040)
-0.039
(0.043)
Industry exp -0.063*
(0.031)
-0.076*
(0.032)
-0.070*
(0.031)
Research exp 0.004
(0.033)
-0.009
(0.034)
-0.001
(0.034)
Entrepreneurial
exp
0.098*
(0.041)
0.116**
(0.043)
0.095*
(0.042)
Education -0.076
(0.153)
0.006
(0.164)
-0.025
(0.161)
Age 0.026
(0.024)
0.028
(0.025)
0.033
(0.025)
Gender -0.999*
(0.504)
-0.957
(0.532)
-0.986
(0.521)
Intercept −3.177∗
(1.256)
−2.307
(1.331)
−3.220∗
(1.306)
Observations 201 201 201
Pseudo 𝑅
!
0.272 0.334 0.310
Note: Standard errors in parentheses. * p<0.05, ** p<0.01, *** p<0.001
48
Table 2.6. Covariates Balance of Entropy Balancing

This table shows how standard differences of covariates are reduced after entropy balancing
methodology. Standard Difference (Pre) is the standard differences of covariates before entropy
balancing, Standard Difference (Post) is the standard differences of covariates after entropy
balancing.

Covariates Standard Difference (Pre) Standard Difference
(Post)
13

Funded by outsiders -0.031 -0.000
Technology Maturity -0.018 -0.000
Business Model -0.184 -0.000
Legal Form -0.203 -0.000
Patent 0.098 -0.000
Venture age 0.037 -0.000
Industry—hardware 0.033 0.000
Industry--software 0.035 0.000
Industry—life science -0.006 -0.000
Number of advisors 0.167 0.000
Industry experience -0.061 -0.000
Research experience 0.140 0.000
Entrepreneurial
experience
-0.059 -0.000
Education 0.284 0.000
Age 0.072 -0.000
Gender 0.047 -0.000

Note: 𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 =
"#$$%&%'(% #' *%+' ,-.(,*%/ 0%.1%%' 2&,-3/
4.+'5+&5 5%6#+.#,' ,$ ,-.(,*% +*,'2 3+&.#(#3+'./

13
keep three decimal digits
49
Table 2.7. Self-financing and Team Size—Results of Entropy Balancing

Probability of Self-
financing
Large team −0.869
∗

(0.425)
Intercept −0.486
(0.319)
Observations 201

Note: Standard errors in parentheses. * p<0.05, ** p<0.01, *** p<0.001
50
Table 2.8. Subsample Ventures—without Mature Outside Financing
Probability of Self-financing
Team Size -0.727***
(0.194)
Funded by Outsiders -1.058*
(0.519)
Tech Maturity 0.239
(0.147)
Business Model 0.410*
(0.171)
Legal form 1.765***
(0.488)
Patent 0.485
(0.590)
Venture age 0.007
(0.006)
Industry—hardware -0.475
(0.526)
Industry—software 0.026
(0.603)
Industry—life science -0.639
(0.739)
Number of advisors -0.021
(0.032)
Industry exp -0.082*
(0.035)
Research exp 0.005
(0.036)
Entrepreneurial exp 0.139**
(0.047)
Education 0.061
(0.168)
Age 0.013
(0.027)
Gender -1.268*
(0.590)
Intercept -2.554
(1.368)
Observations 191
Pseudo 𝑅
!
0.347

Note: Standard errors in parentheses. * p<0.05, ** p<0.01, *** p<0.001
51
Table 2.9. Subsample Ventures—Exclude Ventures with Zero Self-reported Founder
Probability of Self-financing
Team Size -0.788***
(0.204)
Funded by Outsiders -1.212*
(0.501)
Tech Maturity 0.258
(0.145)
Business Model 0.430*
(0.169)
Legal form 1.491**
(0.486)
Patent 0.896
(0.577)
Venture age 0.006
(0.006)
Industry—hardware -0.576
(0.512)
Industry—software -0.364
(0.564)
Industry—life science -0.638
(0.723)
Number of advisors -0.031
(0.038)
Industry exp -0.092**
(0.034)
Research exp 0.007
(0.035)
Entrepreneurial exp 0.112*
(0.044)
Education -0.006
(0.166)
Age 0.031
(0.026)
Gender -0.768
(0.548)
Intercept -2.499
(1.352)
Observations 195
Pseudo 𝑅
!
0.336

Note: Standard errors in parentheses. * p<0.05, ** p<0.01, *** p<0.001

52
Table 2.10. Different Thresholds of Large Team
Probability of Self-financing
(1) (2) (3) (4)
2 founders or
more
-0.999*
(0.458)

3 founders or
more
-1.301**
(0.428)

4 founders or
more
-2.357**
(0.760)

5 founders or
more
-2.765**
(1.028)
Funded by
Outsiders
-1.107*
(0.471)
-1.069*
(0.472)
-1.107*
(0.480)
-1.166*
(0.486)
Tech Maturity 0.242
(0.134)
0.226
(0.138)
0.236
(0.137)
0.219
(0.135)
Business Model 0.331*
(0.158)
0.350*
(0.158)
0.434**
(0.165)
0.445**
(0.164)
Legal form 1.719***
(0.445)
1.628***
(0.460)
1.745***
(0.461)
1.837***
(0.462)
Patent 0.559
(0.536)
0.589
(0.541)
0.619
(0.559)
0.590
(0.549)
Venture age 0.002
(0.005)
0.002
(0.005)
0.004
(0.005)
0.004
(0.005)
Industry—
hardware
-0.551
(0.475)
-0.561
(0.494)
-0.690
(0.498)
-0.447
(0.498)
Industry—
software
-0.493
(0.530)
-0.391
(0.540)
-0.642
(0.562)
-0.527
(0.558)
Industry—life
science
-1.024
(0.696)
-1.008
(0.693)
-0.917
(0.704)
-0.698
(0.696)
Number of
advisors
-0.050
(0.043)
-0.039
(0.043)
-0.027
(0.040)
-0.061
(0.050)
Industry exp -0.073*
(0.031)
-0.070*
(0.031)
-0.074*
(0.033)
-0.075*
(0.033)
Research exp 0.006
(0.033)
0.001
(0.034)
-0.000
(0.034)
-0.008
(0.035)
Entrepreneurial
exp
0.109**
(0.042)
0.095*
(0.042)
0.125**
(0.046)
0.116**
(0.045)
Education -0.024
(0.157)
-0.025
(0.161)
-0.030
(0.161)
-0.074
(0.160)
Age 0.020
(0.024)
0.033
(0.025)
0.026
(0.025)
0.031
(0.025)
Gender -1.001
(0.515)
-0.986
(0.521)
-0.930
(0.526)
-0.878
(0.521)
Intercept -2.400
(1.307)
-3.220*
(1.306)
-3.633**
(1.332)
-3.625**
(1.319)
53
Observations 201 201 201 201
Pseudo 𝑅
!
0.290 0.310 0.319 0.308

Note: Standard errors in parentheses. * p<0.05, ** p<0.01, *** p<0.001

54
Chapter 3: Do M&A Affect Patent Inventors’ Productivity?

Weiran Deng, Yongxiang Wang
14

Abstract
This paper studies whether and how mergers and acquisitions (M&A) affect the productivity of
patent inventors. Career paths of inventors are identified by their patent records. Completed and
withdrawn M&A biddings are exploited as a quasi-experiment. Results show that for both
acquirer inventors and target inventors, the number of patents will drop after experiencing M&A.
When the acquirer and target are in the same industry, the quality of patent also drops for both
acquirer inventors and target inventors. The results support the hypothesis of cultural conflict and
competition for funding among inventors. We did not see evidence supporting collaboration
among inventors or increased productivity due to increased pressure on job security concerns.

Keywords: Innovation, Patent, Mergers and Acquisition, Human Capital, Productivity
JEL Codes: G34, J24, J62

14
Marshall School of Business, University of Southern California
55
Introduction
Since 1996, the number of listed companies in the U.S. declined dramatically, from more
than eight thousand to about four thousand. Research has shown that half of this decline is due to
the increasing number of delists from acquisitions of publicly listed firms (Doidge, Stulz, & M.,
2017). Meanwhile, “acqui-hiring”, or talent acquisition, which refers to the process of acquiring
a company primarily to recruit “some or all of the target company’s at-will employees” (Coyle &
Polsky, 2013), became increasingly common. The popularity of this term is derived from the
phenomenon in the technology sector. Mark Zuckerberg, CEO of Facebook, once stated that
“Facebook has not once bought a company for the company itself. We buy companies to get
excellent people.” In the technology sector, the most valuable and measurable asset that can be
produced by individuals are patents. All the above phenomena that have emerged recently lead
us to think: will patent inventors’ productivity be affected by the mergers and acquisitions they
experienced with the companies they work for?
Mergers and Acquisitions (M&A) can be incentivized for various reasons, and thus
generate very different outcomes and impacts. Generally speaking, M&A can be induced by both
efficient and inefficient purposes. There are two leading theories stating the efficient aspect of
M&A: one is disciplinary theory and the other is synergistic theory. The disciplinary theory
states that buyers are incentivized for potential gains through disciplining targets that are with
good assets but poor performance; synergistic theory suggests that the increasement of the
target’s performance is achieved through forming complementarities between the buyer’s
business and the target’s business (Ogada, Achoki, & Njuguna, 2016; Yeboah, Asirifi, &
Ampadu, 2015). On the other hand, theories of M&A stating the inefficient aspect include
empire-building (Baumol, 1967) (Mueller, 1969) (Mueller, 1993), management-entrenchment
56
(Shleifer & Vishny, 1989), and managers’ overconfidence (Roll, 1986). A widely accepted
opinion is that theories of M&A are not mutually exclusive. A firm could have both efficient and
inefficient incentives for a merger or an acquisition.
As a result, the impact of M&A is mixed. M&A activities can either lead to investment
distortion (Rajan, Servaes, & Zingales, 2000), or more efficient resources allocation through
exploiting winner-picking (Guedj & Scharfstein, 2004), and better resilience to adverse shocks
(Gopalan & Xie, 2011). However, people know very little about the effects of M&A on other
levels of observations except for firm-level impacts. Specifically, and perhaps most importantly,
the individual level impacts. In this paper, we will investigate the impact of M&A on inventors’
productivity to see how such an important group of people is affected by M&A activity and
through which channel.
Microeconomic research on corporate issues have long focused on capitals; for example,
investment decision, pricing strategy, and stock market performance. Recently, however, there
are increased efforts among scholars to shed light on corporate finance from labor economics,
which requires a deeper understanding of human capital in firms. Therefore, instead of
restraining research to firm-level analyses of mergers and acquisitions, going into individual-
level investigation is even more inspiring, and allows us to have a better understanding of
individuals as the ultimate decision-makers.
Inventors are not only workers, but also generators of innovation and engines of the
economy. Given that innovation is at the center of modern theories of economic growth (Solow,
1957) (Aghion & Howitt, 1994), it is crucial to know how firm activities will affect inventor’s
behaviors and incentives.
57
However, estimating the effects of M&A is challenging due to an embedded selection
bias associated with the decision to bid on M&A. As mentioned above, M&A could be
incentivized for mixed reasons. One scenario is that those firms who choose to bid on M&A are
exactly the ones that expect synergies in human capital. The identification strategy of this paper,
therefore, aims at disentangling the treatment effect and the effect from the selection issue. Our
solution is to exploit a quasi-experiment.
Following Seru (2014), we argue that whether M&A deals are completed or withdrawn
can be treated as random. Our treatment group thus comprised of firms completing M&A and the
control group comprised of firms whose M&A bids were withdrawn. The underlying assumption
is that the reasons for the failure of M&A biddings are unrelated to interactions between human
capital in the acquirer and the target.
We use information from the Harvard Business School patent database to get a map of
patents from inventors and the dataset created by Kogan et al. (2017) to map patents to public
firms. Combined with M&A events published on the SDC Platinum and the CRSP/Compustat
files, we were able to get full financial information of all the public firms who have ever been
involved in M&A activities in the sample period. We measured inventors’ productivity by both
the quantity of patents and the quality of patents. Proxy of inventors’ characteristics includes
originality and generality of patents and their exploitative or exploratory nature.
Results show that for both acquirer inventors and target inventors, the number of patents
will drop after experiencing M&A. When the acquirer and target are in the same industry, the
quality of patent also drops for both acquirer inventors and target inventors. The results support
the hypothesis of cultural conflicts and competition for funding (insufficient funding). We did
58
not see evidence supporting collaboration among inventors or increased productivity due to
increased pressure of job security concerns.
The paper is related to several strands of literature. First, by employing an identification
strategy to isolate the treatment effect from M&A and focusing on innovation activities around
the M&A, this paper contributes to M&A literature that explores its impact. Second, this paper
contributes to the discussion on effects of firm activities on innovation. For example, Bernstein
(2015) investigates the effects of going public on innovation. Third, the paper also contributes to
a growing literature that explores the role of individuals in firms. Regardless of the individual-
level feature, this paper is close to Seru (2014), Phillips and Zhdanov (2012), and Sevilir and
Tian (2012), all of which investigate at firm-level. The first one examines the impact of
conglomerate on innovation activity performance (Seru, 2014), the second one shows how the
market of M&A affects firms’ innovation activities (Phillips & Zhdanov, 2012), and the last one
suggests a positive relationship between M&A activities and the firm-level subsequent
innovation performance (Sevilir & Tian, 2012). Considering the individual-level feature, the
most similar work to this study is Kapoor and Lim (2007) and Li and Wang (2020). The former
one focuses on semiconductor firms and thus has limited sample size. The latter one focuses on
M&A events announced from 1981 to 1998, and finds that target inventors perform better
compared to acquirer inventors post-mergers.
The rest of the paper proceeds as follows. we first develop the hypotheses, and then
describes the data. In the “empirical strategy” section, we explain the empirical challenges and
offer a solution. Empirical results are shown as follows. The next section discusses concerns and
next steps. The last section concludes.
Hypotheses Development
59
In this section, we discuss four potential channels that could cause the impacts of M&A
on patent inventors’ productivity.
Cultural conflicts
Every corporate has its own culture. Often, corporate culture develops over time from the
company’s history and experiences (Weber, 1996). It is proved that cultural differences affect
both the volume and the gains of cross-border mergers (Ahern, Daminelli, & Fracassi, 2015).
Corporate culture is even regarded as “the ultimate strategic assets” (Flamholtz & Randle, 2011).
Thus, if there are conflicts in culture between the acquirer and the target, we would
expect that the productivity of target inventors to decline after M&A. The reason is that it is the
target company that changes its ownership and therefore, has to adapt to new rules set by the
acquiring company.
Job security concerns
In order to create value from M&A, the first thing that the acquirer would do is to reduce
costs of running the target (Cohen, 2010). Among all the cost reduction strategies, firing
employees that the new company owner believes redundant is one of the most common ones
(Chatt, Gustafson, & Welker, 2017). As a result, target inventors will have the pressure of losing
jobs and be incentivized to work harder. If this is indeed the motivation, then the effect is more
likely to show through the quantity of patents rather than the quality of patents for target
inventors who stay in the company after M&A, as the quality requires longer time to tell.
Competition for funding after horizontal mergers
Horizontal mergers are mergers that happen between two companies in the same
industry. After such a merger, there will be more inventors competing for funding for one
direction of investment if the firm is financially constrained (Fazzari & Petersen, 1993). If
60
funding for the investment direction remains the previous level, we would expect both acquirer
inventors and target inventors to decrease their productivity. And we won’t observe these effects
following unrelated mergers.
Collaboration
After horizontal mergers, inventors can easily build up networks with a broader set of
inventors in the other company, through which coauthor relationships are formed. “Inventors
contribute more when they are better networked” (Liu, Mao, & Tian, 2017). The collaboration
would generate more efficient match between inventors and thus increase the productivity of
both acquirer inventors and target inventors. We will not observe these effects following
unrelated mergers.
Data
The data in this analysis is constructed from several sources combining information of
patent-inventor-firm combinations, mergers and acquisitions, firm performance and
fundamentals, and characteristics of patent.
Inventors and Patents
Analysis of the inventor-level data is complicated. The main reason is that inventor
names are often unreliable. Some names of inventors are abbreviated and a same name could be
used by several inventors. To overcome this drawback, we use the Harvard Business School
(HBS) patenting database. The database has unique identifiers that come from “refined
disambiguation algorithms”. It differentiates similar inventors according to various
characteristics, instead of simply using names and geographic locations (Lai, Amour, & Fleming,
2010). A patent application could be associated with multiple inventors. Following Bernstein
(2015) which uses the same database, we attribute the patent equally to each inventor. Inventors’
61
relocation is identified through the change of associated companies when patents are created.
Note that we only know the relocations of productive inventors and we assume they matter the
most.
Overall, this dataset covers the universe of over 1.4 million inventors and over 2.1
million U.S. utility patents granted from 1975 to 2010
15
. On average, one inventor has three
patents during the sample period, with the minimum number of patents as one and the maximum
number as 945. Each patent is produced by two inventors on average, with the minimum number
of co-inventors as one and the maximum number as 74.
Patent and Firm
We got the mapping of patents to CRSP and COMPUSTAT firms from Kogan et al.
(2017). The dataset is built on the standard NBER mapping and covers information about all
U.S. publicly traded firms.
Using Kogan et al. (2017) dataset, we are able to map over 948K patents and over 629K
inventors to 6,435 CRSP firms. On average, one inventor has four patents during the sample
period, with the minimum number of patents as one and the maximum number as 875. Each
patent is produced by two inventors on average, with the minimum number of co-inventors as
one and the maximum number as 74.
Note that HBS database covers inventors and patents associated with both public and
private firms. In some scenarios, there are even self-employed inventors. The mapping keeps
only inventors that were related to public firms while applying for patents. By comparing with
numbers in raw HBS data, we can tell that our study covers nearly half of all patents and
inventors from 1975 to 2010. The others were in private companies or self-employed.

15
Information during data processing is summarized in Table 3.1.
62
M&A data
Information on firms’ mergers and acquisitions comes from the SDC Platinum Database.
Between the sample period, January 1975 and December 2010, there are 2.5 million M&A deals
in record, containing both public firms and private firms. In addition to the identities of the
involved parties, the dataset provides information on whether the identities were acquirer or
target, which industry the firm was in; and whether the merger is horizontal or unrelated. The
merger type is defined as ‘horizontal’ if the acquirer’s industry description is the same as its
target’s industry description and defined ‘unrelated’ otherwise. More definitions about M&A are
summarized in the Appendix 3-A.
Using this dataset, we are able to map
16
over 237K patents and over 217K inventors to
1,536 CRSP firms who were acquirers in M&A events; over 143K patents and over 174K
inventors to 1,557 CRSP firms who were targets in M&A events
17
. Some inventors can
experience several M&A events in their lifetime. We only keep the inventors who experience
just once, either in acquirer firms, or in target firms, or each of the types once, in the final
sample.
Measuring Inventor’s Productivity and Other Characteristics
Inventors’ productivity is measured by the number of patents and citations per patent.
Number of patents is the most basic and intuitive measurement of innovative output. Citations
per patent represent the quality of patents. It distinguishes between “breakthrough innovation and
incremental discoveries” (e.g., (Griliches, 1990)). Patent citations have been shown to have

16
The identifier of firms in this dataset is CUSIP, which needs to be converted to match with identifiers in the other
datasets. And this converting process generates a material number of missing values. More about different identifiers
of firms are summarized in Appendix 3-B.
17
We exclude buyback mergers for the study. Buyback mergers refer to the case when the acquirer buys itself.
63
impacts on firm’s market value (Hall, Jaffe, & Trajtenberg, 2005) and firm’s productivity
(Kogan, Papanikolaou, Seru, & Stoffman, 2017).
As to other characteristics of inventors, we consider the average “originality” and
“generality” of the patents that each inventor created. “Originality” and “generality” use the
distribution of forward and backward citations of patents to capture the fundamental nature
(Trajtenberg, Henderson, & Jaffe, 1997) (Bernstein, 2015). Patents that cite a broader array of
technology classes and are being cited by a broader array of technology classes are viewed as
having greater originality and generality respectively. At the inventor-level, “originality” and
“generality” are then the average over patents created by each inventor. Plus, patents’
exploratory and exploitative nature is also considered. Exploratory patents are “radical
innovations” and exploitative patents are “incremental innovations” (Jansen, Bosch, & Volberda,
2006). For each inventor, we calculate the fraction of exploratory (exploitative) patents. More
detailed definitions are shown in the Appendix 3-C.
We use the number of patents and citations per patent as our dependent variables. Other
characteristics are used as control variables in the analysis. Data of all the above measures are
mainly calculated from the database generated by Kogan et al. (2017). For patents’ exploratory
and exploitative property, we collect extra information from Google Patent.
Due to the nature of innovation, patents are not created continuously. More frequently
than not, an inventor is very productive during some specific periods but not in other years. To
overcome this issue, we restrict the analysis to inventors that produce at least a single patent both
before and after the M&A bidding and examine the patenting behavior of inventors in the three
years before and five years after the M&A bidding. As a result, the time frame of our M&A
events is between 1980 and 2005.
64
Firm Characteristics and Financial Information
Firms’ annual accounting data are obtained from the CRSP/COMPUSTAT merged
database. We calculate sales, market equity, R&D intensity, Tobin’s Q, the growth rate of total
assets, return on assets, and Herfindahl index from the data. Detailed definitions are in the
Appendix 3-D. We drop observations whose values are missing for these variables.
Empirical Strategy
Empirical Complication and Design
Identifying the effects of M&A on inventors’ productivity is challenging due to the
selection issue. Firms sometimes do acquisition for human capital in the target firms. The
productivity of inventors in M&A firms might differ from those in non-M&A firms due to some
fundamental characteristics. Under these scenarios, a simple comparison between M&A
inventors and non-M&A inventors in the data captures not only the treatment effect but also the
selection bias.
To remove the selection bias, we need to randomly assign inventors with similar
productivity into M&A and non-M&A firms. We explore a feature of M&A that was not used on
this research topic when we started this study
18
: regarding M&A completion versus M&A
withdrawn as a quasi-experiment. Therefore, we exclude the potential selection through making
the decision to do M&A. The decision to withdraw is exogenous under our assumption.
In sum, we construct a group of firms completing their mergers as “treatment group” and
a group of firms whose mergers failed to go through as “control group”. The two groups together

18
Found out recently that Li and Wang (2020) applies the same empirical strategy. However, the sample period in
that paper is 1981-1998, which is shorter than ours.
65
comprise the sample where the assignment of a firm into completed M&A is random. And
therefore the assignment of inventors into treatment and control group is random.
Methodology
The final sample for analysis comprised of individual inventors who were work for
public companies that have either completed or withdrawn M&As. We keep only those who
have only experience such an M&A once to exclude cases with accumulative effect. As
mentioned in the data section, another filter we use is that inventors must have patents assigned
to the same firm both before and after M&A. By doing this, we avoid the complication of
analyzing inventor’s mobility due to M&A. We focus our analysis on the performance of the
most productive inventors who stay with their firms
19
.
Our outcome variables are the number of patents and citations per patent in five years
after the M&A event. Our independent variable is a dummy variable indicating whether a firm
has completed M&A. Control variables include all firm characteristics in the M&A year, all
inventor characteristics based on patents created in three years before M&A, number of patents
or citations per patent in three years before M&A, plus industry where the firm belongs to and
the year when the M&A takes place. Because our dependent variable is non-negative and skewed
count data, we employ Poisson regression which is regarded to be the most robust model to deal
with this type of dependent variables (Harris, Yang, & Hardin, 2012).
Results

19
Note that target inventors can have patents assigned to the same firm both before and after M&A event if the
target firm is run as a subsidiary after acquisition.
66
To test hypotheses, we estimate the effects among acquirer inventors and target inventors
separately. For each of the groups, we estimate a general effect first, then estimate the impact
only among a subsample of horizontal mergers.
Table 3.3 shows the effects of M&A on acquirer inventors. Columns 1 and 3 are
estimation results on the number of patents and citations per patent without controlling firm and
inventor characteristics. Columns 2 and 4 show the results after controlling firm and inventor
characteristics. Across both specifications, the number of patents and citations per patent
significantly decline after M&A completed. The number of patents drop about 9.4%, citations
per patent drop 1.4%. The quantity of patents got stronger negative effects than the quality of
patents.
Table 3.4 shows the effects of M&A on acquirer inventors in a subsample which only
includes horizontal mergers. Columns 1 and 3 are estimation results on the number of patents
and citations per patent without controlling firm and inventor characteristics. Columns 2 and 4
show the results after controlling firm and inventor characteristics. Across both specifications,
the number of patents and citations per patent significantly decline after M&A completed. The
number of patents drop about 17.4%, citations per patent drop 9.4%. Again, the quantity of
patents got stronger negative effects than the quality of patents. Besides, the magnitudes are
larger than the whole acquirer inventor sample. When the acquirer and the target are in the same
industry, inventors’ productivity got more severe negative impacts.
Table 3.5 shows the effects of M&A on target inventors. Columns 1 and 3 are estimation
results on the number of patents and citations per patent without controlling firm and inventor
characteristics. Columns 2 and 4 show the results after controlling firm and inventor
characteristics. Results show that the number of patents drop about 12.3%, yet citations per
67
patent increase 3.8%. The results implicate that M&A has negative impact on the quantity of
patents produced by target inventors yet has a positive impact on the quality of patents produced
by target inventors.
Table 3.6 shows the effects of M&A on acquirer inventors in a subsample which only
includes horizontal mergers. Columns 1 and 3 are estimation results on the number of patents
and citations per patent without controlling firm and inventor characteristics. Columns 2 and 4
show the results after controlling firm and inventor characteristics. Across both specifications,
the number of patents and citations per patent significantly decline after M&A completed. The
number of patents drop about 14.9%, citations per patent drop 5.4%. When the acquirer and the
target are in the same industry, inventors’ productivity got more severe negative impacts. The
positive effect on quality from M&A on the full target inventors is driven by inventors involved
in horizontal mergers.
In sum, our empirical findings suggest that for both acquirer inventors and target
inventors, the number of patents will drop after experiencing M&A. When the acquirer and
target are in the same industry, the quality of patents also drops for both acquirer inventors and
target inventors.
Since the quantity of patents by target inventors does not increase after M&A, we
exclude the hypothesis of motivation from job security concerns. The mechanism cannot be a
collaboration between acquirer inventors and target inventors neither, which predicts
productivity of inventors from both sides to increase. Our result supports the hypotheses of
cultural conflict and competition for funding. The decline in productivity of both acquirer
inventors and target inventors, especially after horizontal mergers, is consistent with the
prediction from these two mechanisms.
68
The result reveals that the transition caused by firms’ M&A activities has important
implications for the human capital accumulation process. Acquirer firms shall be aware of the
potential decrease in inventor productivity due to cultural conflicts and insufficient funding for
inventors in the same industry. From an individual inventor’s perspective, inventors should have
the expectation on how a new environment after M&A will affect their own productivity and
make wise career decisions to avoid wastes of their human resources.
Discussion
Endogeneity
The empirical strategy of this study relies on the assumption that M&A completed or
withdrawn is randomly assigned. One concern is that the firms that choose to complete M&A
transactions are the ones that expect inventors to have higher productivity. If that is the case, the
empirical design does not solve the endogeneity issue, and the estimation will be biased.
However, the possible bias is toward having positive impacts on inventors’ productivity,
but our results show negative effects on both acquirer and target inventors. Thus, we conclude
that the results we show here is the lower bound of possible negative effects from M&A to
inventors’ productivity.
Inventor mobility
If inventors expected that the new environment after M&A will negatively affect their
productivity, inventors might choose to leave the firm before M&A happen. Thus, another
important and relevant research question is how M&A affect inventors’ mobility. In this study,
we only analyze inventors who have patents both before and after M&A, leaving inventors who
moved in or out the firm due to M&A out of consideration. How M&A affect inventors’ mobility
is a future research direction.
69
Mergers Involving Private Companies
All inventors working for private companies are neglected in this paper. Since the map
from patents to firms is from Kogan et al. database, we are only able to match inventors and
M&A events through the connection of identified U.S. public firms.
However, private companies are crucial for generalizing this research. In terms of M&A,
private companies are very commonly bought by other companies. Moreover, there are a huge
number of inventors working for private companies. In future research, we can do a name
matching directly between the HBS inventor-patent database and the SDC M&A database,
skipping the mapping of firm identification with patents.
Conclusion
This paper investigates the impacts of mergers and acquisitions on patent inventors’
productivity. The most distinguishing feature of the paper is that the analysis is at the individual
level, which is complicated and challenging due to the restrictions on data of inventors. We
combined the HBS inventor and patent database, the patent and firm database from Kogan et al. ,
SDC M&A database, and financial information of firms from CRSP/COMPUSTAT. The paper
contributes to fill the gap of knowledge for the impacts of M&A on individuals.
To deal with the selection bias, this paper exploits a quasi-experiment by constructing
completed M&A as the treatment group and withdrawn M&A as the control group. Results show
that for both acquirer inventors and target inventors, the number of patents will drop after
experiencing M&A. When the acquirer and target are in the same industry, the quality of the
patents also drops for both acquirer inventors and target inventors. The results support the
hypotheses of cultural conflict and competition for funding among inventors. We did not see
70
evidence supporting collaboration among inventors or increased productivity due to increased
pressure on job security concerns.
The results reveal that the transition caused by firms’ M&A activity had important
implications for the human capital accumulation process. Existing research investigating impacts
of M&A found positive effects on firm-level innovation measures. However, that could be the
result of hiring more inventors or buying more patents but not necessarily implicates the
increased individual-level productivity due to M&A.
Our findings have important policy implications. Acquirer firms shall be aware of the
potential decrease in inventor productivity due to cultural conflicts and insufficient funding for
inventors in the same industry. From the individual inventor’s perspective, inventors should have
the expectation on how the new environment after M&A will affect their productivity, and make
wise career decisions to avoid wastes of their human resources.

71
Table 3.1. Data Processing
HBS all inventors Public firm inventors
20

Observations 4,454,862 2,229,270
Distinctive patents 2,179,750 948,553
Distinctive inventors 1,462,619 629,473
Number of patents per inventor (1975
– 2010)
3.05 (6.97) 3.54 (7.98)
Number of inventors per patent 2.04 (1.47) 2.35 (1.64)
Distinctive companies N/A 6,435

M&A Inventors
21

If the firm was a
target
If the firm was an acquirer
Observations 580,724 328,012
Distinctive firms 1,536 1,557
Distinctive patents 237,604 143,928
Distinctive inventors 217,529 174,014
Number of patents per inventor (1975
– 2010)
2.66 (2.34) 1.88 (2.44)
Number of inventors per patent 2.44 (1.64) 2.28 (1.59)

Note: standard deviations are in parentheses.

20
Matched with (Kogan, Papanikolaou, Seru, & Stoffman, 2017)
21
Matched with SDC M&A database.
72
Table 3.2. Summary Statistics
This table provides key summary statistics for the sample used in the final analysis,
comparing individuals in firms that complete M&A with those in firms that withdraw M&A. The
sample covers patent inventors who experienced once and only once M&A bid from 1980 to
2005, and have patents both before and after the M&A bid year. All variables are defined in the
Appendix. Inventor’s productivity measures are calculated over the five years after the M&A
bidding year. Inventor’s characteristic measures are calculated over the three years up to (and
through) the M&A bidding year. Firms’ financial information and M&A characteristics are at the
time of the M&A bidding. *, **, and *** indicate that differences in means are statistically
significant at the 10%, 5%, and 1% level, respectively.

Panel A. Acquirer Inventors

Completed M&A. (n=37,424) Withdrawn M&A (n=2,780) P-value
Mean Median SD Mean Median SD Welch T-
test
Inventor’s Productivity Measures
Number of
Patents
2.786 2.000 4.027 2.603 1.000 4.364 0.032
Citations per
patent
10.525 4.800 18.246 13.872 8.000 18.913 0.000
Firm characteristics
Log sale 9.217 9.678 1.924 8.557 9.354 2.130 0.000
Log ME 8.994 10.051 1.468 8.200 8.265 1.612 0.000
Log RD 0.056 0.033 0.063 0.075 0.056 0.075 0.000
Log Q 0.739 0.669 0.597 0.354 0.191 0.504 0.000
INV 0.101 0.073 0.180 0.080 0.064 0.211 0.000
ROA 0.157 0.159 0.096 0.145 0.152 0.122 0.000
herf 0.352 0.343 0.216 0.330 0.238 0.227 0.000
Inventor characteristics
Originality 0.570 0.625 0.267 0.531 0.591 0.282 0.000
Generality 0.591 0.653 0.266 0.582 0.641 0.265 0.078
Frac of
exploratory
0.593 0.667 0.429 0.703 1.000 0.402 0.000
Frac of
exploitative
0.270 0 0.379 0.199 0 0.346 0.000

73
Panel B. Target Inventors
Completed M&A. (n=21,406) Withdrawn M&A (n=2,637) P-value
Mean Median SD Mean Median SD Welch T-
test
Inventor’s Productivity Measures
Number of
Patents
2.794 1.000 5.251 2.562 1.000 4.643 0.018
Citations per
patent
10.753 5.000 17.890 14.128 7.400 20.454 0.000
Company characteristics
Log sale 8.681 8.867 1.833 8.829 9.291 2.262 0.001
Log ME 8.628 9.001 1.523 8.480 8.690 1.603 0.000
Log RD 0.067 0.039 0.075 0.084 0.059 0.081 0.000
Log Q 0.666 0.492 0.620 0.349 0.259 0.442 0.000
INV 0.059 0.050 0.202 0.112 0.107 0.178 0.000
ROA 0.143 0.148 0.104 0.133 0.141 0.097 0.000
herf 0.340 0.261 0.249 0.356 0.374 0.202 0.000
Inventor characteristics
Originality 0.561 0.613 0.269 0.543 0.600 0.280 0.001
Generality 0.580 0.642 0.273 0.586 0.636 0.268 0.247
Frac of
exploratory
0.653 1.000 0.420 0.669 1.000 0.422 0.065
Frac of
exploitative
0.226 0.000 0.363 0.225 0.000 0.371 0.902

74
Table 3.3. Productivity of Acquirer Inventor
This table reports the results from the quasi-experiment on acquirer inventors (inventors who
were in acquirer companies while M&A was taking place). M&A completed is an indicator
variable that takes the value of one for the treatment group (completed mergers) and zero for the
control group (withdrawn mergers). Poisson regression is employed to examine the change in
both the number of patents and citations per patent of the inventors who are in the firm both
before and after the merger. Control variables include logsale, logME, logRD, logq, INV, ROA,
and herf as company characteristics and generality, originality, fraction of exploratory patent,
and fraction of exploitative patent as inventor characteristics. All company characteristics are
measured in the M&A bid year. All inventor characteristics are measured by the average of
patent measures over the three years before the merger. The unit of observation is at the
inventor-level. Merger dates in the sample correspond to the period 1980-2005. The robust
standard errors are presented in parentheses under each coefficient. *, **, and *** denote
significance at 10%, 5%, and 1%, respectively.

75

Number of Patents Citations per Patent
(1) (2) (3) (4)
M&A completed -0.103***
(0.014)
-0.094***
(0.014)
-0.025***
(0.006)
-0.014**
(0.006)
Log sale 0.029***
(0.006)
0.020***
(0.003)
Log ME -0.082***
(0.007)
-0.036***
(0.003)
Log RD -0.318***
(0.088)
0.636***
(0.044)
Log Q 0.140***
(0.012)
0.006
(0.007)
INV 0.292***
(0.022)
0.102***
(0.012)
ROA 0.359***
(0.053)
0.031
(0.025)
herf 0.602***
(0.042)
-0.020
(0.020)
Originality -0.094***
(0.013)
0.090***
(0.006)
Generality -0.098***
(0.013)
0.106***
(0.007)
Frac of
exploratory
-0.057***
(0.011)
0.048***
(0.006)
Frac of
exploitative
-0.003
(0.013)
0.108***
(0.007)
Number of
patents (pre)
0.041***
(0.000)
0.041***
(0.000)

Citations per
patent (pre)
0.007***
(0.000)
0.007***
(0.000)
constant 1.202***
(0.062)
1.346***
(0.068)
1.992***
(0.040)
1.929***
(0.043)
Industry FE Yes Yes Yes Yes
Year FE Yes Yes Yes Yes
Observations 40,204 40,204 40,204 40,204
Pseudo R2 0.100 0.105 0.356 0.357

76
Table 3.4. Productivity of Acquirer Inventor in Horizontal Mergers
This table reports the results from the quasi-experiment on acquirer inventors (inventors who
were in acquirer companies while M&A was taking place) in horizontal mergers. M&A
completed is an indicator variable that takes the value of one for the treatment group (completed
mergers) and zero for the control group (withdrawn mergers). Poisson regression is employed to
examine the change in both the number of patents and citations per patent of the inventors who
are in the firm both before and after the merger. Control variables include logsale, logME,
logRD, logq, INV, ROA, and herf as company characteristics and generality, originality, fraction
of exploratory patent, and fraction of exploitative patent as inventor characteristics. All company
characteristics are measured in the M&A bid year. All inventor characteristics are measured by
the average of patent measures over the three years before the merger. The unit of observation is
at the inventor-level. Merger dates in the sample correspond to the period 1980-2005. The robust
standard errors are presented in parentheses under each coefficient. *, **, and *** denote
significance at 10%, 5%, and 1%, respectively.

77

Number of Patents Citations per Patent
(1) (2) (3) (4)
M&A completed -0.174***
(0.020)
-0.174***
(0.021)
-0.099***
(0.010)
-0.094***
(0.010)
Log sale 0.067***
(0.011)
0.085***
(0.007)
Log ME -0.111***
(0.013)
-0.097***
(0.006)
Log RD -1.166***
(0.044)
0.147
(0.077)
Log Q 0.237***
(0.022)
0.120***
(0.012)
INV 0.317***
(0.034)
0.261***
(0.018)
ROA -0.217*
(0.087)
0.407***
(0.041)
herf 0.222*
(0.100)
0.335***
(0.044)
Originality -0.102***
(0.021)
-0.002
(0.010)
Generality -0.100***
(0.021)
0.069***
(0.012)
Frac of
exploratory
-0.078***
(0.019)
0.096***
(0.010)
Frac of
exploitative
-0.019
(0.022)
0.099***
(0.012)
Number of
patents (pre)
0.041***
(0.000)
0.041***
(0.000)

Citations per
patent (pre)
0.008***
(0.000)
0.008***
(0.000)
constant 1.537***
(0.119)
1.829***
(0.156)
2.342***
(0.056)
2.146***
(0.075)
Industry FE Yes Yes Yes Yes
Year FE Yes Yes Yes Yes
Observations 15,275 15,275 15,275 15,275
Pseudo R2 0.105 0.110 0.359 0.362

78
Table 3.5. Productivity of Target Inventor
This table reports the results from the quasi-experiment on target inventors (inventors who were
in target companies while M&A was taking place). M&A completed is an indicator variable that
takes the value of one for the treatment group (completed mergers) and zero for the control group
(withdrawn mergers). Poisson regression is employed to examine the change in both the number
of patents and citations per patent of the inventors who are in the firm both before and after the
merger. Control variables include logsale, logME, logRD, logq, INV, ROA, and herf as company
characteristics and generality, originality, fraction of exploratory patent, and fraction of
exploitative patent as inventor characteristics. All company characteristics are measured in the
M&A bid year. All inventor characteristics are measured by the average of patent measures over
the three years before the merger. The unit of observation is at the inventor-level. Merger dates
in the sample correspond to the period 1980-2005. The robust standard errors are presented in
parentheses under each coefficient. *, **, and *** denote significance at 10%, 5%, and 1%,
respectively.

79

Number of Patents Citations per Patent
(1) (2) (3) (4)
M&A completed -0.134***
(0.015)
-0.123***
(0.015)
0.034***
(0.007)
0.038***
(0.007)
Log sale -0.060***
(0.008)
-0.021***
(0.004)
Log ME 0.011
(0.010)
0.035***
(0.005)
Log RD 0.807***
(0.106)
0.485***
(0.053)
Log Q 0.180***
(0.016)
-0.081***
(0.009)
INV 0.199***
(0.025)
0.012
(0.011)
ROA -0.543***
(0.056)
0.098***
(0.028)
herf 0.356***
(0.052)
-0.313***
(0.024)
Originality -0.051**
(0.016)
0.076***
(0.008)
Generality -0.108***
(0.016)
0.117***
(0.009)
Frac of
exploratory
-0.071***
(0.015)
-0.0157*
(0.007)
Frac of
exploitative
0.026
(0.017)
-0.0125
(0.008)
Number of
patents (pre)
0.009***
(0.000)
0.009***
(0.000)

Citations per
patent (pre)
0.005***
(0.000)
0.005***
(0.000)
constant 1.748***
(0.200)
2.099***
(0.204)
1.958***
(0.112)
2.060***
(0.114)
Industry FE Yes Yes Yes Yes
Year FE Yes Yes Yes Yes
Observations 24,043 24,043 24,043 24,043
Pseudo R2 0.135 0.144 0.331 0.332

80
Table 3.6. Productivity of Target Inventor in Horizontal Mergers
This table reports the results from the quasi-experiment on target inventors (inventors who were
in target companies while M&A was taking place) in horizontal mergers. M&A completed is an
indicator variable that takes the value of one for the treatment group (completed mergers) and
zero for the control group (withdrawn mergers). Poisson regression is employed to examine the
change in both the number of patents and citations per patent of the inventors who are in the firm
both before and after the merger. Control variables include logsale, logME, logRD, logq, INV,
ROA, and herf as company characteristics and generality, originality, fraction of exploratory
patent, and fraction of exploitative patent as inventor characteristics. All company characteristics
are measured in the M&A bid year. All inventor characteristics are measured by the average of
patent measures over the three years before the merger. The unit of observation is at the
inventor-level. Merger dates in the sample correspond to the period 1980-2005. The robust
standard errors are presented in parentheses under each coefficient. *, **, and *** denote
significance at 10%, 5%, and 1%, respectively.

81

Number of Patents Citations per Patent
(1) (2) (3) (4)
M&A completed -0.105***
(0.024)
-0.149***
(0.025)
-0.055***
(0.011)
-0.054***
(0.011)
Log sale 0.060***
(0.017)
-0.0101
(0.009)
Log ME -0.082***
(0.019)
0.0112
(0.009)
Log RD 1.619***
(0.201)
0.175
(0.102)
Log Q 0.252***
(0.028)
-0.099***
(0.016)
INV 0.029
(0.042)
-0.069**
(0.022)
ROA 0.016
(0.101)
-0.239***
(0.046)
herf 0.852*
(0.134)
0.582***
(0.063)
Originality -0.008
(0.025)
-0.050***
(0.012)
Generality -0.103***
(0.025)
0.113***
(0.014)
Frac of
exploratory
-0.101***
(0.024)
-0.010
(0.012)
Frac of
exploitative
0.018
(0.027)
0.009
(0.013)
Number of
patents (pre)
0.008***
(0.000)
0.008***
(0.000)

Citations per
patent (pre)
0.010***
(0.000)
0.010***
(0.000)
constant 0.806
(0.671)
1.653*
(0.680)
2.868***
(0.223)
2.605***
(0.229)
Industry FE Yes Yes Yes Yes
Year FE Yes Yes Yes Yes
Observations 10,135 10,135 10,135 10,135
Pseudo R2 0.211 0.215 0.365 0.367

82
Chapter 4: How Investment in Private Equity Affect Economic Inequality in
the U.S.?

Weiran Deng

Abstract
In the past few years, income and wealth inequality in the United States have drawn tremendous
concern. As private equity attracts more and more attention, some people have connected
inequality with the exclusive private equity investment opportunity to accredited investors. In
this paper, I argue that (1) marginal accredited investors do not benefit from the opportunity of
investing in private equity; (2) general partners of startup companies, including general partners
in PE firms and angel investors, gain excess returns from private equity investment, which is one
of the factors driving the increasing income and wealth concentration.

Keywords: Inequality, Private Equity, Government Policy and Regulation
JEL Codes: D63, G24, G28
83
Introduction
In the past few years, income and wealth inequality in the United States drawn
tremendous attention and concern. It is well documented that since the late 1970s, there is an
increasing trend in the concentration of income and wealth. Precisely, by the time of 2010, top
10% households in terms of income (wealth) have more than 45% (75%) of total income
(wealth), top 1% households in terms of income (wealth) have about 20% (40%) of total income
(wealth). Existing explanations include saving rates, CEO compensations, and investment in
risky assets (stocks). Very little attention has been paid to investment in alternative assets that is
mainly consisted of private equity. Recently, as private equity attracts more and more attention,
some people have connected it with inequality. However, the center of this conversation is on the
regulation of accredited investors that prevents average investors who do not pass a certain
income or net worth threshold from investing in private equity.
In this paper, I argue that (1) marginal accredited investors do not benefit from the
opportunity of investing in private equity; (2) general partners of startup companies, including
general partners in PE firms and angel investors, gain excess returns from private equity
investment, which is one of the factors driving the increasing income and wealth concentration.
I will start with a literature review in income and wealth inequality in the U.S., household
portfolio composition, investment in private equity, and an argument proposing the amendment
of accredited investor standard. In the section “reconsidering the amendment of accredited
investor standard”, I show evidence on why marginal accredited investors do not gain excess
returns from private equity investment. Section “the mechanism that investment in private equity
affects inequality” will provide new perspectives linking the increasing income and wealth
84
inequality in the U.S. with the investment in private equity. The following section discusses
issues relevant to the arguments in this paper. The last section concludes.
Literature Review
Literature on Inequality in the United States
Inequality is an issue at the center of economics since Adam Smith (Gilbert 1997;
Rasmussen 2016). Its causes and effects have drawn tremendous attention among economists for
quite a long time. As to causes of inequality, considerations include trade, immigration, inflation,
family structure, age composition, technological change, unionization, compulsory schooling,
minimum wages, and CEO compensation (Card 2009; Jasso and Milgrom 2008; Leigh 2007;
Mishel and Sabadish 2012). High levels of inequality have been proved to have negative
consequences on not only the growth of economy but also stability of society, democracy, public
health, and environment (Cingano 2014; Ford 2019).
The most popular data for measuring inequality was from the Congressional Budget
Office (CBO) and Survey of Consumer Finance (SCF). Though the two sources cover relatively
detailed information about households at all income distribution, they are not long enough to
shed light on economic theories from historical trends, for instance, Kuznets’ influential
hypothesis. Kuznets argues that income inequality first rises with industrialization and then
declines once most of workers transfer to high-productive sectors of the economy. The trend of
inequality should follow an inverse-U shape along the economy’s development process (Kuznets
1955).
To have the longest possible historical series of inequality, Thomas Piketty pioneered the
use of individual tax return data from the IRS. In his paper with Emmanuel Saez, published in
2003, a historical series of top shares of pretax income in the United States is built. The series
85
covers the period from 1913 to 1998. Piketty and Saez reveal that the Great Depression and
World War II had a permanent effect on top capital income owners due to progressive taxations
after the huge shocks (Piketty and Saez 2003). Though this conclusion is interesting in itself, the
most influential finding of this paper was presenting how income is concentrated among the
richest: by 1998, the richest 1% in the U.S. had an income share of 15%. Furthermore, the
income share of top decile in total is as high as 42%. Following that, Wojciech Kopczuk and
Emmanuel Saez published a series of top wealth shares in the United States from 1916 to 2000
based on evidence from estate tax returns. The paper finds similar trends as top income shares,
except that wealth is even more concentrated: by 2000, the top 1% in terms of wealth own more
than 20% of the wealth of all the population (Kopczuk and Saez 2004). Top income or wealth
shares thus became popular measurements of economic inequality.
Before the prevail of the top distribution of income or wealth, the measurement of
inequality has been almost exclusively focused on the Gini coefficient. Research has shown that
top income share has a strong and significant relationship with other measures such as Gini
(Leigh 2007). Thus, top shares can be reasonably regarded as a representation of economic
inequality in general. Since then, the remarkable gains of those at the very top attracted
extensive attention, which is due to the measurement of inequality by share of top 10 or 1 or
even 0.1 percentile of income or wealth.
The series updated to the most recent years reveals an even more severe inequality issue
in the United States. By updating their series of top income shares up to 2010, Piketty and Saez
address that the top decile income share has risen from less than 35 percent during the 1970s to
over 45 percent in 2010. Top percentile income share has risen from less than 10 percent in the
86
1970s to about 20 percent (Piketty and Saez 2013)
22
. In 2014, Thomas Piketty published a book,
“Capital in the Twenty-First Century”, which soon became the international bestseller. In the
book, Piketty presents the dynamics of the distribution of income and wealth since the eighteenth
century by showing trends of top income and wealth share. He argues that rising inequality is
almost unavoidable under capitalism as the rate of return on capital is larger than the growth rate
of the economy (Piketty 2014). These findings push the discussion on inequality into a peak that
never had before.
After that, Thomas Piketty and his coauthors pay more attention to improving their
previous measures. Gabriel Zucman and Emmanuel Saez refine wealth inequality measurement
by applying the “income capitalization method” (Saez and Zucman 2016). The income
capitalization method aims to recover the distribution of wealth from the distribution of capital
income flows. The results confirm that the wealth concentration in the U.S. is high compared
with other countries and has considerably increased in recent decades. By 2012, the top decile
wealth share has risen from 65 percent in the early 1980s to over 75 percent. The top percentile
wealth share has reached 42%, mostly driven by the top 0.1 percent wealth share
23
. From 1978 to
2012, the top 0.1 percent wealth share grew from 7% to 22%, back to the level of inequality in
the early 20
th
century (Zucman 2019).
In 2018, Piketty, Saez, and Zucman published a paper together. Their primary purpose is
to study economic growth and inequality in a consistent framework, which requires the
aggregation of individual income to be the same as national income. To achieve the goal, they
combine data from tax returns, surveys, and national accounts and apply a “distributional
national accounts” methodology. Specifically, they start with individual tax return data and then

22
See Figure 4.1.
23
See Figure 4.2.
87
distribute the forms of income only in national accounts to each tier (top 1 percent, etc.) (Piketty,
Saez, and Zucman 2018, 2019). This new series confirms that income is extremely concentrated
at the top.
Some voices challenge Piketty’s statement. One strand of challenges is on data sources.
Economists from Federal Reserve Board noticed the gap between the measurement of top
income and wealth shares from tax return data used by Piketty and his coauthors and from survey
data by Survey of Consumer Finances (SCF). For example, the top 1 percent wealth share has
increased 13 percent from 1992 to 2013 using tax data, while rising only six percent by SCF
estimates. Similarly, the top 1 percent income share has increased ten percentage points since
1992 by Piketty’s estimates, while the SCF shows only eight percentage points increase. By
reconciling the differences in concepts and measurements around wealth and income, a new
estimate shows a more moderate increase in inequality (Bricker et al. 2016).
The other strand of the challenges is on assumptions that Piketty and his coauthors have
made. For example, a very recent paper presents a lower upward trend of top income share after
correcting technical tax issues and social issues (such as declining marriage rates) (Auten and
Splinter 2019). Other papers refine the income capitalization approach in Saez and Zucman
(2016) by further considering the heterogeneity of returns within asset classes. The results also
moderate the level of inequality, but again, do not change the fact that wealth is concentrated at
the very top (Bricker and Volz 2019; Smith, Zidar, and Zwick 2019). Piketty, Saez and Zucman
made a response to these challenges, arguing how their estimates remain valid (Piketty et al.
2019). As the debate attracted so much attention, there was even a court hearing held in October
2019 on measuring economic inequality in the United States. Economists who were invited to
the hearing addressed in their testimony that “there is no consensus in the research literature on
88
the measurement, level, and recent change of inequality” (Holtz-Eakin 2019). However, by any
measure, economic inequality in the United States has increased significantly over the past 40
years (Heather Boushey 2019).
Regardless of which the most precise measurement of inequality is, the idea that
inequality in the United States has significantly increased since the 1970s is already one of the
most powerful statements of our time. Broad acceptance of this view has fueled concerns with all
the negative consequences, including a greater concentration of political power and more rent-
seeking (Lindsey and Teles 2017; Stiglitz 2012), or increased bargaining power of top capital
owners (Piketty, Saez, and Stantcheva 2014). With the huge impact they have made, Piketty and
his co-authors’ research has become part of the political discourse. Piketty suggests wealth taxes
of 90 percent on billionaires, which is documented in his newly published book “Capital and
Ideology” (Piketty 2020). Saez and Zucman published a book titled “The Triumph of injustice:
How the Rich Dodge Taxes and How to Make Them Pay”, in which they propose a highly
progressive annual tax on wealth (Saez and Zucman 2019). The idea has been adopted by
Elizabeth Warren and Bernie Sanders, two former leading candidates for the Democratic
nomination for the American presidency.
Literature on Household Portfolio Composition
Regarding income and wealth, it is essential to know what their composition is before
trying to figure out what explains their distribution among the population. Income has two
components: labor income (wages) and capital income. Wealth is assets net of liabilities, while
assets come from the decision that households make to allocate their income. According to the
Survey of Consumer Finance (SCF), assets and liabilities are categorized into three groups,
financial assets, nonfinancial assets, and debts. Financial assets include liquid accounts,
89
certificates of deposit, savings bonds, other bonds, stocks, mutual funds, retirement accounts,
cash-value life insurance, trusts and other managed assets, and other financial. Nonfinancial
assets include primary residence, investment real estate, business equity, and other nonfinancial.
Under debts, there are mortgage and home equity, loans for investment on real estate, credit card
balances, and other debt.
Table 4.3, 4.4, and 4.5 are from Bricker et al. (2019) calculated from data published by
the Survey of Consumer Finance (SCF). Table 4.3 shows that from 1989 to 2016, nearly all
families owned assets of some kind. We pay particular attention to “business”, which covers net
equities in all types of privately-owned businesses. The percentage of families that hold this type
of assets varies from 21.1% to 17.4% throughout the years, higher than the percentage of
families that directly hold stocks or mutual funds. Note that, according to the definition of
“business” in the SCF, the household may have an active management role in the business or
may only invest in the business. Thus, the percentage of families that hold “business” assets
covers both entrepreneurs and private company investors.
Table 4.4 shows that mean asset holdings nearly doubled during the sample period,
reaching $787K in 2016. The mean value of assets under the “business” category also nearly
doubled, reaching $184,200 in 2016. Business asset is an asset class with the second-largest
mean value, only below the primary residence.
Who owns the business assets in the end? Table 4.5 shows the average assets and debts
for each asset group. In general, assets are highly concentrated in the top 1% asset share
households. The top 1% have 30 times more assets than the assets held by an average household,
which is true throughout the years and is in an increasing trend. Furthermore, figure 4.5 shows
the composition of assets of the wealthiest 1 percent by year. From 1989 to 2016, privately held
90
business assets kept being the largest asset class that the wealthiest 1 percent held in their
portfolios
24
. In summary, it is the wealthiest 1 percent who own most of the business assets.
There is inconsistency in the top wealth composition under alternative specifications,
though. Specifically, the allocation of wealth following the capitalization method in Saez and
Zucman (2016) and that from SCF are different. SCF estimates always deliver a higher
concentration in business assets in each survey year. This difference is partially due to different
valuation concepts (Bricker and Volz 2019). For investment in privately held businesses, the
SCF measures it in terms of its market value while the other measures it in terms of book value.
Market value considers unrealized capital gains, which makes it often higher than the book value
of the same business. Bricker and Volz (2019) show that the wealthiest families’ asset portfolios
across different datasets for different countries all skew toward business assets. Research also
shows that unrealized capital gains share a larger proportion of wealthy families’ portfolios over
time (Avery, Grodzicki, and Moore 2015). Thus, this paper takes the point of top wealth
composition from SCF estimates.
Literature on Private Equity
Having assets in private business could either mean the household having an active
management role in the business, or only investing in the business. The former is more relevant
to return from entrepreneurship activity, which is not the main issue discussed in this paper. This
section will review existing literature about investment in private businesses, which is mainly
through private equity funds.
Essentially, private equity is capital that goes to unlisted companies in exchange for
equity. Unlisted company, or say private company, is a business that does not offer shares to the

24
Though surpassed by “equities” slightly (one percentage point) in 1998 and 2001 survey.
91
general public on the stock market exchange. Private equity includes two main types of
investments: venture capital (VC) and buyouts. Venture capital investments focus on startups
with high growth potential. Buyout investments acquire publicly traded companies with poor
performance and then take them private
25
.
The business model of private equity firms is as follows – raise capital from external
capital contributors, invest the capital in a series of private equity transactions, sell those
investments, and return the proceeds to the external capital contributors while keeping a share of
the total profits. The private equity firm is called general partner (GP), and the external capital
contributors are called limited partners (LPs). Each GP may manage one or a few funds with
different investment focuses in terms of geography or industry or other deal characteristics. Each
fund invests in 10 to 12 portfolio companies and takes about 5 to 7 years to exit. The percentage
of profits that GP keeps includes both a management fee and a performance fee. Typically,
management fees consist of 2% of assets under management and performance fees consists of
20% of the profits from exited investments. Performance fees are also called “carried interest” or
“carry”.
The PE industry has grown enormously in the United States. According to the PitchBook,
249 U.S. buyout and other PE funds (excluding VC) in total raised $315 billion in 2019, making
5,433 deals that worth $756 billion (PitchBook 2020a). 308 U.S. venture capital funds raised $51
billion in 2019, making 11,923 deals worth $136 billion (PitchBook & NVCA 2020)
26
.
The high growth rate of the PE industry is mainly due to the high returns in its early
years. The mean arithmetic returns from PE investments could be as high as 698% (Cochrane

25
In some databases such as PitchBook, VC funds are separated from PE funds. PE funds in PitchBook include
buyout, growth/expansion, diversified private equity, mezzanine, and restructuring/turnaround.
26
See Figure 4.4 and Figure 4.5.
92
2005), which is stronger than any other asset class. However, this abnormally high return suffers
from the survivor-bias: payoffs can be observed only if portfolio companies have survived. Plus,
the illiquid nature of private equity funds disables continuous measurement on their
performance. The returns are only realized when the funds are no longer active. Numerous
efforts have been made to offer more precise estimates on the performance of private equity
funds. So far, there are two main strands of literature on the performance of private equity. The
first strand is about risk and return. The second strand is about performance persistence.
The most typical measure of PE performance is Internal Rate of Return (IRR), which is
essentially a discount rate making the present value of a series of investment equal to the present
value of the returns on those investments. The literature has developed new methods of assessing
the performance of PE funds, such as constructing a venture capital index and applying new
econometrics models. Specifically, Kaplan and Schoar propose to use public market equivalent
(PME) as a better market performance measure (Kaplan and Schoar 2005). Cochrane uses a
maximum likelihood approach to correct the selection bias (Cochrane 2005). Hwang, Quigley,
and Woodward build a VC index from repeat valuations from sales (Hwang, Quigley, and
Woodward 2005). Korteweg and Sorensen propose a risk and return model using Bayesian
Markov Chain Monte Carlo (MCMC) (Korteweg and Sorensen 2010). All of the research
contributed to improve PE performance assessment, but the goal of establishing a universally
accepted performance measure has not been reached yet.
Regarding PE risk and return, existing literature generally agrees that PE funds
outperform the public market gross of fees. For example, Ljungqvist and Richardson document
excess returns over the S&P 500 Index of close to 6% annually, net of all fees (Ljungqvist and
Richardson 2003). Phalippou and Gottschalg document 3% outperformance gross of fees and 3%
93
underperformance net of fees over the S&P 500 Index for liquidated PE funds (Phalippou and
Gottschalg 2009). Lopez de Silanes, Phalippou, and Gottschalg estimate a median IRR of 21%
and a median PME of 1.3 gross of fees (Lopez-de-Silanes, Phalippou, and Gottschalg 2010). In
more recent articles, Harris, Jenkinson, and Kaplan document more than 3% annually
outperformance over the S&P 500 for buyout funds, while VC funds outperformed public
equities in the 1990s but underperformed in the 2000s (Harris, Jenkinson, and Kaplan 2014). In
general, PE funds have large and positive excess returns during the 1990s, but the returns from
public and private equities are getting closer during the 2000s (Kartashova 2014).
What drives PE return? Macroeconomic factors and public equity market movements
have been shown to have influences on PE return (Gompers et al. 2008; Phalippou and Zollo
2006). Besides these, higher PE returns could also be caused by the skill of GP and the
relationship between the LP and the GP (Ewens, Jones, and Rhodes-Kropf 2013; Ewens and
Rhodes-Kropf 2015). Marquez, Nanda, and Yavuz argue that higher returns are related to
restricted fund size (Marquez, Nanda, and Yavuz 2015). Along this line, Korteweg and
Sorensen’s work confirms that smaller funds have greater long-term persistence than larger funds
(Korteweg and Sorensen 2017).
Regarding the performance persistency of PE funds, the evidence is mixed, most articles
have documented some degree of persistence. Kaplan and Schoar document that returns persist
strongly for up to two funds, which is likely due to the quality of GP (Kaplan and Schoar 2005).
Later on, Phalippou argues that the Kaplan and Schoar did not use “an ex-ante measure of the
performance of earlier funds” so that the persistence is overstated (Phalippou 2010). Using a
unique database containing cash flow data of buyout funds, Braun and coauthors find that
persistence has declined as the PE sector becomes more competitive (Braun, Jenkinson, and
94
Stoff 2015). In a more recent work, Korteweg and Sorensen use a variance decomposition model
to and find a high long-term persistence. Top quantile PE firms have annual expected returns that
are 7 to 8 percentage points higher than those of bottom quantile PE firms, net of fees (Korteweg
and Sorensen 2017).
Literature on Accredited Investor
An existing argument connecting inequality with private equity focuses on a security law
about “accredited investors”. Essentially, this securities law sets a threshold in terms of income
and net worth for investing in private equities.
The notion of the accredited investor has a long history. Bender (2015) has a detailed
review on how the notion developed to the current version. In a sum, the accredited investor
concept first appeared in The Securities Act of 1933, which requires purchasers of unregistered
securities offering to meet a standard. The SEC determined the standard using a subjective
analysis until Congress enacted the Small Business Investment Incentive Act of 1980, which
added section 4(a)(5) to the existing Securities Act. To implement this section, the SEC in 1981
defined the accredited investor standard in Rule 501 of Regulation D
27
. Individuals are
characterized as accredited investors if: “their income exceeds $200,000 in each of the two most
recent years (or $300,000 in joint income with their spouse) and they reasonably expect to reach
the same income level in the current year; or their net worth exceeds $1 million (individually or
jointly with a spouse). In addition, directors, executive officers, and general partners of the issuer
selling the securities are accredited investors for purposes of that issuer.”
On December 21, 2011, the SEC adopted Dodd-Frank net worth standard that excludes
the value of a person’s primary residence when determine whether a person qualifies as an

27
Regulation D is a private offering exemption, comprising Securities Act Rules 500 to 506.
95
accredited investor based on having a net worth above $1 million (Securities and Commission
2011). This change is a response to the increasing concern after the 2008 financial crisis, which
argues that investment in private placements is highly risky and should not allow investors who
cannot bear the loss to participate.
The SEC has estimated the number of households that qualify under existing accredited
investor criteria. Data in table 4.6 provides an estimate of the overall pool of qualifying
households in the United States. In 2019, roughly 16.0 million households qualified under
existing accredited investor criteria
28
, which is about 13% of the population. It does not,
however, represent the actual number of accredited investors that do or would invest in the
exempt offerings.
However, voices asking for refining the accredited investor standard continued and
moved toward an opposite direction. In December 2015, the SEC issued a staff report on the
accredited investor definition, with the objective of expanding the eligible pool of sophisticated
investors. In October 2017, the U.S. Department of the Treasury prepared a report that included
recommendations to revise the accredited investor definition. The 2017 Treasury Report stated
that the definition could be broadened to include: “(a) any investor who is advised on the merits
of making a Regulation D investment by a fiduciary, such as an SEC- or state-registered
investment adviser; and (b) financial professionals, such as registered representatives and
investment adviser representatives, who are considered qualified to recommend Regulation D
investments to others” (Securities and Exchange Commission 2019). The SEC then released an
amendment proposal in 2019. The proposal is still under consideration but has already received
supportive comments from the public and professionals.

28
Including households qualify based on either individual income threshold, or joint income threshold, or net worth.
96
The primary rationale for expanding the eligible pool of accredited investors is that the
accredited investor requirement favors economic inequality in the United States. Although the
regulation was aimed at protecting non-accredited investors who lack financial sophistication,
this also means that the possible high return from investing in private capital markets can only be
enjoyed by “accredited investor”.
Kevin G. Bender advocates to “give the average investor the keys to the kingdom” by
analyzing how the federal securities laws facilitate wealth inequality. He reviews securities laws,
including Securities Act Section 4(a)(2), Regulation D, the Accredited Investor, Investment
Company Act Section 3(c)(7) and the Qualified Purchaser, and Secondary Market Transfers and
Rule 144. He also shows how average risk premium from private equity funds (buyout funds),
venture capital funds, hedge funds, or angel investments are often much higher than that of the
S&P 500 and bonds. He, therefore, argues that “securities laws have deprived average investors
of the right to have higher returns from PE/VC funds and to have adequate portfolio
diversification” (Bender 2015).
Jeff Thomas argues that redefining the notion of accredited investor is “one small step for
the SEC, one giant leap for our economy” (Thomas 2019). He states that converting current non-
accredited investors into accredited investors would create fair investment opportunities and
provide an extra source of capital for entrepreneurial ventures.
Stuart Ford writes in another paper that the current levels of inequality are themselves the
product of government action. Namely, federal securities laws facilitate growing wealth
inequality is one of the evidences. He concludes that there is a need for a wealth inequality
amendment (Ford 2019).
Reconsidering the Amendment of Accredited Investor Standard
97
Currently, the relationship between inequality and accredited investors standard is argued
in a way as if all accredited investors will invest in private equity funds and can get positive
returns higher than that from investments in the public equity market.
However, before drawing such a conclusion, we should at least ask: who actually invests
in private equity funds? Does everyone invest in those funds benefit from this exclusive
opportunity? In this section, I argue that actually only very few among the limited partners of
private equity funds are accredited investors. Furthermore, even if they invest, the probability
that accredited investors can have excess returns than the public equity market is very low. Thus,
slowing down inequality in the U.S. should not be the purpose of expanding the pool of
accredited investors.
Who are the Limited Partners?
Evidence shown below uses data from PitchBook. I collect data of all funds labeled as
“private equity” or “venture capital” located in the United States. There are 16,847 funds in total
invested by 3,802 limited partners. Table 4.7 summarizes the frequencies of each type of limited
partner. Foundations, corporation pension, insurance company, and public pension funds take up
more than half of all limited partners. The only type of LPs constrained by qualifying as
accredited investors is “high-net-worth investors”, which is less than 7% among all limited
partners in the sample.
Compared with the relatively massive base of qualified accredited investors, which is 16
million households as the estimation in Table 4.6 shows, investors who actually invest in private
equity could be just a tiny portion, even with consideration of selection to the sample. Arguing
that a change in the threshold of accredited investors would have a visible impact on inequality
lacks the requisite understanding of limited partners of private equity funds.
98
The Skewness of Private Equity Performance
The key assumption justifying the expansion in accredited investor pool is that the
investment returns from private equity is higher than that from public market investments. The
assumption is true, but only on average. We should not ignore the most significant feature of the
PE investment: skewness. Investment returns from private equity have a much larger standard
deviation than those from public equity. The average of PE investment return is always higher
than the median of PE investment return. Investors could both win a lot and lose a lot by making
bets in private equity. In PitchBook, there are 4,243 funds reported their returns data, among all
16,847 funds that are labeled as “private equity” or “venture capital” located in the United
States
29
. 15% of these 4,243 funds have no higher than zero internal rate of return (IRR), and
36% of these funds have IRRs less than 10%
30
.
This above evidence shows that only a portion of limited partners of private equity funds
can get returns higher than that from public stock market investments. What is worse, there is
performance consistency at least to some extent in private equity return, which means several
private equity firms can outperform in many of their funds. Then the question actually comes to:
who are exclusive to those outperformers?
Research about the performance persistence of private equity has been summarized in the
literature section. The most recent report from Bain & Company studies a pool of 113 private
equity firms that each have raised $5 billion or more since 2000. The author defines a PE firm as
an outperformer if at least 80% of a firm’s funds ranked in the top two quartiles of the industry
performance over that period. It turns out that 28 PE firms out of the 113 are outperformers, and

29
Fund returns data from PitchBook is composed of LP reports. For the majority of LPs such as public pensions and
investment consultants, only net value of IRR is available, so fund IRR will be considered net of fee here.
30
The average S&P 500 annual returns from 1980 to 2019 is 10.13%.
99
they had many more deals with an IRR above 15% (32% of their portfolios vs. 18% for the
laggards) and far fewer write-offs (5% vs. 8%) (Macarthur et al. 2020).
When a PE firm is regarded as an outperformer, its bargaining power becomes much
larger than limited partners while making investment commitments, as the supply of capital is
larger than the demand of capital to those PE firms. The literature has shown that consistent
outperformers usually have relatively smaller fund size to guarantee the quality of the fund
management (Marquez et al. 2015). A result of this is that the consistent outperformers select
money from certain people to accept as commitments. As limited partners do not play an active
role in fund management, the best PE firms do not select LPs by their skills. Therefore, the
criterion that top PE firms use to accept commitments is most likely to be the amount of money
each LP invest, which can reduce the number of limited partners and also reduce the cost of
management. For example, venture capital firm Sequoia is known for its early investments in
companies including Google, LinkedIn, WhatsApp, and Dropbox. According to CNBC, Sequoia
has set the minimum investment for its global growth fund as $250 million (Levy 2018). This bar
might be easy to reach for institutional LPs, but not for individual investors. Even if we assume
that a high-net-worth investor contributes 100% of his/her wealth into Sequoia, this means that
the investor must have a net worth above the threshold of the top 0.01% households
31
and is of
course far above the threshold set for accredited investors.
The Pool of Accredited Investors is Enlarging Over Time
We have seen in the literature review that both the income inequality and the wealth
inequality in the U.S. have climbed up rapidly since the 1980s. Besides, the number of accredited
investors grows over time as the monetary criteria did not change since the accredited investor

31
Refer to Table 4.2.
100
standard was set up. In Table 4.6, the estimation by the SEC shows that the share of qualified
accredited investors as of all U.S. households has increased from 1.6% to 13% from 1938 to
2019.
If enlarging the pool of accredited investors can slow down the increasing trend of
economic inequality in the U.S., why did we not see it in the past 30 years? It is not convincing
to argue that the inequality issue will be even more severe if the pool of accredited investors did
not enlarge over time.
To sum up, I state that the marginal accredited investors do not benefit from the
investment opportunities in the private equity market. Amendment of expanding the accredited
investors’ pool should not use slowing down inequality in the U.S. as the ground.
The Mechanism that Investment in Private Equity Affects Inequality
The existing literature has documented that income and wealth in the U.S. have
concentrated increasingly. Among the richest households, ownership of private business equity
makes up a large proportion of their asset portfolio. Private business equity is mostly owned by
the richest households. Research in private equity performance generally agrees that the
investment return from private equity is on average higher than public market equivalents. If we
connect all these findings together, it is natural to hypothesize that income and wealth
concentration in the U.S. is related to private equity investment, arguably because the
opportunity to invest in private equity is practically exclusive to a certain group of people.
The notion of accredited investor is clearly the most apparent threshold that we can
target. However, as the previous section shows, accredited investors who are at the margin
cannot gain excess returns from private equity investment. The question then shifts to figure out
who is exclusive to the excess return of private equity investment.
101
In this section, I will offer my opinion on how private equity investment affects the
income and wealth inequality in the United States and then add evidence about the decline in the
number of IPOs in the U.S. to demonstrate the severity of the issue.
General Partners Gain Excess Returns from Investment in Private Equity
General Partners in Private Equity Firms
General partners in PE firms are never prevented from investing in private equity, no
matter what income or net worth level they have at the start of their career. The nature of their
job allows them to have exclusive access to high returns through the “carries”. “Carries” are fees
that PE firms charge to their limited partners. It often takes 20% of all profits that PE firms
generate. Since 1980, general partners in PE firms have made up a growing share of the highest
income groups (Scigliuzzo, Butler, and Bakewell 2019). Kaplan and Rauh have shown that
“there are more private equity managers make at least $100 million annually than investment
bankers, top financial executives, and professional athletes combined” (Kaplan and Rauh 2013).
Their estimation also shows that private equity fees jumped to $34 billion per year (in 2010
dollars) between 2005 to 2011 from $1 to $2 billion annually back in the late 1980s.
Other statistics also shows that salaries of PE general partners are astonishingly high.
Table 4.8 is from the 2019 publication of the website “efinancial careers”. It documents that the
total remuneration for managers in U.S. PE firms is $850k averagely, and the number is as high
as $1.6 million for managing directors and partners. With such an income, a general partner in a
PE firm can easily make herself one of the richest in terms of the top 1% or even the top 0.1%
income share group
32
. In Appendix 4-A, I list all individuals in 2019 U.S. Forbes 400 ranking
that hold primary positions in private equity firms as general partners. The data source is the

32
Refer to Table 4.1.
102
PitchBook and Forbes 2019 ranking. There are 46 individuals in total. Seventeen of them are
general partners in venture capital firms. The others are general partners in buyout firms
33
.
Private equity is the source of wealth for more than half of them.
This exclusive access to high returns from private equity is unrelated to the accredited
investor standard. It is rooted in the current operating structure in the PE industry. The high
salary of PE firm executive explains the trend of increasing wage income share among the
richest households as documented in existing research (Piketty and Saez 2003, 2013). Figure 4.6
is from Piketty and Saez (2013). It shows that wage income gained an increasing share in the
income composition of the top groups with the top decile from 1929 to 2007. Current
explanations of this trend focus on increasing CEO compensation and increasing return from
entrepreneurial activity. More attention should be paid on PE firm general partners.
Angel Investors
An angel investor is an individual investor who provides both finance and business
expertise to a company in which he invested. An angel investor is also a general partner of the
company. By searching the names of all 400 individuals in the 2019 Forbes list of the U.S. in
PitchBook, I found that most of them invest in private companies as angel investors. Only 60 out
of the 400 individuals do not have any record of investment associated with private equity.
According to PitchBook, among all U.S. high-net-worth individuals who invest in private
equity, 7,328 are angel investors, and 312 are limited partners. This evidence shows that to invest
in private equity as an. individual, a more conventional way is to invest as an angel investor.
Angel investors have to diversify their portfolios by making more bets in the private market, as a
single angel investment does not have the advantage of diversification as an investment in fund

33
Labeled as private equity in PitchBook.
103
has. This means that angel investors must be extremely wealthy. Angel investment is, therefore,
highly correlated with wealth inequality.
By combining all the evidence, I state that general partners, including general partners in
PE firms and angel investors investing in PE, have exclusive access to the excess return from PE
investment. General partners in PE firms benefit from the rule in the PE industry, which allows
them to keep the “carry” as part of the salaries. Angel investors investing in PE benefit from the
exclusivity to diversify PE portfolio by an extremely large amounts of wealth.
The Decline in the Number of Domestic IPO
Figure 4.7 shows the number of U.S. listings from 1980 to 2018. Data is from the World
Bank database. Since 1996, the number of U.S. listings dropped dramatically, from 8,025 to
about half of this level. The fact is related to the relationship between investment in PE and
economic inequality as the shrink of the number of U.S. public companies means shrinking
investment opportunities for public market investors.
Doidge and his co-authors are among the first who document this trend. The authors state
that this abnormally low listing rate cannot be explained by regulatory changes (Doidge, Karolyi,
and Stulz 2013). Both the high delist rate and low new listing rate contribute to this status.
(Doidge, Karolyi, and Stulz 2017). Gao, Ritter and Zhu, at the same time, notice that the average
number of companies that went public per year in the U.S. dropped from 300 to 99 between 1980
and 2000. The authors argue that the change is like explained by the increasing benefits from
selling out to larger organizations, compared with operating as an independent firm (Gao, Ritter,
and Zhu 2013). From another perspective, Ewens and Farre-Mensa highlight the correlation
between the decline in IPOs and the development of the PE sector. They argue that the
104
deregulation of securities laws
34
has increased the supply of private capital to late-stage private
startups. Founders with increased bargaining power versus investors now prefer to let their
startups to stay private longer (Ewens and Farre-Mensa 2020).
In 2013, the term “unicorn” was coined by Aileen Lee to refer to private companies with
valuations of $1 billion or more. Figures 4.8 shows that the cumulative unicorn valuation has
reached $500 billion until 2018. From figures 4.9 and 4.10, we can see that valuations of VC-
backed startups are rising during the investments and at the exits via either IPO or acquisition
(PitchBook & NVCA 2019). However, the growth in valuation did not continue after exits. In
2019, eight IPOs that raised more than $1 billion, but many of them struggled to stay above their
offer prices. Specifically, Lyft, SmileDirectClub, and Uber have negative returns as high as more
than 35% since their IPOs. Furthermore, there are only three companies among the eight that
have positive returns since debut, which are Tradeweb, Pinterest, and Chewy (Strauss 2019). It is
clear that more companies experience most of their growth while staying private, demonstrating
the high investment return that certain investors can get from and only from investing in private
equity funds.
Discussion
Public Pension Funds Invest in Private Equity
The public pension fund is one of the largest limited partners of PE firms. Investment in
PE is regarded as a way to diversify portfolios and gain higher returns by many pension funds. It
is arguable whether pensions make the investment in PE an equal investment opportunity and
thus weaken the impact of PE on inequality.

34
in particular the National Securities Market Improvement Act (NSMIA) of 1996.
105
The California Public Employees’ Retirement System (CalPERS) is the largest public
pension fund in the United States. As of December 2019, CalPERS allocate 6.6% of its $395
billion of the Public Employees’ Retirement Fund (PERF) into private equity. The private equity
program fund performance review published on its official website shows that the since
inception net IRR is 10.7% as of September 2019 (CalPERS 2019). The return is not
significantly higher than from public market investments. Academic research also shows that
public pension funds did not perform very well in their PE allocations (Lerner, Schoar, and
Wongsunwai 2007).
Based on the above evidence, it is reasonable to state that public pension funds were not
able to generate excess returns from their PE allocations. Thus, public pension funds did not help
to weaken the inequality issue caused by unequal investment opportunities.
Private Market Investment Opportunity through Crowdfunding
The Jumpstart Our Business Startups Act (JOBS Act) was signed into law in April 2012.
It is intended to encourage small businesses in the United States to use crowdfunding to issue
securities and allow ordinary individuals to have the ability to invest in startup companies. The
rules went into effect on May 16, 2016 after the SEC adopted final rules. Crowdfunding could be
a way to equalize investment opportunities. However, it is too early to draw any conclusion as
we have not seen enough evidence on crowdfunding performance.
Social Networks
PE firms and angel investors provide both capital and expertise to the companies in
which they invest. Competitive startups with high potential have the bargaining power to select
their investors.
106
An investor’s expertise can be easily identified if the investor is a successful CEO of a
huge organization when the investor is an angel. Otherwise, competitive startups are likely to
select their investors through personal connections. Common education or working experience
between startup founders and general partners may lead to a higher probability of deal
generation. This hypothesis is a research direction worth to be explored.
World Trend
The trend of increasing income and wealth inequality is not only experienced by the
United States. Globally, almost all countries exhibit this trend, but at different speeds. The
literature has documented that in English speaking countries plus India and China, the inequality
has increased more than in other countries (Alvaredo et al. 2018; Atkinson, Piketty, and Saez
2011). It will be both interesting and important to figure out how private equity development is
related to inequality in the worldwide.
Conclusion
This paper reviews the literature on income and wealth inequality in the U.S., household
portfolio composition, investment in private equity, and an argument proposing the amendment
of accredited investor standard. More importantly, it provides new perspectives to link the
increasing income and wealth inequality in the U.S. with the investment in private equity. I argue
that (1) marginal accredited investors do not benefit from the opportunity of investing in private
equity; (2) general partners of startup companies, including general partners in PE firms and
angel investors, gain excess returns from private equity investment, which is one of the factors
driving the increasing income and wealth concentration.
Among accredited investors, only a tiny portion of them actually invest in private equity
funds. The skewness of private equity performance determines that not all PE funds can have
107
positive returns, let alone returns that beat the public equity market. The consistent
outperformers among PE funds select investors by setting minimum commitment amount, an
amount that is much higher than the accredited investor standard.
General partners of PE firms keep roughly 20% of the profits from investment as part of
their salaries. In other words, they are guaranteed with pays higher than the net of fee IRR that
ordinary limited partners get. Among the top 400 wealthiest Americans ranked by Forbes 2019,
46 of them currently hold primary positions in PE firms as general partners. Private equity is the
source of wealth for more than half of these 46 people. Furthermore, most of these 400
wealthiest Americans invest in private equity as angel investors. Only 60 of them do not have
any private equity investment record in the PitchBook database. Angel investors provide both
capital and expertise to the companies in which they invest in. They are also general partners as
they make transactions with the companies directly. Being an angel investor requires an even
larger wealth bar as the investment risk can only be diversified by making many large-size
investments.
The conclusion has substantial policy implications. It suggests that expanding the pool of
accredited investors will not significantly weaken income or wealth concentration. Policymakers
shall pay more attention to the operational structure inside the private equity industry. For
instance, regulating the pay structure of PE firms’ partners and managers and lowering the
minimum commitments required by the top PE funds. Some ongoing policies might play a role
in equalizing the playing field. For example, as of May 2016, the JOBS Act allows ordinary
individuals to invest in startups through crowdfunding. However, there is not enough evidence to
show the performance of crowdfunding until now. Another way to distribute the excess return
from private equity to ordinary individuals is through public pension funds. Some public pension
108
funds have already invested in private equity. However, the investment return is not higher than
that from the public equity allocation. Improving public pension funds’ ability to select private
equity funds might be a way to reduce the economic inequality caused by the exclusive access to
high private equity returns.

109
Figure 4.1. Top Income Shares in the United States, 1917-2010
Figure 4.1 (a). The Top 10% Income Share in the United States, 1917-2010

Source: Piketty and Saez (2013). The figure plots the top 10% income share in the United States
from 1917 to 2010. The unit is the family (single adult person aged 20 or more, with or without
children dependents, or married couple with or without dependents). Capital gains are included.

25%
30%
35%
40%
45%
50%
1917
1920
1923
1926
1929
1932
1935
1938
1941
1944
1947
1950
1953
1956
1959
1962
1965
1968
1971
1974
1977
1980
1983
1986
1989
1992
1995
1998
2001
2004
2007
2010
Share of total income going to Top 10%
110
Figure 4.1 (b). Decomposing the Top 10% U.S. Income Share into Three Groups, 1917-
2010

Source: Piketty and Saez (2013). The figure plots the top 1% and next 4% (top 5-1%) and the
next 5% (top 10-5%) income shares in the United States from 1917 to 2010. The unit is the
family (single adult person aged 20 or more, with or without children dependents, or married
couple with or without dependents). Capital gains are included.

5%
7%
9%
11%
13%
15%
17%
19%
21%
23%
25%
1917
1920
1923
1926
1929
1932
1935
1938
1941
1944
1947
1950
1953
1956
1959
1962
1965
1968
1971
1974
1977
1980
1983
1986
1989
1992
1995
1998
2001
2004
2007
2010
Share of total income accruing to each group
Top 10-5% Top 5-1% Top 1%
111
Table 4.1. Threshold and Average Incomes in Top Income Groups in the United States in
2010

Percentile
Threshold
Income
Threshold
Income Groups Number of
Families
Average Income
in Each Group
Full Population 156,167,000 $51,550
Bottom 90% 140,550,300 $29,840
Top 10% $108,024 Top 10-5% 7,808,350 $125,627
Top 5% $150,400 Top 5-1% 6,246,680 $205,529
Top 1% $352,055 Top 1-0.5% 780,835 $418,378
Top 0.5% $521,246 Top 0.5-0.1% 624,668 $798,120
Top 0.1% $1,492,175 Top 0.1-0.01% 140,550 $2,802,020
Top 0.01% $7,890,307 Top 0.01% 15,617 $23,846,950

Source: Piketty and Saez (2013). Computations based on income tax return statistics. Income
defined as market income (annual gross income reported on tax returns excluding all government
transfers and before individual income taxes), including realized capital gains.

112
Figure 4.2. Top Wealth Shares in the United States, 1917-2012
Figure 4.2 (a). The Top 10% Wealth Share in the United States, 1917-2012

Source: Saez and Zucman (2016). The figure plots the wealth share of the top 10% in the United
States from 1917 to 2012 using the capitalization method. The unit is the family (single adult
person aged 20 or more, with or without children dependents, or married couple with or without
dependents).

60%
65%
70%
75%
80%
85%
90%
1917
1920
1923
1926
1929
1932
1935
1938
1941
1944
1947
1950
1953
1956
1959
1962
1965
1968
1971
1974
1977
1980
1983
1986
1989
1992
1995
1998
2001
2004
2007
2010
Share of total household wealth
113
Figure 4.2 (b). Decomposing the Top 10% U.S. Wealth Share into Three Groups, 1917-2012

Source: Saez and Zucman (2016). The figure plots the top 1% and next 4% (top 5-1%) and the
next 5% (top 10-5%) wealth shares in the United States from 1917 to 2012 using the
capitalization method. The unit is the family (single adult person aged 20 or more, with or
without children dependents, or married couple with or without dependents).

5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
55%
1917
1920
1923
1926
1929
1932
1935
1938
1941
1944
1947
1950
1953
1956
1959
1962
1965
1968
1971
1974
1977
1980
1983
1986
1989
1992
1995
1998
2001
2004
2007
2010
Top 10-5% Top 5-1% Top 1%
114
Figure 4.2 (c). Top 1%, Top 1-0.1%, and Top 0.1% U.S. Wealth, 1913–2012

Source: Saez and Zucman (2016). The figure plots the share of total household wealth owned by
the richest 1%, 1-0.1%, and 0.1% of families in the United States from 1913 to 2012 using the
capitalization method. The unit is the family (either a single person aged 20 or above or a
married couple, in both cases with children dependents if any).

5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
55%
1913
1916
1919
1922
1925
1928
1931
1934
1937
1940
1943
1946
1949
1952
1955
1958
1961
1964
1967
1970
1973
1976
1979
1982
1985
1988
1991
1994
1997
2000
2003
2006
2009
2012
Top 1% Top 1-0.1% Top 0.1%
115
Table 4.2. Threshold and Average Wealth in Top Wealth Groups in the United States in
2012

Percentile
Threshold
Wealth
Threshold
Wealth Groups Number of
Families
Average Wealth
in Each Group
Full Population 160,700,000 $343,000
Bottom 90% 144,600,000 $84,000
Top 10% $660,000 Top 10-1% 14,463,000 $1,310,000
Top 1% $3,960,000 Top 1-0.1% 1,446,300 $7,290,000
Top 0.1% $20,600,000 Top 0.1-0.01% 144,600 $39,700,000
Top 0.01% $111,000,000 Top 0.01% 16,070 $371,000,000

Source: Saez and Zucman (2016). This table reports statistics on the wealth distribution in the
United States in 2012 obtained by capitalizing income tax returns. The unit is the family (either a
single person aged 20 or above or a married couple, in both cases with children dependents if
any). Fractiles are defined relative to the total number of families in the population.

116
Table 4.3. Percent of Families that Hold Assets, 1989–2016
1989 1992 1995 1998 2001 2004 2007 2010 2013 2016
Any asset 94.7 95.8 96.4 96.8 96.7 97.9 97.7 97.4 97.9 99.4
Financial
asset
88.9 90.3 91.2 93.1 93.4 93.8 93.9 94.0 84.5 98.5
Liquid
asset
85.5 86.9 87.4 90.6 91.4 91.3 92.1 92.5 93.2 98.0
CDs
35
19.9 16.7 14.3 15.3 15.7 12.7 16.1 12.2 7.8 6.5
Mutual
36

fund
7.2 10.4 12.3 16.5 17.7 15.0 11.4 8.7 8.2 10.0
Stocks
37
16.8 17.0 15.2 19.2 21.3 20.7 17.9 15.1 13.8 13.9
Bonds
38
5.7 4.3 3.1 3.0 3.0 1.8 1.6 1.6 1.4 1.2
Managed
account
3.6 4.0 3.9 5.9 6.6 7.3 5.8 5.7 5.2 5.5
DC
pension
39

or IRA
40

37.1 40.1 45.3 48.9 52.8 49.9 53.0 50.4 49.2 52.1
Other
financial
55.0 51.2 50.0 45.1 43.7 41.5 38.5 33.1 30.7 30.9
Non-
financial
89.3 90.8 90.9 89.9 90.7 92.5 92.0 91.3 91.0 90.8
Primary
residence
63.9 63.9 64.7 66.2 67.7 69.1 68.6 67.3 65.2 63.7
Second+
home
13.1 12.7 11.8 12.8 11.3 12.5 13.8 14.3 13.2 13.8
Business 21.1 20.8 19.2 18.7 19.0 19.1 19.0 18.5 16.7 17.4
Other
nonfin.
84.5 86.4 85.0 83.6 85.5 87.0 87.7 87.2 86.8 85.8
Any debt 72.3 73.2 74.5 74.1 75.1 76.4 77.0 74.9 74.5 77.1
Mortgage 39.5 39.1 41.0 43.1 44.6 47.9 48.7 47.0 42.9 41.9
Mortg.
(2
nd
+
home)
5.1 5.7 4.7 5.0 4.6 4.0 5.5 5.3 5.2 5.6
Credit
card
balance
39.7 43.7 47.3 44.1 44.4 46.2 46.1 39.4 38.1 43.9
Other
installment
52.5 50.2 50.3 48.5 48.5 49.8 49.8 49.2 50.0 52.4

35
CDs: Certificate of deposit.
36
Directly held.
37
Directly held.
38
Directly held.
39
DC pension: Defined-contribution plan.
40
IRA: Individual retirement account.
117

Source: Bricker et al. (2019). Calculated from Federal Reserve Board’s Survey of Consumer
Finances (SCF).

118
Table 4.4. Mean Asset Holdings, 1989–2016 (thousands of 2016 dollars)
1989 1992 1995 1998 2001 2004 2007 2010 2013 2016
Any asset 395.3 365.8 388.9 486.6 612.2 672.0 758.3 659.5 645.3 787.1
Financial
asset
120.6 115.7 143.0 198.2 258.3 240.9 257.9 250.1 263.0 334.7
Liquid
asset
23.0 20.1 19.9 22.5 29.4 31.6 28.2 33.2 34.9 39.4
CDs
41
12.3 9.3 8.0 8.5 7.9 8.9 10.4 9.8 5.2 4.9
Mutual
42

fund
6.4 8.8 18.1 24.6 31.3 35.2 40.8 37.4 39.0 77.8
Stocks
43
18.1 19.0 22.3 44.9 55.5 42.1 45.9 35.1 41.7 45.7
Bonds
44
12.3 9.7 8.9 8.5 11.7 12.7 10.7 11.1 8.5 9.4
Managed
account
7.9 6.3 8.4 16.9 27.2 19.1 16.8 15.5 19.9 25.9
DC
pension
45

or IRA
46

26.0 29.9 40.5 55.1 74.9 78.1 90.4 95.4 102.2 119.2
Other
financial
14.8 12.5 16.9 17.2 20.5 13.4 14.9 12.6 11.6 12.4
Non-
financial
274.7 250.2 245.9 288.3 353.8 431.1 500.4 409.4 382.2 452.4
Primary
residence
126.3 117.5 116.7 135.5 165.8 216.8 240.4 194.4 176.4 191.9
Second+
home
22.4 21.2 19.6 24.5 28.6 42.6 53.5 45.8 43.2 49.5
Business 103.9 93.1 86.4 104.6 132.6 143.1 177.6 142.9 137.1 184.2
Other
nonfin.
22.2 18.4 23.2 23.6 26.8 28.6 28.9 26.4 25.5 26.9
Any debt 48.4 52.9 56.8 69.2 73.8 100.5 112.4 108.2 94.0 95.1
Mortgage 33.4 38.1 41.5 49.4 55.5 75.6 84.0 80.2 69.4 66.0
Mortg.
(2
nd
+
home)
3.7 5.4 4.3 5.2 4.6 8.5 11.4 10.6 8.4 9.0
Credit
card
balance
1.4 1.7 2.2 2.7 2.5 3.0 3.9 3.1 2.3 2.5
Other
installment
9.9 7.6 8.7 11.9 11.2 13.4 13.1 14.4 13.9 17.6

41
CDs: Certificate of deposit.
42
Directly held.
43
Directly held.
44
Directly held.
45
DC pension: Defined-contribution plan.
46
IRA: Individual retirement account.
119
Memo: net
worth
346.9 312.9 332.1 417.3 538.4 571.5 645.9 551.3 551.3 692.0

Source: Bricker et al. (2019). Calculated from Federal Reserve Board’s Survey of Consumer
Finances (SCF).

120
Table 4.5. Average Assets and Debts of Asset Groups, 1989–2016 (in thousands of 2016
dollars)
Asset
percentil
e
2016 2013 2010 2007 2004
asset debt asset debt asset debt asset debt asset debt
<25 9.0 12.4 8.4 9.8 9.1 9.3 11.3 8.9 11.5 9.3
25-50 96.9 41.8 91.9 41.1 108.4 51.9 140.2 53.7 120.7 47.8
50-60 227.6 94.9 212.9 97.8 234.8 105.
8
296.0 109.
7
257.8 99.2
60-70 327.9 111.
4
302.9 110.
6
327.9 136.
5
412.1 135.
3
380.2 130.
6
70-80 495.8 133.
7
456.0 139.
2
469.7 164.
1
601.5 164.
9
571.7 151.
1
80-90 935.3 173.
0
810.1 174.
5
831.3 166.
7
951.3 188.
3
914.9 166.
3
90-99 3191.2 264.
7
2627.2 257.
5
2740.1 317.
3
2996.0 335.
6
2529.5 251.
5
Top 1 27448.
8
643.
9
20523.
4
587.
2
19674.
0
703.
5
22403.
0
669.
6
19834.
5
879.
5
Average 787.1 95.1 645.3 94.0 659.5 108.
2
758.3 112.
4
672.0 100.
5

Asset
percentil
e
2016 2013 2010 2007 2004
asset debt asset debt asset asset debt asset debt
<25 11.5 7.3 9.6 7.4 9.2 6.5 7.0 4.9 5.2 4.5
25-50 108.3 40.0 100.2 34.7 85.7 26.2 72.1 20.8 70.7 21.0
50-60 227.7 78.4 204.5 69.4 175.7 53.9 159.6 49.4 164.5 47.7
60-70 333.3 84.3 284.0 88.9 242.4 79.6 227.8 68.2 235.3 58.7
70-80 493.4 114.
0
400.8 101.
3
325.8 86.6 317.1 70.7 348.4 64.9
80-90 814.7 124.
6
645.1 113.
3
510.1 104.
2
496.3 96.8 567.4 105.
6
90-99 2388.6 190.
3
1781.9 190.
0
1347.5 138.
6
1424.
3
159.
0
1546.8 126.
8
Top 1 17938.
4
463.
4
14516.
2
428.
4
11779.
5
366.
3
9757.
4
361.
6
10549.
3
289.
4
Average 612.2 73.8 486.6 69.2 388.9 56.8 365.8 52.9 395.3 48.4

Source: Bricker et al. (2019). Calculated from Federal Reserve Board’s Survey of Consumer
Finances (SCF).
121
Figure 4.3. Composition in Shares of Assets of the Wealthiest 1 Percent, 1989-2016

Source: Bricker and Volz (2019). Calculated from Federal Reserve Board’s Survey of Consumer
Finances (SCF).

0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
1989 1992 1995 1998 2001 2004 2007 2010 2013 2016
Equities Business Interest Housing Pensions
122
Figure 4.4. Private Equity Fundraising Activity in the United States, 2006-2019
Figure 4.4 (a). Buyout and Others
47
Fundraising Activity in the United States, 2006-2019

Source: “US PE Breakdown” (PitchBook 2020a).

47
Other PE includes growth/expansion, diversified private equity, mezzanine, and restructuring/turnaround.

0
50
100
150
200
250
300
$-
$50
$100
$150
$200
$250
$300
$350
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
Capital Raised ($B) Fund Count
123
Figure 4.4 (b). VC Fundraising Activity in the United States, 2006-2019

Source: “PitchBook-NVCA Venture Monitor” (PitchBook & NVCA 2020).

0
50
100
150
200
250
300
350
400
$-
$10
$20
$30
$40
$50
$60
$70
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
Capital raised ($B) Fund count
124
Figure 4.5. Private Equity Deal Activity in the United States, 2006-2019
Figure 4.5 (a). Buyout and Others
48
Deal Activity in the United States, 2006-2019

Source: “US PE Breakdown” (PitchBook 2020a).

48
Other PE includes growth/expansion, diversified private equity, mezzanine, and restructuring/turnaround.

0
1,000
2,000
3,000
4,000
5,000
6,000
$-
$100
$200
$300
$400
$500
$600
$700
$800
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
Deal value ($B) Deal count
125
Figure 4.5 (b). VC Deal Activity in the United States, 2006-2019

Source: “PitchBook-NVCA Venture Monitor” (PitchBook & NVCA 2020).

0
2000
4000
6000
8000
10000
12000
14000
$-
$20
$40
$60
$80
$100
$120
$140
$160
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
Deal value ($B) Deal count
126
Table 4.6. Households Qualifying under Existing Accredited Investor Criteria
Basic for
Qualifying
as
Accredited
Investor
1983 1989 2019
Number of
qualifying
households
Qualifying
households
as % of
U.S.
households
Number of
qualifying
households
Qualifying
households
as % of
U.S.
households
Number of
qualifying
households
Qualifying
households
as % of
U.S.
households
Individual
income
threshold
($200K)
0.4M 0.5% 4.3M 4.7% 11.2M 8.9%
Joint
income
threshold
($300K)
N/A N/A 2.1M 2.3% 5.8M 4.6%
Net worth
($1M)
1.2M 1.4% 4.5M 4.8% 11.8M 9.4%
Overall
number of
qualifying
households
1.3M 1.6% 6.8M 7.3% 16.0M 13.0%

Source: SEC “Amending the ‘Accredited Investor’ Definition” (2019). Calculated from SCF
data.

127
Table 4.7. Type and Frequency of Limited Partner
LP Type Freq. Percent Cum.
Foundation 647 17.02 17.02
Corporate Pension 631 16.60 33.61
Insurance Company 498 13.10 46.71
Public Pension Fund 308 8.10 54.81
Corporation 269 7.08 61.89
High-net-worth Investor 245 6.44 68.33
Fund of Funds 188 4.94 73.28
Union Pension Fund 156 4.10 77.38
Direct Investment 154 4.05 81.43
Endowment 135 3.55 84.98
Banking Institution 106 2.79 87.77
Money Management Firm 74 1.95 89.72
Private Investment Fund 68 1.79 91.50
Investment Advisor 48 1.26 92.77
Economic Development Agency 45 1.18 93.95
Wealth Management Firm 45 1.18 95.13
Other Limited Partners 43 1.13 96.27
Family Office (Single) 32 0.84 97.11
Real Estate Investment Company 29 0.76 97.87
Family Office (Multi) 25 0.66 98.53
Government Agency 25 0.66 99.18
Sovereign Wealth Fund 21 0.55 99.74
Secondary LP 9 0.24 99.97
University (Non-endowment) 1 0.03 100.00
Total 3,802 100.00

Data source: PitchBook.

128
Table 4.8. Average Private Equity Pay in the U.S.
Rank Base Salary Total annual
cash
compensation
Long-term
incentive/carries
interest
Total
remuneration
Analyst $86k $114k n/a n/a
Associate $107k $152k $45k $160k
Senior associate $127k $184k $65k $215k
Direct/principal $272k $428k $545k $850k
Managing
director/partner
$420k $668k $1,195k $1,623k

Source: efinancial career website. Data from Preqin.

129
Figure 4.6. Income Composition of Top Groups within the Top Decile in 1929 and 2007
Figure 4.6 (a). Income Composition of Top Groups within the Top Decile in 1929

Source: Piketty and Saez (2013).

0.000
0.100
0.200
0.300
0.400
0.500
0.600
0.700
0.800
0.900
1.000
P90-95 P95-99 P99-99.5 P99.5-99.9 P99.9-99.99 P99.99-100
Wage Income Capital Income Entrepreneurial Income
130
Figure 4.6 (b). Income Composition of Top Groups within the Top Decile in 2007

Source: Piketty and Saez (2013).

0.000
0.100
0.200
0.300
0.400
0.500
0.600
0.700
0.800
0.900
1.000
P90-95 P95-99 P99-99.5 P99.5-99.9 P99.9-99.99 P99.99-100
Wage Income Capital Income Entrepreneurial Income
131
Figure 4.7. Number of U.S. Listings, 1980-2018

Data Source: The World Bank Database.

4000
4500
5000
5500
6000
6500
7000
7500
8000
8500
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
132
Figure 4.8. U.S. Unicorn Count and Aggregated Post-valuation ($B), 2006-2018

Data Source: “Unicorn Report” (Pitchbook 2018)

$0
$100
$200
$300
$400
$500
$600
0
20
40
60
80
100
120
140
160
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Cumulative Unicorns (#) Aggregate Unicorn Post-Valuations ($B)
133
Figure 4.9. U.S. VC Pre-money Valuation ($M) by stage, 2006-2019
Figure 4.9 (a). U.S. VC Median Pre-money Valuation ($M) by stage, 2006-2019

Data Source: “PitchBook-NVCA Venture Monitor” (PitchBook & NVCA 2020)

$0
$10
$20
$30
$40
$50
$60
$70
$80
$90
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
Angel Seed Early VC Late VC
134
Figure 4.9 (a). U.S. VC Average Pre-money Valuation ($M) by stage, 2006-2019

Data Source: “PitchBook-NVCA Venture Monitor” (PitchBook & NVCA 2020)

$0
$50
$100
$150
$200
$250
$300
$350
$400
$450
$500
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
Angel Seed Early VC Late VC
135
Figure 4.10. Quartile Breakdown for Valuation at Exit for U.S. VC, 2006-2019
Figure 4.10 (a). Quartile Breakdown for Valuation at Exit via IPO for U.S. VC, 2006-2019

Data Source: “US VC Valuations Report” (PitchBook 2020)

$0
$500
$1,000
$1,500
$2,000
$2,500
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
75th Percentile Median 25th Percentile Average
136
Figure 4.10 (b). Quartile Breakdown for Valuation at Exit via Aquisition for U.S. VC,
2006-2019

Data Source: “US VC Valuations Report” (PitchBook 2020)

$0
$50
$100
$150
$200
$250
$300
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
75th Percentile Median 25th Percentile Average
137
References
Abadie, Alberto, Alexis Diamond, and And Jens Hainmueller. 2010. “Synthetic Control Methods
for Comparative Case Studies: Estimating the Effect of California’s Tobacco Control
Program.” Journal of the American Statistical Association 105(490):493–505.
Aghion, P., and P. Howitt. 1994. “Growth and Unemployment.” The Review of Economic Studies
61(3):477–94.
Almus, Matthias, and Dirk Czarnitzki. 2003. “The Effects of Public R and D Subsidies on Firms’
Innovation Activities: The Case of Eastern Germany.” Journal of Business and Economic
Statistics 21(2):226–36.
Alvaredo, F., L. Chancel, T. Piketty, E. Saez, and G. Zucman. 2018. “World Inequality Report
2018 Executive Summary.” World Inequality Lab 20.
Ashenfelter, Orley, Daniel Hosken, Michael Vita, and Matthew Weinberg. 2011. “Retrospective
Analysis of Hospital Mergers.” International Journal of the Economics of Business 18(1):5–
16.
Atkinson, Anthony B., Thomas Piketty, and Emmanuel Saez. 2011. “Top Incomes in the Long
Run of History.” Journal of Economic Literature 49(1):3–71.
Auken, Howard E. Van, and Lynn Neeley. 1996. “Evidence of Bootstrap Financing among Small
StartUp Firms.” Journal of Entrepreneurial and Small Business Finance.
Auten, Gerald, and David Splinter. 2019. “Income Inequality in the United States: Using Tax
Data to Measure Long-Term Trends.” Working Paper 1–43.
Avery, Robert B., Daniel J. Grodzicki, and Kevin B. Moore. 2015. “Death and Taxes: An
Evaluation of the Impact of Prospective Policies for Taxing Wealth at the Time of Death.”
National Tax Journal 68(3):601–32.
138
Baker, Jonathan B. 2011. “Comcast/NBCU: The FCC Provides a Roadmap for Vertical Merger
Analysis.” Antitrust 25(2).
Baptista, Rui, Murat Karaöz, and Joana Mendonça. 2014. “The Impact of Human Capital on the
Early Success of Necessity versus Opportunity-Based Entrepreneurs.” Small Business
Economics 42(4):831–47.
Becker, Gary S. 1964. Human Capital: A Theoretical and Empirical Analysis, with Special
Reference to Education. Vol. 43. Columbia University Press.
Beckman, Christine M. 2006. “The Influence of Founding Team Company Affiliations on Firm
Behavior.” Academy of Management Journal 49(4):741–58.
Bellavitis, Cristiano, Igor Filatotchev, Dzidziso Samuel Kamuriwo, and Tom Vanacker. 2017.
“Entrepreneurial Finance: New Frontiers of Research and Practice: Editorial for the Special
Issue Embracing Entrepreneurial Funding Innovations.” Venture Capital 19(1–2):1–16.
Belz, Andrea, Richard J. Terrile, Fernando Zapatero, Michael Kawas, and Aleksandar Giga.
2019. “Mapping the ‘Valley of Death’: Managing Selection and Technology Advancement
in NASA’s Small Business Innovation Research Program.” IEEE Transactions on
Engineering Management.
Belz, Andrea, and Fernando Zapatero. 2019. “Impact of Government Grants on Venture Capital
Funding of Deep Technology University Spinoffs.” Working Paper.
Bender, Kevin G. 2015. “Giving the Average Investor the Keys to the Kingdom : How the
Federal Securities Laws Facilitate Wealth Inequality.” Journal of Business & Security Law
15(2).
Berry, Steven, James Levinsohn, and Ariel Pakes. 1995. “Automobile Prices in Market
Equilibrium.” Econometrica 63(4):841–90.
139
Bertoni, Fabio, Massimo G. Colombo, and Luca Grilli. 2011. “Venture Capital Financing and the
Growth of High-Tech Start-Ups: Disentangling Treatment from Selection Effects.”
Research Policy 40(7):1028–43.
Block, Joern H., Massimo G. Colombo, Douglas J. Cumming, and Silvio Vismara. 2018. “New
Players in Entrepreneurial Finance and Why They Are There.” Small Business Economics
50(2):239–50.
Braun, Reiner, Tim Jenkinson, and Ingo Stoff. 2015. “How Persistent Is Private Equity
Performance? Evidence from Deal-Level Data.” SSRN Electronic Journal.
Bricker, Jesse, Alice M. Henriques, Jake A. Krimmel, and John E. Sabelhaus. 2016. “Measuring
Income and Wealth at the Top Using Administrative and Survey Data.” Finance and
Economics Discussion Series 2015(030):1–63.
Bricker, Jesse, Kevin B. Moore, and Jeffrey P. Thompson. 2019. “Trends in Household Portfolio
Composition.” Finance and Economics Discussion Series 2019(069):1–51.
Bricker, Jesse, and Alice Henriques Volz. 2019. “Why Has Wealth Concentration Grown in the
United States ? A Re-Examination of Data from 1998-2016.” 1–36.
Brush, Candida G., Linda F. Edelman, and Tatiana S. Manolova. 2012. “Ready for Funding?
Entrepreneurial Ventures and the Pursuit of Angel Financing.” Venture Capital 14(2–
3):111–29.
CalPERS. 2019. “Private Equity Program Fund Performance Review.”
Card, David. 2009. “Immigration and Inequality.” NBER Working Paper Series.
Carter, Nancy M., Mary Williams, and Paul D. Reynolds. 1997. “Discontinuance Among New
Firms in Retail: The Influence of Initial Resources, Strategy, and Gender.” Journal of
Business Venturing.
140
Carter, Richard B., and Howard E. Van Auken. 2005. “Bootstrap Financing and Owners’
Perceptions of Their Business Constraints and Opportunities.” Entrepreneurship and
Regional Development 17(2):129–44.
Cassar, Gavin. 2004. “The Financing of Business Start-Ups.” Journal of Business Venturing
19(2):261–83.
Chandler, Gaylen N., and Steven H. Hanks. 1998. “An Examination of the Substitutability of
Founders Human and Financial Capital in Emerging Business Ventures.” Journal of
Business Venturing 13(5):353–69.
Choi, Joonkyu, Nathan Goldschlag, John C. Haltiwanger, and J. Daniel Kim. 2019. “Founding
Teams and Startup Performance.” SSRN Electronic Journal.
Cingano, Federico. 2014. “Trends in Income Inequality and Its Impact on Economic Growth.”
OECD Social, Employment, and Migration Working Papers (163):0_1,5-59.
Cochrane, John H. 2005. “The Risk and Return of Venture Capital.” Journal of Financial
Economics 75(1):3–52.
Cole, C. W. 2009. “Financing an Entrepreneurial Venture: Navigating the Maze of Corporate,
Securities, and Tax Law.” University of Missouri-Kansas City Law Review 78(1):473–502.
Colombo, Massimo G., and Luca Grilli. 2005. “Founders’ Human Capital and the Growth of
New Technology-Based Firms: A Competence-Based View.” Research Policy 34(6):795–
816.
Conti, Annamaria, Jerry Thursby, and Marie Thursby. 2013. “Patents as Signals for Startup
Financing.” Journal of Industrial Economics 61(3):592–622.
Cooney, Thomas M. 2005. “Editorial: What Is an Entrepreneurial Team?” International Small
Business Journal 23(3):226–35.
141
Cooper, Arnold C., F. Javier Gimeno-Gascon, and Carolyn Y. Woo. 1994. “Initial Human and
Financial Capital as Predictors of New Venture Performance.” Journal of Business
Venturing 9(5):371–95.
Crawford, Gregory S., Robin S. Lee, Michael D. Whinston, and Ali Yurukoglu. 2018. “The
Welfare Effects of Vertical Integration in Multichannel Television Markets.” Econometrica
86:891–954.
Croce, Annalisa, José Martí, and Samuele Murtinu. 2013. “The Impact of Venture Capital on the
Productivity Growth of European Entrepreneurial Firms: ‘Screening’ or ‘Value Added’
Effect?” Journal of Business Venturing 28(4):489–510.
Davidsson, Per, and Benson Honig. 2003. “The Role of Social and Human Capital among
Nascent Entrepreneurs.” Journal of Business Venturing 18(3):301–31.
Denis, David J. 2004. “Entrepreneurial Finance: An Overview of the Issues and Evidence.”
Journal of Corporate Finance 10(2):301–26.
Denoo, Lien, Anneleen Van Boxstael, and Andrea Belz. 2019. “Help, I Need Somebody, Not
Just Anybody? The Role of Advisors During Technology Venture Emergence.” Academy of
Management Proceedings 2019(1):11928.
Dimov, Dimo. 2010. “Nascent Entrepreneurs and Venture Emergence: Opportunity Confidence,
Human Capital, and Early Planning.” Journal of Management Studies 47(6):1123–53.
Doidge, Craig, G. Andrew Karolyi, and René M. Stulz. 2013. “The U.S. Left behind? Financial
Globalization and the Rise of IPOs Outside the U.S.” Journal of Financial Economics
110(3):546–73.
Doidge, Craig, G. Andrew Karolyi, and René M. Stulz. 2017. “The U.S. Listing Gap.” Journal of
Financial Economics 123(3):464–87.
142
Dunne, Timothy C., Brent B. Clark, John P. Berns, and William C. McDowell. 2019. “The
Technology Bias in Entrepreneur-Investor Negotiations.” Journal of Business Research
105(August):258–69.
Dyer, Jeffrey H., Hal B. Gregersen, and Clayton Christensen. 2008. “Entrepreneur Behaviors,
Opportunity Recognition, and The Origins of Innovation Ventures.” Strategic
Entrepreneurship Journal 306(2008):285–306.
Ensley, Michael D., and Craig L. Pearce. 2001. “Shared Cognition in Top Management Teams:
Implications for New Venture Performance.” Journal of Organizational Behavior
22(2):145–60.
Ewens, Michael, and Joan Farre-Mensa. 2020. “The Evolution of the Private Equity Market and
the Decline in IPOs.” Review of Financial Studies, Forthcoming.
Ewens, Michael, Charles M. Jones, and Matthew Rhodes-Kropf. 2013. “The Price of
Diversifiable Risk in Venture Capital and Private Equity.” Review of Financial Studies
26(8):1853–89.
Ewens, Michael, and Matthew Rhodes-Kropf. 2015. “Is a VC Partnership Greater Than the Sum
of Its Partners?” Journal of Finance 70(3):1081–1113.
Farrell, Joseph, Paul A. Pautler, and Michael G. Vita. 2009. “Economics at the FTC:
Retrospective Merger Analysis with a Focus on Hospitals.” Review of Industrial
Organization 35(4):369–85.
Federal Communications Commission. 1992. “Cable Television Consumer Protection and
Competition Act of 1992.” Public Law 102-385.
Ford, George S. 2018. “A Retrospective Analysis of Vertical Mergers in Multichannel Video
Programming Distribution Markets: The Comcast-NBCU Merger.” SSRN Electronic
143
Journal (December):1–23.
Ford, Stuart. 2019. “The Need for a Wealth Inequality Amendment.” West Virginia Law Review
122(1):4.
Ganotakis, Panagiotis. 2012. “Founders’ Human Capital and the Performance of UK New
Technology Based Firms.” Small Business Economics 39(2):495–515.
Gao, Xiaohui, Jay R. Ritter, and Zhongyan Zhu. 2013. “Where Have All the IPOs Gone ?” The
Journal of Financial and Quantitative Analysis 48(6):1663–92.
Giga, Aleksandar. 2019. “Founders’ Commitment and Firm Financing: Multiple ‘Skins in the
Game’ as a Signal to Investors.” SSRN Electronic Journal (February).
Gilbert, Geoffrey. 1997. “Adam Smith on the Nature and Causes of Poverty.” Review of Social
Economy 55(3):273–91.
Gompers, Paul, Anna Kovner, Josh Lerner, and David Scharfstein. 2008. “Venture Capital
Investment Cycles: The Impact of Public Markets.” Journal of Financial Economics
87(1):1–23.
Grossman, Sanford J., and Oliver D. Hart. 1986. “The Costs and Benefits of Ownership: A
Theory of Vertical and Lateral Integration.” Journal of Political Economy 94(4):691–719.
Hainmueller, Jens. 2012. “Entropy Balancing for Causal Effects: A Multivariate Reweighting
Method to Produce Balanced Samples in Observational Studies.” Political Analysis
20(1):25–46.
Haleblian, Jerayr, and Sydney Finkelstein. 1993. “Top Management Team Size , CEO
Dominance , and Firm Performance : The Moderating Roles of Environmental Turbulence
and Discretion Author ( s ): Jerayr Haleblian and Sydney Finkelstein Source : The Academy
of Management Journal , Vol . 36 , No . 4 ( Aug .,.” The Academy of Management
144
36(4):844–63.
Haltiwanger, John. 2015. “Job Creation, Job Destruction, and Productivity Growth: The Role of
Young Businesses.” Annual Review of Economics 7(1):341–58.
Haltiwanger, John, Ron S. Jarmin, and Javier Miranda. 2013. “Who Creates Jobs? Small Versus
Large Versus Young.” The Review of Economics and Statistics 85(2):298–306.
Harris, Robert S., Tim Jenkinson, and Steven N. Kaplan. 2014. “Private Equity Performance:
What Do We Know?” Journal of Finance 69(5):1851–82.
Harrison, Richard, and Colin Mason. 2017. “Backing the Horse or the Jockey? Due Diligence,
Agency Costs, Information and the Evaluation of Risk by Business Angel Investors.”
International Review of Entrepreneurship 15(3):269–90.
Hart, Oliver, and Jean Tirole. 1990. “Vertical Integration and Market Foreclosure.” Brookings
Papers on Eco- Nomic Activity: Microeconomics.
Heather Boushey. 2019.“Hearing on ‘Measuring Economic Inequality in the United States.’”
Helmers, Christian, Mark Rogers, and Alan Gilbert Building. 2009. “Does Patenting Help High-
Tech Start-Ups ? Does Patenting Help High-Tech Start-Ups ?” 61(0):0–25.
Helslop, Louise A., Eileen McGregor, and May Griffith. 2001. “Development of a Technology
Readiness Assessment Measure: The Cloverleaf Model of Technology Transfer.” Journal of
Technology Transfer 103(3):239–48.
Holtz-Eakin, Douglas. 2019.“Hearing to Discuss Controversy Over Measuring Economic
Inequality.”
Hosken, Daniel S., and Christopher T. Taylor. 2007. “The Economic Effects of The Marathon-
Ashland Joint Venture : The Importance of Industry Supply Shocks and Vertical Market
Structure.” The Journal of Industrial Economics LV(3):419–51.
145
Huyghebaert, Nancy, and Linda M. Van De Gucht. 2007. “The Determinants of Financial
Structure: New Insights from Business Start-Ups.” European Financial Management
13(1):101–33.
Hwang, Min, John M. Quigley, and Susan E. Woodward. 2005. “An Index For Venture Capital,
1987-2003.” Contributions in Economic Analysis & Policy 4(1).
Islam, Mazhar, Adam Fremeth, and Alfred Marcus. 2018. “Signaling by Early Stage Startups:
US Government Research Grants and Venture Capital Funding.” Journal of Business
Venturing 33(1):35–51.
Jasso, Guillermina, and Eva M. Meyersso. Milgrom. 2008. “Distributive Justice and CEO
Compensation.” Acta Sociologica 51(2):123–43.
Jay Barney. 1991. “Frim Resources and Sustained Competitive Advantage.” Journal of
Management.
Jehn, Karen A. 1997. “A Qualitative Analysis of Conflict Types and Dimensions in
Organizational Groups.” 42(3):530–57.
Jin, Linlin, Kristen Madison, Nils D. Kraiczy, Franz W. Kellermanns, T. Russell Crook, and Jing
Xi. 2017. “Entrepreneurial Team Composition Characteristics and New Venture
Performance: A Meta-Analysis.” Entrepreneurship: Theory and Practice 41(5):743–71.
de Jong, Ad, Michael Song, and Lisa Z. Song. 2013. “How Lead Founder Personality Affects
New Venture Performance: The Mediating Role of Team Conflict.” Journal of Management
39(7):1825–54.
Kaplan, Steven N., and Joshua Rauh. 2013. “It’s the Market: The Broad-Based Rise in the
Return to Top Talent.” Journal of Economic Perspectives 27(3):1–16.
Kaplan, Steven N., and Antoinette Schoar. 2005. “Private Equity Performance: Returns,
146
Persistence, and Capital Flows.” Journal of Finance 60(4):1791–1823.
Kaplan, Steven N., Berk A. Sensoy, and Per StrÖmberg. 2009. “Should Investors Bet on the
Jockey or the Horse? Evidence from the Evolution of Firms from Early Business Plans to
Public Companies.” Journal of Finance 64(1):75–115.
Kartashova, Katya. 2014. “Private Equity Premium Puzzle Revisited.” American Economic
Review 104(10):3297–3334.
Kerr, William, and Ramana Nanda. 2009. “Financing Constraints and Entrepreneurship.” NBER
Working Paper Series 1–44.
Klotz, Anthony C., Keith M. Hmieleski, Bret H. Bradley, and Lowell W. Busenitz. 2014. “New
Venture Teams: A Review of the Literature and Roadmap for Future Research.” Journal of
Management 40(1):226–55.
Kopczuk, Wojciech, and Emmanuel Saez. 2004. “Top Wealth Shares in the United States, 1916-
2000: Evidence From Estate Tax Returns.” NBER Working Paper Series Working Paper
Series 1916–2000.
Korteweg, Arthur, and Morten Sorensen. 2010. “Risk and Return Characteristics of Venture
Capital-Backed Entrepreneurial Companies.” Review of Financial Studies 23(10):3738–72.
Korteweg, Arthur, and Morten Sorensen. 2017. “Skill and Luck in Private Equity Performance.”
Journal of Financial Economics 124(3):535–62.
Kuznets, Simon. 1955. “Economic Growth and Income Inequality.” The American Economic
Review 31(2):386–408.
Kwoka, John. 2015. “Mergers, Merger Control, and Remedies: A Retrospective Analysis of U.S.
Policy.”
Kwoka, John, and Evgenia Shumilkina. 2010. “The Price Effect of Eliminating Potential
147
Competition: Evidence from an Airline Merger.” Journal of Industrial Economics
58(4):767–93.
Leigh, Andrew. 2007. “How Closely Do Top Income Shares Track Other Measures of
Inequality ?” The Economic Journal 117(524).
Lerner, Josh, Antoinette Schoar, and Wan Wongsunwai. 2007. “Smart Institutions, Foolish
Choices: The Limited Partner Performance Puzzle.” Journal of Finance 62(2):731–64.
Levy, Ari. 2018. “Venture Firm Sequoia Is Building a $12 Billion War Chest after SoftBank
Raises Stakes.” CNBC, 1–7.
Liao, Jianwen, Harold Welsch, and Chad Moutray. 2008. “Start-Up REsources and
Entrepreneurial Discontinuance: The Case of Nascent Entrepreneurs.” Journal of Small
Business Strategy 17(2):89–103.
Lindsey, Brink, and Steven M. Teles. 2017. The Captures Economy: How the Powerful Enrich
Themselves, Slow Down Growth, and Increase Inequality. Oxford University Press.
Ljungqvist, Alexander, and Matthew P. Richardson. 2003. “The Cash Flow, Return and Risk
Characteristics of Private Equity.” SSRN Electronic Journal.
Lopez-de-Silanes, Florencio, Ludovic Phalippou, and Oliver Gottschalg. 2010. “Giants at the
Gate: On the Cross-Section of Private Equity Investment Returns.”
Macarthur, Hugh, Josh Lerner, State Street Global Markets, and State Street Private Equity
Index. 2020. Public vs . Private Equity Returns : Is PE Losing Its Advantage ?
Mankins, John C. 1995. “Technology Readiness Level.” White Paper.
Mankins, John C. 2009. “Technology Readiness Assessments: A Retrospective.” Acta
Astronautica 65(9–10):1216–23.
Markova, Sonja, and Tatjana Petkovska-Mircevska. 2009. “Financing Options for
148
Entrepreneurial Ventures.” Economic Interferences XI(26):597–605.
Marquez, Robert, Vikram Nanda, and M. Deniz Yavuz. 2015. “Private Equity Fund Returns and
Performance Persistence.” Review of Finance 19(5):1783–1823.
Mathieu, John, Travis M. Maynard, Tammy Rapp, and Lucy Gilson. 2008. “Team Effectiveness
1997-2007: A Review of Recent Advancements and a Glimpse into the Future.” Journal of
Management 34(3):410–76.
Minola, Tommaso, and Marco Giorgino. 2008. “Who’s Going to Provide the Funding for High
Tech Start-Ups? A Model for the Analysis of Determinants with a Fuzzy Approach.” R and
D Management 38(3):335–51.
Mishel, Lawrence, and Natalie Sabadish. 2012. “CEO Pay and The Top 1%: How Executive
Compensation and Financial-Sector Pay Have Fueled Income Ineqaulity.” Economic Policy
Institue 1–7.
Morrissette, Stephen G. 2007. “A Profile of Angel Investors.” The Journal of Private Equity
10(3):52–66.
Muñoz-Bullon, Fernando, Maria J. Sanchez-Bueno, and Antonio Vos-Saz. 2015. “Startup Team
Contributions and New Firm Creation: The Role of Founding Team Experience.”
Entrepreneurship and Regional Development 27:80–105.
Mustapha, Achibane, and Jamal Tlaty. 2018. “The Entrepreneurial Finance and the Issue of
Funding Startup Companies.” European Scientific Journal, ESJ 14(13):268.
Nelson, Teresa. 2003. “The Persistence of Founder Influence: Management, Ownership, and
Performance Effects at Initial Public Offering.” Strategic Management Journal 24(8):707–
24.
Nicolaou, Nicos, Scott Shane, Lynn Cherkas, and Tim D. Spector. 2009. “Opportunity
149
Recognition and the Tendency to Be an Entrepreneur: A Bivariate Genetics Perspective.”
Organizational Behavior and Human Decision Processes 110(2):108–17.
NSF. 2020. “Directory of I-Corps Nodes and Sites.” 1–2. Retrieved
(https://www.nsf.gov/news/special_reports/i-corps/about.jsp).
Oranburg, S. C. 2016. “Start-up Financing.” Start-Up Creation: The Smart Eco-Efficient Built
Environment 57–73.
Ordover, Janusz A., Garth Saloner, and Steven C. Salop. 1990. “Equilibrium Vertical
Foreclosure.” American Economic Review 80(1):127–42.
Ozgen, Eren, and Robert A. Baron. 2007. “Social Sources of Information in Opportunity
Recognition: Effects of Mentors, Industry Networks, and Professional Forums.” Journal of
Business Venturing 22(2):174–92.
Peters, Craig. 2006. “Evaluating the Performance of Merger Simulation: Evidence from the U.S.
Airline Industry.” Journal of Law and Economics 49(2):627–49.
Phalippou, Ludovic. 2010. “Venture Capital Funds: Flow-Performance Relationship and
Performance Persistence.” Journal of Banking and Finance 34(3):568–77.
Phalippou, Ludovic, and Oliver Gottschalg. 2009. “The Performance of Private Equity Funds.”
Review of Financial Studies 22(4):1747–76.
Phalippou, Ludovic, and Maurizio Zollo. 2006. “What Drives Private Equity Fund
Performance?” Working Papers -- Financial Institutions Center at The Wharton School
(November):1–29.
Piketty, Thomas. 2014. Capital in the 21 St Century. The Belknap Press of Harvard University
Press.
Piketty, Thomas. 2020. Capital and Ideology.
150
Piketty, Thomas, and Emmanuel Saez. 2003. “Income Inequality in The United States, 1913–
1998.” The Quarterly Journal of Economics.
Piketty, Thomas, and Emmanuel Saez. 2013. “Top Incomes and the Great Recession: Recent
Evolutions and Policy Implications.” IMF Economic Review 61(3):456–78.
Piketty, Thomas, Emmanuel Saez, and Stefanie Stantcheva. 2014. “Optimal Taxation of Top
Labor Incomes: A Tale of Three Elasticities.” American Economic Journal: Economic
Policy 6(1 B):230–71.
Piketty, Thomas, Emmanuel Saez, and Gabriel Zucman. 2018. “Distributional National
Accounts: Methods and Estimates for the United States.” The Quarterly Journal of
Economics 133(2):553–609.
Piketty, Thomas, Emmanuel Saez, and Gabriel Zucman. 2019. “Simplified Distributional
National Accounts.” WID.World Working Paper 109(January):289–95.
Pitchbook. 2018. “UNICORN Report.”
PitchBook. 2020a. US PE Breakdown.
PitchBook. 2020b. “US VC Valuations Report.”
PitchBook & NVCA. 2019. Venture Monitor Q4 2019.
PitchBook & NVCA. 2020. PitchBook-NVCA Venture Monitor.
Poschke, Markus. 2013. “Who Becomes an Entrepreneur? Labor Market Prospects and
Occupational Choice.” Journal of Economic Dynamics and Control 37(3):693–710.
Rasmussen, Dennis C. 2016. “The Problem With Inequality , According to Adam Smith.” The
Atlantic 1–15.
Robb, Alicia M., and David T. Robinson. 2014. “The Capital Structure Decisions of New
Firms.” Review of Financial Studies 27(1):153–79.
151
Rocha, Vera, Anabela Carneiro, and Celeste Amorim Varum. 2015. “Serial Entrepreneurship,
Learning by Doing and Self-Selection.” International Journal of Industrial Organization
40:91–106.
Roper, Stephen. 1998. “Entrepreneurial Characteristics, Strategic Choice and Small Business
Performance.” Small Business Economics 11(1):11–24.
Rossi, Matteo. 2016. “The Impact of Age on Firm Performance: A Literature Review.”
Corporate Ownership and Control 13(2CONT1):217–23.
Saez, Emmanuel, and Gabriel Zucman. 2016. “Wealth Inequality in the United States Since
1913: Evidence from Capitalized Income Tax Data.” The Quarterly Journal of Economics
131(May):269–76.
Saez, Emmanuel, and Gabriel Zucman. 2019. The Triumph of Injustice: : How the Rich Dodge
Taxes and How to Make Them Pay. W. W. Norton & Company.
Sah, Raaj Kumar, and Joseph E. Stiglitz. 1986. “The Architecture of Economic Systems:
Hierarchies and Polyarchies.” The American Economic Review 76(4):716–27.
Scherer, Robert, James Brodzinski, and Frank Wiebe. 1990. “Entrepreneur Career Selection and
Gender: A Socialization Approach.” Journal of Small Business Management 28(2):37.
Scherr, Frederick C., Timothy F. Sugrue, and Janice B. Ward. 1993. “Financing the Small Firm
Start-Up: Determinants of Debt Use.” The Journal of Entrepreneurial Finance 3(1):17–36.
Scigliuzzo, Davide, Kelsey Butler, and Sally Bakewell. 2019. “Everything Is Private Equity
Now.” Bloomberg Businessweek, 1–16.
Securities and Exchange Commission. 2019. “Concept Release on Harmonization of Securities
Offering Exemptions (Conformed to Federal Register Version).”
Securities, U. S., and Exchange Commission. 2011. “Net Worth Standard for Accredited
152
Investors.” 413(6).
De Silva, Muthu, Jeremy Howells, and Martin Meyer. 2018. “Innovation Intermediaries and
Collaboration: Knowledge–Based Practices and Internal Value Creation.” Research Policy
47(1):70–87.
Smith, Matthew, Owen M. Zidar, and Eric Zwick. 2019. “Top Wealth in the United States: New
Estimates and Implications for Taxing the Rich.” Working Paper (1752431).
Solow, Robert M. 1957. “Technical Change and the Aggregate Production Function.” The
Review of Economics and Statistics 39(3):312–20.
Solow, Robert M. .. 1956. “A Contribution to the Theory of Economic Growth.” The Quarterly
Journal of Economics 70(1):65–94.
Spengler, Joseph J. .. 1950. “Vertical Integration and Antitrust Policy.” Journal of Political
Economy 58(4):347–52.
Stiglitz, Joseph E. 2012. The Price of Inequality: How Today’s Divided Society Endangers Our
Future. Vol. 27. W. W. Norton & Company.
Strauss, Daniel. 2019. “Here’s How Every New IPO Bigger than $1 Billion Has Performed in
2019.” 500:1–11.
Thomas, Jeff. 2019. “Redefining Accredited Investor: That’s One Small Step for The SEC, One
Giant Leap for Our Economy.” Mich. Bus. & Entrepreneurial L. Rev., Forthcoming 1–21.
Thornhill, Stewart, and Raphael Amit. 2003. “Learning about Failure: Bankruptcy, Firm Age,
and the Resource-Based View.” Organization Science 14(5):497–509.
Toft-Kehler, Rasmus, Karl Wennberg, and Phillip H. Kim. 2014. “Practice Makes Perfect:
Entrepreneurial-Experience Curves and Venture Performance.” Journal of Business
Venturing 29(4):453–70.
153
Tzabbar, Daniel, and Jaclyn Margolis. 2017. “Beyond the Startup Stage: The Founding Team’s
Human Capital, New Venture’s Stage of Life, Founder-CEO Duality, and Breakthrough
Innovation.” Organization Science 28(5):857–72.
United States District Court for The District of Columbia. 2017.“Case 1:17-Cv-02511 Document
1.”
Wernerfelt, Birger. 1984. “A Resource-Based View of the Firm.” Strategic Management Journal
5(2):171–80.
Xu, Yiqing. 2017. “Generalized Synthetic Control Method: Causal Inference with Interactive
Fixed Effects Models.” Political Analysis 25(1):57–76.
Ye, Qian. 2017. “Bootstrapping and New-Born Startups Performance: The Role of Founding
Team Human Capital.” Global Journal of Entrepreneurship 1(2):53–71.
Zucman, Gabriel. 2019. “Global Wealth Inequality.” Annual Review of Economics 11(1):109–38.

154
Appendices
Appendices for Chapter 1
Appendix 1-A. Definition of Variables
Variable Definition
Affiliate revenue per Avg Sub/ Month Average affiliate revenue per average
subscriber per month
Inner market subscribers Number of inner market subscribers
attributable to the network
Outer market subscribers Number of outer market subscribers
attributable to the network
Network satellite subscribers Number of direct broadcast satellite
subscribers attributable to the network
Subscribers Number of subscribers attributable to the
network. For Pay networks, this figure
represents Total Combined Units (a
summation of inner market, outer market, and
satellite subscribers)
Average subscribers Average number of subscribers attributable to
the network. For Pay networks, this figure
represents total combined unit
Affiliate revenue Affiliate revenue from multichannel providers
Gross advertising revenue Revenue from carrying advertisements, before
expenses of ad agency
Net advertising revenue Revenue from carrying advertisements, after
expenses of ad agency
Other operating revenue Operating revenue derived from all other
sources
Operating revenue, net All operating revenue associated with the
business (a summation of affiliate revenue,
net advertising revenue and other operating
revenue, we use as total revenue)
Programming expenses Direct cost of creating, acquiring and
distributing content and services
Operating SG&A expense Selling, general and administrative expenses
incurred in the normal operating businesses of
the company
Total operating expenses All operating expenses associated with the
business (a summation of programming
expenses operating SG&A expenses)
Cash flow Revenue less SG&A and programming,
excludes depreciation of PP&E and
amortization of goodwill
Cash flow margin Percent of revenue attributable to cash flow
155
Profit Affiliate revenue minus programming
expenses
Bargaining power Programming expenses divided by operating
revenue, net

156
Appendix 1-B. Compare Two Sets of Treatment Effects
The unique structure of the U.S. television industry makes the regulation on
discriminative pricing being only effective on cable distributors
49
. Therefore, the observed price
effects after vertical integration between a cable distributor and a content provider is an
aggregate effect of both vertical integration and regulation. In contrast, the observed price effects
after vertical integration between a satellite distributor and a content provider is an effect purely
from vertical integration. Identification in this paper relies on comparing these two kinds of
treatment effects.
We will show price effects through graphs to illustrate how we use the comparison to
draw conclusions. Let’s first assume for a cable; proportion 𝛼 is competitor, (1−𝛼) is non-
competitor. Optimally, it would want to increase 𝑎 to competitor, decrease 𝑏 to non-competitor.
Then the observed average change of price would be 𝛼∗𝑎+(1−𝛼)∗(−𝑏). If 𝛼∗𝑎 is greater
than (1−𝛼)∗(−𝑏), prices increase, if 𝛼∗𝑎 is less than (1−𝛼)∗(−𝑏), prices decrease.
Figure 1-B-1 illustrates this point.
Under the scenario that PAR is effective, price after vertical integration should therefore
polarizes to the one either for competitor or non-competitor. Assume the weight that cable put on
its competitor is 𝑤
(,*
, weight on its non-competitor is 𝑤
'(
. If 𝑤
(,*
is greater than 𝑤
'(
, we
would observe price polarized to that for cable’s competitor. If 𝑤
(,*
is less than 𝑤
'(
, we would
observe price polarized to that for cable’s non-competitor. These two price levels are shown by
the upper and lower lines in Figure 1-B-2. For satellite distributors, the observed average price
will always increase since every distributor in the market is a competitor. This is shown in Figure
1-B-3.

49
Only cable distributors have both competitors and non-competitors.
157
Under this setting, comparison between two treatment effects from vertical integration
(one with PAR, one without PAR) becomes clear: we could regard effects after satellite’s
vertical integration as a baseline, if observed average price increases more in the case that a
channel is acquired by a cable distributor than the case that a channel is acquired by a satellite
distributor, then PAR harms consumer welfare and vice versa. These are shown in figures 1-B-4
and 1-B-5.
Figure 1-B-1. Average Affiliate Fee—Cable, No PAR

158
Figure 1-B-2. Average Affiliate Fee—Cable under PAR

159
Figure 1-B-3. Average Affiliate Fee—Satellite

160
Figure 1-B-4. Average Affiliate Fee—Comparison 1

161
Figure 1-B-5. Average Affiliate Fee—Comparison 2

162
Appendix 1-C. Synthetic Results
Table 1-C-1. Unit Weights for “The Travel Channel”
Network Unit Weight
A&E (US) 0.680
Animal Planet (US) 2.032
Antena 3 (US) 4.997
BET Her (US) 1.978
Bloomberg Television (US) 1.019
C-SPAN (US) 2.210
Canal Sur (US) 2.579
Channel 4 San Diego -7.085
CMT (US) 1.585
CNN en Español (US) 2.085
Discovery Life Channel (US) 1.122
Disney Channel (US) -0.562
ESPN Classic (US) 1.467
ESPN2 (US) -2.324
ESPNews (US) 0.948
EVINE (US) 2.053
Food Network (US) 2.405
Freeform (US) 2.623
FUSE (US) 2.457
FX Network (US) 1.795
FXM (US) 1.560
FXX (US) 1.530
Galavision (US) 2.013
Great American Country (US) 1.811
Hallmark Channel (US) 2.352
HGTV (US) 1.970
History (US) 1.502
INSP (US) 2.129
Lifetime Television (US) 0.363
MSG+ -11.769
MTV (US) -3.344
MTV2 (US) 1.976
NBC Sports Boston 0.608
NBC Sports Philadelphia -20.853
NBCSN (US) -2.105
New England Sports Network -13.614
Nickelodeon/Nick At Nite (US) -0.960
Outdoor Channel (US) 2.403
Ovation (US) 2.294
Paramount Network (US) 0.702
POP (US) 2.690
163
TCM (US) 1.515
TNT (US) -3.228
TV Land / TV Land Classic (US) 1.709
VH1 (US) 0.454
WGN America (US) 2.230

164
Table 1-C-2. Predictor Balance for “The Travel Channel”
Predictor Treated unit Synthetic unit
Bargaining Power 0.443 0.442
Cash Flow per Avg
Sub/Month
5.965 5.494
Cash Flow Margin 2.468 2.455
Net Ad Revenue per Avg
Sub/Month
73.821 73.464
Affiliate Fee per Avg
Sub/Month at 2006
0.074 0.072
Affiliate Fee per Avg
Sub/Month at 2001
0.051 0.049
Affiliate Fee per Avg
Sub/Month at 1997
0.035 0.034

165
Table 1-C-3. Unit Weights for “ROOT Sports Northwest”
Network Unit Weight
A&E (US) -0.156
Canal Sur (US) -0.227
CMT (US) -0.199
Discovery Life Channel (US) -0.203
Disney Channel (US) -0.516
ESPN (US) 2.413
ESPN2 (US) 0.083
EVINE (US) -0.246
Freeform (US) -0.171
Galavision (US) -0.239
Hallmark Channel (US) -0.218
INSP (US) -0.246
Lifetime Television (US) -0.127
MSG+ 0.555
MTV (US) -0.094
MTV2 (US) -0.212
NBC Sports Boston -0.029
New England Sports Network 0.400
Nickelodeon/Nick At Nite (US) -0.037
Paramount Network (US) -0.156
POP (US) -0.252
TNT (US) 0.242
VH1 (US) -0.165
WGN America (US) -0.204

166
Table 1-C-4. Predictor Balance for “ROOT Sports Northwest”
Predictor Treated unit Synthetic unit
Bargaining Power 0.648 0.603
Cash Flow per Avg
Sub/Month
372.114 649.214
Cash Flow Margin 19.151 19.257
Net Ad Revenue per Avg
Sub/Month
271.190 695.081
Affiliate Fee per Avg
Sub/Month at 2008
2.069 2.206
Affiliate Fee per Avg
Sub/Month at 2003
1.570 1.367
Affiliate Fee per Avg
Sub/Month at 1998
0.448 0.621
Affiliate Fee per Avg
Sub/Month at 1993
0.383 0.451

167
Table 1-C-5. Unit Weights for “AT&T Sports Pittsburgh”
Network Unit Weight
A&E (US) -0.179
Canal Sur (US) -0.251
CMT (US) -0.219
Discovery Life Channel (US) -0.231
Disney Channel (US) -0.462
ESPN (US) 2.496
ESPN2 (US) 0.073
EVINE (US) -0.270
Freeform (US) -0.191
Galavision (US) -0.262
Hallmark Channel (US) -0.241
INSP (US) -0.270
Lifetime Television (US) -0.151
MSG+ 0.604
MTV (US) -0.108
MTV2 (US) -0.235
NBC Sports Boston -0.043
New England Sports Network 0.577
Nickelodeon/Nick At Nite (US) -0.044
Paramount Network (US) -0.160
POP (US) -0.275
TNT (US) 0.257
VH1 (US) -0.183
WGN America (US) -0.231

168
Table 1-C-6. Predictor Balance for “AT&T Sports Pittsburgh”
Predictor Treated unit Synthetic unit
Bargaining Power 0.577 0.583
Cash Flow per Avg
Sub/Month
541.992 619.077
Cash Flow Margin 28.347 30.548
Net Ad Revenue per Avg
Sub/Month
206.338 482.094
Affiliate Fee per Avg
Sub/Month at 2008
2.045 2.167
Affiliate Fee per Avg
Sub/Month at 2003
1.650 1.423
Affiliate Fee per Avg
Sub/Month at 1998
1.110 0.922
Affiliate Fee per Avg
Sub/Month at 1993
0.655 0.656

169
Appendix 1-D. Treated and Counter Factual Averages
Figure 1-D-1. Treated and Counter Factual Averages on “The Travel Channel”

170
Figure 1-D-2. Treated and Counter Factual Averages on “ROOT Sports Northwest”

171
Figure 1-D-3. Treated and Counter Factual Averages on “AT&T Sports Pittsburgh”

172
Figure 1-D-4. Treated and Counter Factual Averages on “GSN”

173
Appendices for Chapter 2
Appendix 2-A. Survey Question on Technology Maturity

Table 2-A-1. Definition of Technology Maturity Outcomes

Level Meaning
1 Effect demonstrate in laboratory
2 Application identified
3 Technology and application validated in laboratory environments
4 Technology and application validated in relevant environments
5 Prototype ready
6 Product fully realized

Table 2-A-2 Frequency of Technology Maturity
Level Frequency Percent Cumulative Percent
1 41 19.71 19.71
2 50 24.04 43.45
3 45 21.63 65.38
4 17 8.17 73.56
5 43 20.67 94.23
6 12 5.77 100.00
Total 208 100.00

174
Appendix 2-B. Survey Question on Business Model Development
To what extent do you agree with the following statements as they pertain to this business?
1. Potential customer segments have been explored.
2. The target customer segments have been identified.
3. Alternative value propositions have been identified.
4. The value proposition has been defined.
5. Alternative distribution channels have been explored.
6. The distribution channels have been determined.
7. Alternative types of relationship the business will have with customers has been defined.
8. The type of relationship the business will have with customers has been defined.
9. Potential revenue streams have been evaluated.
10. Customers' willingness to pay has been researched.
11. The revenue structure for this business has been defined.
12. The key resource needs of this business have been evaluated.
13. The key resource needs of this business have been determined.
14. The scope of activities that this business will undertake has been considered.
15. The key activities this business will undertake have been defined in detail.
16. Alternate partnerships that this business will establish have been explored.
17. The key partnerships that this business will establish have been defined in terms of which
partners the firm needs.
18. The key activities that these partners will perform have been defined.
19. The key resources that these partners will provide have been defined.
20. The costs of providing the product or service have been researched.
21. The cost structure of the business has been determined

175
Appendices for Chapter 3
Appendix 3-A. Definition of M&A Features
Status of Transaction
Completed The transaction has closed.
Intended The acquirer has announced that they propose or expect to make an
acquisition, generally used for Repurchases.
Pending The transaction has been announced but has not been completed or
withdrawn.
Seeking Buyer The target company has announced plans to seek out a buyer or buyers
for its assets or the company itself.
Rumor Reports about a likely transaction have been published in the media, but
no formal announcement has been made by either the target or acquirer.
Discontinued
Rumor
Target company has formally denied the rumor of an acquisition or
merger.
Withdrawn The target or acquirer in the transaction has terminated its agreement,
letter of intent, or plans for the acquisition or merger.
Seeking Buyer
Withdrawn
The target in the transaction has terminated its plans to seek out a buyer
or buyers for its assets, stock, or the company itself.
Type of Mergers
Friendly
Merger
Occurs when one corporation acquires another with both boards of
directors approving the transaction.
Hostile Merger Occurs when the acquiring corporation attempts to take the target
corporation without the agreement of the target corporation’s board of
directors.
Horizontal
Merger
Two companies that are in direct competition and share the same
product lines and markets.
Unrelated
Merger
Two companies that have no common business areas.

176
Appendix 3-B. Corporate Identifiers
Identifier Description
CUSIP The identifier used in the SDC M&A database
GVKEY The identifier used in COMPUSTAT
PERMNO The identifier used in Kogan et al. patent database. PERMNO identifies
stocks. Some companies have more than one PERMNO.

177
Appendix 3-C. Definition of Inventor’s Characteristics
Variable Description
Generality A patent that is cited by a broader array of technology classes is viewed as
having greater generality. Generality is calculated as one minus the
Herfindahl index of citing patents, which captures the dispersion across
technology classes of patents using the patent. To account for cases with a
small number of patents within technology classes
50
, we use the bias
correction described in Jaffe and Trajtenberg (2002). As an inventor
characteristic, generality measures the average generality of patents that
the inventor has created.
Originality A patent that cites a broader array of technology classes is viewed as
having greater originality. Originality is calculated as one minus the
Herfindahl index of cited patents, which captures the dispersion of the
patent citations across technology classes. To account for cases with a
small number of patents within technology classes, we use the bias
correction described in Jaffe and Trajtenberg (2002). As an inventor
characteristic, originality measures the average originality of patents that
the inventor has created.
Fraction of
exploratory
patent
Less than or equal to forty percent of citations a patent makes are existing
knowledge
51
. As an inventor-characteristic, the fraction of exploratory
patent measures the average fraction of exploratory patent that the
inventor has created.
Fraction of
exploitative
patent
More than or equal to sixty percent of citations a patent makes are
existing knowledge. As an inventor-characteristic, the fraction of
exploitative patent measures the average fraction of exploitative patent
that the inventor has created.

50
Technology class: A technology class is a detailed classification of the USPTO, which clusters patents based on
similarity in the essence of their technological innovation. Technological classes are often more detailed than
industry classifications, consisting of about 400 main (three-digit) patent classes, and over 120,000 patent
subclasses.
51
Existing knowledge: All firm patents prior to this one, and all patents that were cited by the firm’s patents from
the last 5 years.
178
Appendix 3-D. Company Characteristics
Variable Description
Sale Sales over turnover (net).
Market equity (ME) Annual price close (fiscal) times common share outstanding.
Research and
development intensity
(RD)
Research & Development expenditures, scaled by market
equity.
Return on assets (ROA) Net income over total assets.
Asset growth (INV) Growth rate of total assets.
Tobin’s Q (Q) Total market value of firm divided by total asset value of
firm.
Herfindahl Index (herf) Measure the size of a firm in relation to the industry in terms
of sale.

179
Appendices for Chapter 4
Appendix 4-A. General Partners in Private Equity Firms among U.S. Forbes 400, Ranked
by Net Worth in 2019
Name Net Worth Firm Type
Stephen Schwarzman $17.7B Blackstone Source of wealth
Pierre Omidyar $13.1B
Omidyar Technology
Ventures
Current primary
position
Jerry Jones $8.6B
Blue Star Innovation
Partners
Current primary
position
Leon Black $7.7B
Apollo Global
Management
Source of wealth
John Doerr $7.5B Kleiner Perkins Source of wealth
Israel Englander $6.6B
Innovatus Capital
Partners
Current primary
position
George Roberts $6.1B KKR & Co. Source of wealth
Henry Kravis $6B KKR & Co. Source of wealth
Dan Cathy $5.7B Engage VC
Current primary
position
Tom Gores $5.6B Platinum Equity Source of wealth
Sam Zell $5.5B
Equity Group
Investments
Current primary
position
Whitney MacMillan $5.1B
Western NIS
Enterprise Fund
Current primary
position
Jeffrey Skoll $5.2B
Pacific Sequoia
Holdings
Current primary
position
Dirk Ziff $5B Ziff Capital Partners
Current primary
position
Robert F. Smith $5B Vista Equity Source of wealth
Joshua Harris $4.4B
Apollo Global
Management
Source of wealth
Neil Bluhm $3.9B Walton Street Capital
Current primary
position
Gouglas Leone $3.8B Sequoia Capital Source of wealth
David Bonderman $3.7B TPG Source of wealth
Marc Rowan $3.6B
Apollo Global
Management
Source of wealth
Michael Moritz $3.5B Sequoia Capital Source of wealth
Anthony Pritzker $3.5B The Pritzker Group
Current primary
position
David Steward $3.5B Kingdom Capital
Current primary
position
J.B. Pritzker $3.4B
Pritzker Group
Venture Capital
Current primary
position
180
Sid Bass $3.1B
Sid R. Bass
Associates
Current primary
position
William Conway, Jr. $3.1B Carlyle Group Source of wealth
Daniel D'Aniello $3.1B Carlyle Group Source of wealth
David Rubinstein $3.1B Carlyle Group Source of wealth
Orlando Bravo $3B Thoma Bravo Source of wealth
Haim Saban $2.9B Saban Capital Group
Current primary
position
Penny Pritzker $2.8B PSP partners
Current primary
position
Antony Ressler $2.8B
Apollo Global
Management
Source of wealth
Mark Stevens $2.7B Sequoia Capital Source of wealth
Jim Coulter $2.6B TPG Source of wealth
John Pritzker $2.6B Geolo Capital
Current primary
position
Jerry Yang $2.6B AME Could Ventures
Current primary
position
John Fisher $2.5B Sansome Partners
Current primary
position
James Breyer $2.6B
Accel Partners,
Breyer Capital
Source of wealth
Peter Thiel $2.3B
Founders Fund,
Mithril Capital
Management, Thiel
Capital
Current primary
position
Bruce Karsh $2.2B
Oaktree Capital
Management
Source of wealth
Howard Marks $2.2B
Oaktree Capital
Management
Source of wealth
Brian Sheth $2.2B Vista Equity Source of wealth
Kavitark Ram
Shriram $2.2B Sherpalo Ventures
Source of wealth
Alex Gores $2.1B The Gores Group Source of wealth
Bradley Jacobs $2.1B Jacobs Private Equity
Current primary
position
Vinod Khosla $2.1B Khosla Ventures Source of wealth

Data Source: Forbes and PitchBook.

Abstract (if available)

Abstract “Changes call for innovation, and innovation leads to progress.” —Li Keqiang (Premier of the People’s Republic of China). ❧ Innovation, which shapes the future of the world, is the key to growth and prosperity. This dissertation explores topics around innovation from multiple perspectives. In terms of the company-lifecycle while doing innovation, the dissertation covers both high-tech startups and dominant companies. Regarding people involved in innovation activities, this dissertation pays attention to startup founders, patent inventors, and private company investors. From the point of innovation policy, this dissertation provides insights on both the regulation on dominant companies and the protection to average investors. ❧ Chapter 1 investigates the effect of the Program Access Rule (PAR), an antitrust law on input prices in the U.S. television industry, through applying the Generalized Synthetic Control Method. Results show that vertical integration itself increases input prices, while combining it with the PAR can hold prices almost unchanged. The empirical strategy we propose has the potential to be implemented in a broader context, especially in the fast-growing high-tech industry where vertical integration raised the most concern. ❧ Chapter 2 explores nascent high-tech ventures prior to incorporation. Self-financing is the most common financing strategy for ventures in their earliest years. We find that entrepreneurial teams are more likely to invest in their own businesses if their founding teams are smaller. Our finding suggests that the founding team’s human capital can potentially substitute for its financial commitment. From a policy perspective, it becomes important to determine how to provide both human and financial resources to nascent high-tech ventures. In the strategy domain, this has important implications for early decision-making since the lack of funding may have subsequent consequences. ❧ Chapter 3 studies whether and how mergers and acquisitions (M&A) affect the productivity of patent inventors. Career paths of inventors are identified by their patent records. Completed and withdrawn M&A biddings are exploited as a quasi-experiment. Results show that for both acquirer inventors and target inventors, the number of patents will drop after experiencing M&A. When the acquirer and target are in the same industry, the quality of patent also drops for both acquirer- and target inventors. The results support the hypothesis of cultural conflict and competition for funding among inventors. We did not see evidence supporting collaboration among inventors or increased productivity due to increased pressure on job security concerns. ❧ Chapter 4 sheds light on a current debate on government policy and regulation. In the past few years, income and wealth concentration in the United States have drawn tremendous concern. As private equity attracts more and more attention, some people have connected inequality with the exclusive private equity investment opportunity to accredited investors. I argue that (1) marginal accredited investors do not benefit from the opportunity of investing in private equity

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Essays in behavioral and entrepreneurial finance

PDF

Three essays in international macroeconomics and finance

PDF

Essays on firm investment, innovation and productivity

PDF

Early-warning systems for crisis risk

PDF

Essays in financial intermediation

PDF

Essay on monetary policy, macroprudential policy, and financial integration

PDF

Essays on government intervention in financial crises

PDF

Essays on information and financial economics

PDF

Essays on innovation, human capital, and COVID-19 related policies

PDF

Workplace organization and asset pricing

PDF

Essays in macroeconomics and macro-finance

PDF

Increasing faculty engagement in entrepreneurial behavior to advance regional economic development: an innovation study

PDF

Essays on monetary policy and international spillovers

PDF

Essays on urban and real estate economics

PDF

The Silicon Valley startup ecosystem in the 21st century: entrepreneurial resilience and networked innovation

PDF

Essays on health insurance programs and policies

PDF

Economic policy-making in a changing environment: the politics of success in Peru vis-à-vis China

PDF

Behavioral approaches to industrial organization

PDF

Emerging market financial crises, investors and monetary policy

PDF

Essays in asset pricing

Asset Metadata

Creator Deng, Weiran (author)

Core Title Innovation: financial and economics considerations

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Economics

Publication Date 07/12/2020

Defense Date 05/01/2020

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

Tag entrepreneurial finance,entrepreneurship,Inequality,innovation,OAI-PMH Harvest,private equity,regulation

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Ranciere, Romain (committee chair), Zapatero, Fernando (committee chair), Aizenman, Joshua (committee member), Korteweg, Arthur (committee member)

Creator Email weirand@amazon.com,weirande@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-327688

Unique identifier UC11663920

Identifier etd-DengWeiran-8660.pdf (filename),usctheses-c89-327688 (legacy record id)

Legacy Identifier etd-DengWeiran-8660.pdf

Dmrecord 327688

Document Type Dissertation

Rights Deng, Weiran

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 a...

Repository Name University of Southern California Digital Library

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