The Global Financial Crisis and its Effect on Founders' Ownership in Initial Public Offerings : a Comparative Study of Intangible and Traditional Firms

J

2017

T

HE

G

LOBAL

F

INANCIAL

C

RISIS AND ITS

E

FFECT

ON

F

OUNDERS

’

O

WNERSHIP IN

I

NITIAL

P

UBLIC

O

FFERINGS

A

C

S

I

T

F

Rony Attie

M

ASTER

T

HESIS

M

S

F

ASSET MANAGEMENT

T

HE

G

LOBAL

F

INANCIAL

C

RISIS AND ITS

E

FFECT ON

F

OUNDERS

’

O

WNERSHIP IN

I

NITIAL

P

UBLIC

O

FFERINGS

A

C

OMPARATIVE

S

TUDY ON

I

NTANGIBLE AND

T

RADITIONAL

F

IRMS

R

ONY

A

TTIE

11385391

MS. EVGENIA ZHIVOTOVA

The thesis is submitted in partial satisfaction of the requirements for the

degree of Master of Science in Finance

U

A

–

A

B

S

June 2017

Pages: 61

S

TATEMENT OF

O

RIGINALITY

This document is written by Rony Attie who declares to take full responsibility for the con-tents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of comple-tion of the work, not for the contents.

A

BSTRACT

This study aims to link the global financial crisis to the founders’ ownership in a sample of 272 IPOs. Additionally, the tangibility of a firms’ asset is carefully considered. Among many of the 2008 Global Financial Crisis consequences, the banks’ tightening of liquidity constraints is a critical point. This study is the first to explore the crisis’ impact on owner-ships during an IPO, and an extension to the thin academic literature on founders. Through OLS, logistic, and difference-in-difference regressions, and a unique hand-collected dataset, the results point towards a strong statistically significant effect of the financial crisis on founders’ ownership in an IPO. Indeed, in an IPO, founders diluted on average 6.75% more of their ownerships as a result of the financial crisis. In addition, founders are approximately 4 times more likely to dilute at least half of their respective ownerships post-crisis than pre-crisis during an IPO. However, the tangibility of a firms’ asset does not play a major role in this study. These findings mainly imply that an event such as the global financial crisis, which led to the tightening of bank’s liquidity constraints, have a strong and significant im-pact on founders’ ownership in an IPO.

T

ABLE OF

C

ONTENTS

List of Acronyms ... 8

I Introduction ... 10

II Overview of the Global Financial Crisis ... 13

1 The Build-Up ... 13

2 The Banks’ Response ... 15

III Literature Review ... 17

1 Previous Research ... 17

A. From the Crisis to the Banks’ Response ... 17

B. Initial Public Offerings ... 18

C. Shares Ownerships ... 19

D. Importance of Founders ... 20

E. Distinction Between Intangible and Traditional Firms ... 20

2 Hypotheses ... 21

IV Methodology ... 23

1 Descriptive Variable ... 23

2 Intangible Proxy ... 28

3 Regression Information ... 30

4 Goodness of Fit ... 33

V Data ... 34

1 Data Gathering ... 34

2 Descriptive Statistics ... 36

A. General Overview ... 36

B. Subsample Statistics ... 40

VI Results & Discussions ... 44

1 OLS Regressions ... 44

2 Logistic Regressions ... 50

3 Difference-in-Difference Regressions ... 53

VII Robustness Checks ... 57

1 Econometric Issues ... 57

A. Multicollinearity ... 57

B. Heteroscedasticity & Autocorrelation ... 58

C. Endogeneity ... 59

2 Revisited Regressions ... 60

D. OLS Regression ... 60

VIII Conclusion ... 67

1 Summary ... 67

2 Limitations ... 68

3 Recommendations ... 69

References ... 71

Appendices ... 75

L

IST OF

A

CRONYMS

ABCP Asset Backed Commercial Paper

AIC Akaike Information Criterion

ADR American depositary Receipts

BIC Bayesian Information Criterion

CDO Collateralized Debt Obligations

CEO Chief Executive Officer

CFO Chief Financial Officer

FCIC Financial Crisis Inquiry Commission

FRED Federal Reserve Economic Data

GAAP Generally Accepted Accounting Principles

GICS Global Industrial Classification Standards

IPO Initial Public Offering

JOBS Jumpstart Our Business Startups

NASDAQ National Association of Securities Dealers Automated Quotations NINJA No Income, No Job & Assets

NPM Net Profit Margin

NYSE New York Stock Exchange

OECD Organization for Economic Cooperation and Development

OLS Ordinary Least Square

PP&E Property, Plant & Equipment R&D Research & Development

ROA Return on Assets

SEC Security Exchange Commission

SG&A Selling, General & Administrative

SIC Standard Industrial Classification

SPV Special Purpose Vehicles

I

–

I

NTRODUCTION

The global financial crisis is a “once-in-a-century credit tsunami”, referring to the words pronounced by the Federal Reserve chairman at the time, Dr. Alan Greenspan. This crisis begins in the summer of 2007 and is an unexpected event in the eyes of most people. What originally is a mortgage problem spilled over the worlds’ financial markets, as Longstaff (2010) proves, threatening the economy of powerful countries, and in particular, the United States of America. Unsurprisingly, this leads to volatile and uncertain market conditions, and the most bearish market since the great depression in 1929. In consequence, the Initial Public Offering (IPO) market enters a “cold” period. Banks’ profits shrink drastically, which result in a decrease in their risk tolerance. Thus, deleveraging and improving their capital structure is primordial for banks, especially after the significant loss of trust and confidence in the public’s eyes.

Among the numerous economic, social, and psychological consequences of the glob-al financiglob-al crisis, the banks’ tightening of liquidity constraints is centrglob-al to this study’s re-search. Forsberg (2015), and Ivashina and Scharfstein (2010) separately show a decrease in the amount of lending due to the bank’s new policies in the post-crisis period. These policies increase the difficulties of borrowing funds.

Other than debt, companies raise funds in a variety of ways. A major motivation of going public is to raise funds through equity. Indeed, Pagano, Panetta, and Zingales (1998) find that companies appear to go public to rebalance their accounts. Myers and Majluf’s (1984) pecking order theory lists internal financing as the cheapest way to raise funds, fol-lowed by debt, and then equity. As the financial crisis makes it harder for businesses to bor-row funds from banks, insiders – such as founders – realize that they have to increasingly dilute their respective ownership in an IPO to raise the amount of capital needed. In addition and as a result of their uncollateralized assets, firms with a higher proportion of intangible assets face further difficulties to borrow funds compared to firms with a higher proportion of tangible assets. Throughout this paper, the former is referred to as “intangible firms”, and the latter as “traditional firms”. In line with this argument, Bates, Kahle, and Stulz (2009), find that intangible firms are more active in precautionary saving in comparison to

tradition-al ones. A logictradition-al reason is that founders of intangible firms are aware of the firms’ risky nature and difficulty demanding a loan from a bank; thus, are more inclined to save cash.

Founders of businesses are major pioneers of a working society; they play a signifi-cant role in a country’s economy and competitive market. In her findings, Nelson (2003) suggests the importance of founders’ in a company’s performances and in the eyes of the public market by showing a higher market reaction to founder-led firms in an IPO. Yet, to this day, the academic literature on founders is still very thin. Throughout this paper, found-ers’ ownership change is to be understood as the percentage increase/decrease from the pre- to the post-offering ownerships. A decrease (increase) in founders’ ownership change signi-fies that founders are diluting more (less) of their respective ownership in an IPO.

Previous empirical papers cover the IPO market pre- and post-crisis, in terms of vol-ume, underpricing, proceed, etc. Other papers discuss the ownership separation in an IPO, focusing for instance on Chief Executive Officers (CEO) and Venture Capitalists’ (VC) roles. Finally others study the difference between intangible and traditional firms in the light of precautionary savings. However, the academic world lacks a discussion concerning founders’ ownership in an IPO and its effect due to the crisis, while considering the natural differences between intangible and traditional firms. The literature gap I intend to close comes from answering the following main question:

“How has the global financial crisis affected the founders’ ownership in an IPO, and how does it differ between intangible and traditional firms?”

Interestingly, this paper is the first to undergo this specific study in order to add fur-ther knowledge to this area of research. Three different methodologies are used to answer the study’s main empirical question. First, Ordinary Least Square (OLS) regressions are run to observe whether or not there is a substantial effect of the financial crisis on founders’ ownership change in an IPO. Then, logistic regressions are run to study the odds ratio of founders diluting at least half of their respective ownership post-crisis, as compared to pre-crisis in an IPO. Last, the difference-in-difference regressions are done for the sole purpose of studying the difference in the crisis’s impact on founders’ ownership change for intangi-ble as compared to traditional firms. The study’s sample is separated between intangiintangi-ble and

traditional firms using the perpetual inventory method and an approach inspired by Peters and Taylor (2017).

This paper contributes to the body of the literature by studying the impact of the global financial crisis on founders’ ownership in IPOs. Four major contributions are extract-ed from this paper. First, with the help of a uniquely handpickextract-ed dataset, this study improves the thin academic literature on founders by specifically studying the crisis’ impact on their respective ownership in IPOs. Second, this paper explores a previously un-studied impact of the crisis – which is the change in percentage ownership. This not only adds to the academic literature on IPOs, but also presents a new economic consequence of the global financial cri-sis. Third, I point out a significant inaccuracy in the Compustat database that should raise attentiveness to any past and future researches done involving SG&A expenses. This study overcomes this difficulty by handpicking the necessary data. Last but not least, the method-ology this study uses to separate intangible from traditional firms is considered contributive to the literature. Indeed, with the help of previous empirical papers1, this study develops a reliable appraisal of a company’s stock of intangible assets.

The rest of the paper is structured as follow: Section II briefly introduces the global financial crisis, including its origination and its consequence on bank’s liquidity constraints. Section III reviews previous literature surrounding this study’s area of research, as well as explicitly states its hypotheses. Section IV describes the variable of interests and details the methodologies in place to test the hypotheses. Section V provides details on the gathering of data and its descriptive statistics. Section VI unveils the study’s results and discusses its economic interpretation. Section VII embodies the robustness checks by presenting econo-metric issues, then by revisiting previous regressions. Section VIII culminates the study with the conclusion, limitations, and ideas for future researches.

1 _{Peters and Taylors (2017), Falato, Kadyrzhanova, and Sim’s (2013), Jaffe, and Trajtenberg (2001), Lev and} Radhakrish-nan (2005), and Corrado, Hulten, and Sichel (2009).

II

–

O

VERVIEW OF THE

G

LOBAL

F

INANCIAL

C

RISIS

This section, divided into two sub-sections, presents an overview of the global financial cri-sis. First, I present the crisis’ origination and securitization processes, and how these phe-nomena led to a highly volatile and uncertain market environment. Then, I illustrate how the crisis leads to the banks’ tightening of liquidity constraints.

1

–

T

B

-

In order to thoroughly understand the banks’ tightening of liquidity constraints, this sub-section attempts to demystify and briefly summarize the origination and build-up to the global financial crisis of 2008. The following explanation is based on previous empirical pa-pers, articles, as well as theoretical knowledge. Most of the information is centered on re-searches by Levitin and Wachter (2010), Comiskey and Madhogarhia (2009), Spiegel (2011), and the Financial Crisis Inquiry Commission (2011) (FCIC). It is worth noting that this study does not intend to blame anyone for the crisis, but instead shows how the crisis leads to the loss of trust and confidence. To make the explanation easier to follow and un-derstand, the below figure is used as a support.

FIGURE 1

THE CRISIS’ORIGINATION &SECURITIZATION

This figure represents the origination and securitization processes of the financial crisis. It should be noted that this is not an exhaustive, bur rather a simplified picture of the real event.

It all starts with borrowers – most specifically mortgagors – that take in mortgages from commercial banks. The money lent appears in the commercial banks’ balance sheets as assets due to the mortgagors’ planned interest payments.

As a result of the great depression of 1929, the Glass-Steagall Act is passed in 1933 that separates the participation of commercial banks from investment banks’ businesses, and vice-versa. However, in 1999, the Glass-Steagall Act is repealed, and therefore allows mer-gers to happen between major commercial and investment banks. The latter purchase the loan book that is on the former’s balance sheet.

After this purchase, the investment banks set up Special Purpose Vehicles (SPV) to sell them their loan books, with the aim to remove the loans from the investment banks’ bal-ance sheets. The SPVs use these loans to issue a type of bond; asset-backed security. As its name states, these bonds are backed by assets, which are, in the case of the 2008 global fi-nancial crisis, mortgages. The bondholders purchase the bonds from the SPVs, and the pro-ceeds are transferred all the way back to the originating firms, the commercial banks.

With new proceeds, the commercial banks issue more loans. A major and important detail that contributes to the 2008 crisis is the shift of the commercial banks’ fundamental model. Indeed, the commercial banks become more focused on lending money instead of attracting money from depositors. When commercial banks issue mortgages to all secured borrowers, they start to issue mortgages to borrowers with low credit ratings, such as No In-come, No Job or Assets (NINJA) individuals. These mortgages are in fact called Subprime mortgages, and have a higher interest rate compared to conventional mortgages because of their risky nature.

The investment banks, that initially purchase the loan book from the commercial bank, decide to separate the loans into different tranches based on the repayment likelihood – seniority of tranches. This process is referred to as securitization. For the ease of explana-tion, let’s consider three tranches rated as follow; AAA, AA, and BBB, from the most to the least secure. The most senior tranche (AAA) is used to back the bonds issued by the SPVs. The middle tranche, AA, is divided into three sub-tranches similar to the original division. What happens here is another major contribution to the 2008 financial crisis, because the original middle tranche leads to the creation of a senior sub-tranche. This is referred to as a Collateralized Debt Obligation (CDO). This senior sub-tranche is also used to back the

bonds issued by the SPVs. On the 9th of August 2007, BNP Paribas freezes three of their funds due to issues valuing CDOs. Adam Applegarth, Northern Rock’s chief executive, says that day is “the day the world changed”. During that same month, the outstanding asset backed commercial papers (ABCP) shrink by $190 billion. Covitz, Liang, and Suarez (2013) consider this event to largely contributes to the financial crisis.

The credit rating agencies play a significant and quasi-regulatory role in this crisis because they are in charge of objectively rating the bonds issued by the SPVs. The credit agencies justify giving the bonds issued by the SPVs an AAA rating. According to the report “Guide to European CMBS” released by Barclays Capital in 2006, 74% of all Commercial Mortgage Backed Securities are rated AAA in 2005, which is above the historical average. In fact, a lot of unsecured bonds are rated as secured, and therefore lead unknowledgeable bondholders to purchase bad mortgage-backed securities.

Before the crisis, the investment banks purchase insurance from the world’s biggest insurance company, AIG. When the low credit rating borrowers start to default on their re-spective mortgages, house prices start to decrease, and the value of the mortgage-backed se-curities decrease as well. The insurance companies’ assets shrink as a consequence. Hank Paulson, the U.S. Treasury secretary at the time, says in May 2008 “I do believe that the worst is likely to be behind us”. However, this is not completely true. In fact, the U.S. expe-riences the biggest and largest bankruptcy ever – Lehmann Brothers is left to fail on the 15th of September 2008. However, the government bails out AIG the following day, as its failure would have an even more catastrophic effect on the economy.

2

–

T

B

’

R

Using the FCIC’s own words, “[…] this financial crisis was avoidable”. Numerous econom-ic, social, and psychological consequences come as a result of the global financial crisis. The confidence and the trust that banks portray start to deteriorate in the eyes of investors and the general public, creating a volatile and uncertain market that the world has not experienced since the great depression of 1929. Bloomberg studies the after tax and minority interests profits of the five largest commercial banks in the United States. In this study, as shown in Figure 2, a significant decrease of profits, from around $25 billion to -$25 billion, is ob-served from 2006 to 2008. This dramatic drop in earnings leads to the decrease of banks’

risk tolerance. Therefore, and after an increase in regulations, these financial institutions have to deleverage and improve their capital structure as soon as possible, resulting in the tightening of liquidity constraints.

FIGURE 2

THE BANKS’PROFITS

This figure represents the 5 largest U.S. and several European commercial banks’ profit from the beginning of 2002 to the end of 2008. The graph was directly obtained from Bloomberg.

III

–

L

ITERATURE

R

EVIEW

This section, divided into two sub-sections, deeply scrutinizes previous empirical researches in order to show a clear gap in the literature. First, on one hand, past findings are used to link the global financial crisis and the difficulties of borrowing, and on the other hand, to motivate the study’s research idea. Last, an explicit statement of the hypotheses concludes this part of the paper.

1

–

P

R

A – FROM THE CRISIS TO THE BANK’S RESPONSE

Longstaff (2010) uses the subprime crisis as an opportunity to study its spillover on the fi-nancial markets. Indeed, he finds strong evidence that the mortgage crisis contaminated the worlds’ financial markets. As previously touched upon, the consequences of the crisis are numerous. For instance, Hoffman, Post, and Pennings (2011) state that investors changed their behavior and are more risk averse due to the crisis. Moreover, the Organization for Economic Cooperation and Development (OECD) (2011) finds that investor’s risk appetite decreases as a result of the financial crisis. In addition, and in accordance to Guerrieri and Lorenzioni (2011), individuals are forced to deleverage and increase their cash savings due to the post-crisis market environment. As a result, and as FCIC states, markets become vola-tile and uncertain. Indeed, Claessens et al. (2010) mention that uncertainties lead to further market confusion and disturbance following the crisis.

As Section II mentions, banks’ sensitivity to risk increases as a result of the global financial crisis. Following their profit losses, banks need to recover; and thus, deleverage and improve their capital structure. Also, due to the fact that banks become even more regu-lated, they have to tighten their liquidity constraints, and therefore, make it harder for busi-nesses to obtain loans. Forsberg (2015) argues that the financial crisis has a substantial im-pact on the capital market and, among others, heavily reduce the lending by financial institu-tions. Complementary, Ivashina and Scharfstein (2010) show that the amount of lending from financial institutions drops by 47% during 2008. These authors also state that lending is reduced throughout all loans and credit lines due to the extensive market volatility the

cri-sis brought with it. Hence, the cricri-sis decreases the amount of lending financial institutions are willing to give out and instead, make these institutions highly risk-averse. Thus, borrow-ing becomes harder for businesses, implyborrow-ing that raisborrow-ing funds through debt become more unavailable for companies after the crisis.

So far, I have linked the global financial crisis to banks’ tightening their liquidity constraints; hence, making loans harder for businesses to obtain. Now, I will use past litera-ture to help motivate the study’s research idea.

B – INITIAL PUBLIC OFFERINGS

Previous studies find several consequences of the 2008 financial crisis on IPOs. Following the crisis, the yearly volume of IPOs decreases. However, according to Henry and Gregoriou (2013), the yearly volume of post-crisis IPOs almost returns to its pre-crisis value in 2013. But in terms of sales, the authors affirm that post-crisis IPOs are significantly greater. It is interesting to note what the major causes are that lead to the fluctuation of IPO volumes dur-ing and after the crisis. Even before the crisis occurs, Lowry (2003) affirms that firms' de-mands for capital and investor sentiment are two main causes for the IPO volume fluctua-tion. Also, Ritter and Welsh (2002) find that market conditions are the most important factor in a company’s decision to be listed on the public stock market. Going public is an expen-sive and time-consuming process. For instance, the Sarbanes-Oxley Act of 2002 increase the cost of going public by 90 basis points of gross proceeds according to Kaserer, Mettler, and Obernberger (2011). Also, a lot of time is spent on all the mandatory reports, such as 10-K and 8-K, that companies must submit to the Security Exchange Commission (SEC) before a specific deadline.

The main risk that companies face when going public is underpricing. This risk is usually referred to as “leaving money on the table”. Ibbotson (1975) and Ritter (1984) pro-vide convincing epro-vidence that IPOs are on average underpriced. This phenomenon is not al-ways unintentional, as Beatty and Ritter (1986) find that underwriters do enforce the under-pricing equilibrium. The decision to go public has for long confused scholars and practition-ers. Once, the decision to go public is viewed as a natural process of company growth. How-ever, as it becomes evident that those large companies do not go public despite their growth, scholars begin to search for new reasons. Today, there are a large variety of examined and

validated factors to why companies pursue an IPO. Zingales (1995) and Mello and Parsons (2000) mention a motivation for an IPO – which is for insiders to cash out. Ang and Brau (2003) test the motivation of insiders, such as founders, to cash out, and found that they sell shares in an IPO for personal gain. Pagano, Panetta, and Zingales (1998) find that companies appear to go public to rebalance their accounts. With that said, amongst the several reasons to initiate an IPO, the main motivation of going public is to raise capital, whether it is for the founders’ personal gain or the financing of future opportunities.

It is assumed that the issuance of equity is the most expensive way to raise funds. This assumption follows the pecking order theory that Myers and Majluf (1984) introduce, stating that businesses prioritize their financing strategies; internal funds is always the first choice, then debt, and last equity issuance. It is worth noting that this is a theory and neither a fact nor a law. In a survey done by Brau and Fawcett (2006), the pecking order of financ-ing, as a motivation of going public, receive low support by Chief Financial Officers (CFO). However, the timing of this survey is to be considered. This study is done in 2006, when the cost of debt is low, which surely influence the findings. We assume that if this survey is to be conducted again post-crisis, after the banks’ tightening of liquidity constraints, the peck-ing order of financpeck-ing would receive stronger support from CFOs. In the post-crisis period, where banks’ liquidity constraints are at its highest because of the level of volatility and un-certainty, a company looking to raise funds can not rely as much on bank lending. Campello, Graham, and Harvey (2010) find that the 2008 credit crisis leads many firms to bypass at-tractive investment opportunities as a result of the complications of external borrowing. With that said, one can draw the conclusion that a company looking to raise capital is more likely to resort to equity financing after the crisis, compared to the pre-crisis period.

C – SHARES OWNERSHIPS

When a company issues additional equity to raise more funds during an IPO, it is logical that internal ownership dilutes. Prior studies investigate the ownership topic during IPOs. Bren-nan and Franks (1997) find that, in less than 7 years following the IPO, two-thirds of the of-fering company’s shares are sold to outside shareholders. This event leads to further separa-tion between ownership and control. Other studies on ownership, such as the one by Meg-ginson and Weiss (1991), affirm the positive role of VC firms. The latter’s presence in a

business’ ownership structure lowers the cost of going public and increases the going-public company’s net proceeds. Jain and Kini (1995), discover that IPOs backed by a VC firm dis-play higher operating performance, and generally have a higher market value.

D – IMPORTANCE OF FOUNDERS

Despite numerous studies available on board and directors’ ownership, institutional ship, internal ownership, and public ownership, the academic literature on founders’ owner-ship is very thin. One reason is the accessibility of data. To my knowledge, there is no data-base containing founders’ ownership data apart from the SEC’s EDGAR datadata-base.

A specific study on founders is important in reference to N. Wasserman’s (2012) book, “The Founder’s Dilemmas”. This Harvard Business School professor develops the “rich” versus “king” trade-off that founders make. A “king” has strong control over the company and financial gains will be below potential, and a “rich” has little control over the company and financial gains will be close to potential. It is interesting to observe how founders respond to this dilemma as a result of the global financial crisis. Prior literature finds founders to bring distinct and value-added skills and features to the business in terms of accounting performance and stock market valuation. Anderson and Reeb (2003) find that founder-CEOs lead to superior accounting profitability measures. This supports Morck, Schleifer, and Vishny (1988) claim that founder-CEOs bring innovative and value-enhancing expertise to the business. In terms of market valuation, Nelson (2003) finds that the stock market’s reaction to founder-led firms is higher than for non-founder-led firms. These findings suggest the importance of founders’ in the company’s performances and in the eyes of the public market.

E –DISTINCTION BETWEEN INTANGIBLE AND TRADITIONAL FIRMS

Finally, discerning between traditional and intangible firms is interesting. A traditional firm is one with a high percentage of physical assets, such as equipment and vehicles. An intan-gible firm is one with a high percentage of non-physical assets, such as patents, trademarks, goodwill and copyrights. The results of any financial empirical paper could significantly vary when this distinction is used because of the natural differences between these two types of firms. In the majority of today’s academic literature, this separation is taken for granted.

Studies by Gamayuni (2015) and the OECD (2011) show that intangible firms are consid-ered more risky than tangible firms. Logically, intangible firms therefore face more difficul-ties than traditional firms raising funds through bank loans. This affirmation derives from the fact that intangible firms do not have as much collateral assets as its counterparts, mak-ing it harder to raise debt. Commercial banks require a high level of collateral assets, specif-ically after the event of the financial crisis, which can unfortunately not be met for to the same extent for intangible firms as for traditional firms. Bates, Kahle, and Stulz (2009) sup-port this claim and state that firms with more intangible assets tend to be more active in pre-cautionary savings, and therefore hold more cash, compared to its counterparts. This implies that intangible firms perceive themselves as more risky than tangible firms. Precautionary savings are cash savings resulting from future income’s uncertainty. Intangible firms are more active in precautionary savings than traditional ones as they are aware of the difficul-ties taking in loans with a low stock of collateralized assets. Moreover, Falato, Kadyrzhano-va, and Sim (2013) find that intangible firms hold more cash to preserve financial flexibility. Different approaches are used to separate a sample of firms into traditional and intangible firms. This paper’s approach is inspired from Peters and Taylor (2017). In their study, the authors find that intangible capital responds slower to changes in investment opportunities than traditional capital. My approach towards this separation is developed in Section IV.

2

–

H

I have now motivated the importance of the study’s research question, and can explicitly state the hypotheses:

(1) In IPOs, founders’ ownership change2 is more likely to be lower post-crisis than pre-crisis.

(2) In IPOs, founders are more likely to dilute at least half of their respective owner-ship post-crisis than pre-crisis.

(3) In IPOs, the effect of the financial crisis on founders’ ownership change is stronger for intangible firms, compared to traditional firms.

2 _{Reminder: Founders’ ownership change is the percentage increase/decrease from the pre- to the post-offering} owner-ships. A decrease (increase) in founders’ ownership change signifies that founders are diluting more (less) of their respec-tive ownership in an IPO.

I would also like to mention that I thought the other way around. One can argue that the crisis lead to an increase, and not a decrease, in founders’ ownership change. The ration-al is that the volatile and uncertain market created by the crisis makes potentiration-al shareholders more risk averse to invest in newly public companies. Hence, it leads company insiders, such as founders, to end up with a higher ownership change post-crisis than pre-crisis. How-ever, in an IPO, the amount of shares issued is typically determined ex-ante, meaning that the level of founders’ ownership change is not directly affected by investor sentiment. If the founders have more shares than initially planned, because the underwriter cannot attract enough shareholders, then the IPO is considered a failure. Hence, this approach is not valid.

IV

–

M

ETHODOLOGY

This section, divided into three sub-sections, details information concerning the methodolo-gy used in this study. First, an explanation of every variable needed for the regressions is discussed. Then, a deeper look is taken into the approach used to separate intangible and tra-ditional firms. Last, the regression descriptions and equations used to test the hypotheses are presented.

1

–

D

V

This study requires mostly quantitative data, although some qualitative data is also present. All financial crisis data, company specific data and most IPO characteristics data are quanti-tative. The qualitative data needed is the name of the underwriters in every IPO, used in or-der to rank them according to their prestige. The following paragraphs describe our depend-ent, independdepend-ent, and control variables.

A – FIRM CHARACTERISTICS

v Industry Fixed Effect

This control variable represents a dummy for each sector. The sectors are divided based on their respective Global Industry Classification Standard (GICS), as shown in Appendix – Table A.1. I use industry fixed effects, as I believe industries are important factors that can affect regression results. Some industries might find it easier to borrow money from the bank, whereas some might find it harder. As Ritter and Loughran (2004) find, technology companies are the riskiest. Logically, this makes it harder for them to borrow money from the bank after the crisis compared to their peers’ industries. Also, Benveniste, Busaba, and Wilhelm (2002) argue that when one firm goes public, other firms within the same industry learn about their own valuations, which might encourage them to go public in their turn. Moreover, Ljungqvist and Wilhelm (2003) mention the high information asymmetry that is in technology companies compared to the other industries.

v Founders’ Ownership Change

This variable is a dependent variable. It represents the percentage change in founders’ own-ership pre- and post-IPO offering. Solely taking into consideration the post-offering owner-ship is not a representative variable, as the change between pre and post-offering reflects more how founders behave due to crisis’ impact on banks’ liquidity constraints. A decrease (increase) in founders’ ownership change is to be understood as founders diluting more (less) of their ownerships in an IPO.

𝐹𝑜𝑢𝑛𝑑𝑒𝑟𝑠!_{𝑂𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝 𝐶ℎ𝑎𝑛𝑔𝑒} ! =

(!"#$%&!!"_{!"#$!!""#$%&'!}!!"#$%&!!"_{!"#!!""#$%&'!})

!"#$%&!!"_{!"#!!""#$%&'!} (1)

v More Than Half

This variable is a dummy dependent variable. It takes the value of 0 if founders dilute less than half of their ownership in an IPO, and 1 otherwise. In other words, it takes the value of 0 if founders’ ownership change is bigger than -50%, and 1 if smaller than -50%.

v CEO Ownership

This control variable represents the ownership of the CEO post-offering. Some founders might want to be superior to the CEO of the firm in terms of ego, power and voting right. Therefore, the amount of shares a CEO keep post-offering might impact the decision of the founders to keep more or less shares to himself. I assume that the higher the CEO ownership is, the lesser founders would dilute their shares; thus, increasing founders’ ownership change. The natural logarithmic of this variable is used in the regressions to correct for skewness of the data.

v Company Age

This control variable represents the age of a company, from its foundation to the day of its public offering. According to Ritter (1984), younger firms are riskier. Older firms however have more information, which reduces information asymmetry, based on Ritter’s (1991) findings. As the financial crisis created a volatile and uncertain market environment, I be-lieve the younger and riskier the firm is, the more founders would dilute their ownership;

hence, decreasing founders’ ownership change. Previous respected empirical papers, such as Megginson and Weiss’ (1991) study, also used the age of the company as a control variable. The natural logarithmic of this variable is used in the regressions to correct for skewness of the data.

𝐴𝑔𝑒_! = ln 1 + 𝐼𝑃𝑂 𝑌𝑒𝑎𝑟_!− 𝐹𝑜𝑢𝑛𝑑𝑎𝑡𝑖𝑜𝑛 𝑌𝑒𝑎𝑟_! (2) v Total Assets

This control variable represents a proxy for the size of a company at the time of the public offering. Other proxies can be used to estimate the size of private firms, such as the numbers of employees or total sales. Kurshev and Strebulaev (2015) found firm size to be strongly positively related to capital structure. As equity is part of a firm’s capital structure, a firm’s size should therefore affect founders’ ownership change. I believe the bigger the firm is, the more founders will have to dilute their shares; hence, decreasing founders’ ownership change. The natural logarithmic of this variable is used in the regressions to correct for skewness of the data.

v Cash Ratio

This control variable is a measure of a company’s liquidity. It illustrates the company’s abil-ity to cover its short-term debt. A ratio above 1 is a positive sign, whilst a ratio under 1 shows an insufficiency of cash to re-pay short term debts. In this study, the variable is used to control for the amount of cash a company has before going public, as it might influence a founder’s ownership decision. I assume the more cash the company has, the lesser founders need to dilute their ownerships; thus, increasing founders’ ownership change. The natural logarithmic of this variable is used in the regressions to correct for skewness of the data.

𝐶𝑎𝑠ℎ 𝑅𝑎𝑡𝑖𝑜_!" = ln (!"#! !"# !"#! !"#!"#$%&'(!"

!"##$%& !"#!"#"$"%&!" ) (3)

v Current Ratio

This control variable is another measure of a company’s liquidity. Unlike the cash ratio, the current ratio illustrates a company’s ability to cover its short-term debts using its current

sets. Here, it is assumed that a company might not have a lot of cash, but enough current as-sets to be financially healthy. A ratio above 1 indicates that the company is financially healthy. I assume the more the company is financially healthy, the lesser founders need to dilute their ownerships; hence, increasing founders’ ownership change. The natural loga-rithmic of this variable is used in the regressions to correct for skewness of the data.

𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝑅𝑎𝑡𝑖𝑜_!" = ln ( !"##$%& !""#$"!"

!"##$%& !"#$"%"&"'(!") (4)

v Net Profit Margin

This control variable represents the profitability of a firm. The higher the number, the more profitable a firm is. This might have an impact on founder’s ownership change because it shows how good a company is at converting sales into net earnings available for sharehold-ers. And as the founder is a future potential shareholder, I assume the more the company is profitable, the lesser founders want to dilute their ownerships; therefore, increasing found-ers’ ownership change.

𝑁𝑒𝑡 𝑃𝑟𝑜𝑓𝑖𝑡 𝑀𝑎𝑟𝑔𝑖𝑛_!" =!"# !"#$%&/!"##!"

!"#$% !"#$%!" (5)

v Return on Assets

This control variable is a measure of a firm’s performance. As Kothari, Leone, and Wasley (2005) say, the return on assets is the most powerful tool to measure a company’s perfor-mance. Similar to a company’s profitability, its performance could also influence the found-er’s ownership decision. I assume the better the company is performing, the lesser founders need to dilute their ownerships; thus, increasing founders’ ownership change.

𝑅𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝐴𝑠𝑠𝑒𝑡𝑠_!" = !"# !"#$%&/!"##!"

B – IPO CHARACTERISTICS

v Underwriter Rank

This control variable embodies the prestige of the underwriters in charge of the IPO. I be-lieve this to have an impact on our dependent variable, as founders are aware of the reputa-tion of the underwriters the company going public decides to hire. I assume that the better the underwriters’ reputation, the lesser founders dilute their ownerships; thus, increasing founders’ ownership change. The ranking is based on Ritter and Loughran’s (2004) ranking and varies between 0 and 9.1, where the former is the worst and the latter is the best. As most IPOs have multiple underwriters, the average of the underwriters’ ranking is computed.

v Venture Capital

This control variable is a dummy that takes the value of 1 if the IPO is backed by a VC firm, and takes a value of 0 otherwise. VC firms are highly involved in a company’s management from all aspects. Past studies show the positive impact of a VC firm in an IPO. For instance, Megginson and Weiss (1991) find a significant positive relation between the presence of a VC firm from one side, and the cost of going public as well as the IPO net proceeds from the other side. I believe the presence of such a positively seen institution to encourage founders to keep more shares; thus, increasing founders’ ownership change.

v Underpricing

This variable will be computed for the sole purpose of observation. It will not be used in any of the regressions. It corresponds to the underpricing level using the offer price and the clos-ing price of the first public day.

𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔_! =!"#$%&'()%*+!"!!!""#$%$&'#!

C – MARKET CHARACTERISTICS

v Financial Crisis

This variable is the independent dummy variable, as I test its effect on founders’ ownership change. It takes the value of 0 when the IPO happens before August 2007, and the value of 1 when the IPO happens after February 2009.

v Loans Change

This variable is an independent variable that is only introduced and used during robustness checks. Loans are referred to as monthly commercial and industrial loans issued by U.S. commercial banks. I am interested in the monthly change of loan issuance. Forsberg (2015), and Ivashina and Scharfstein (2010) show a decrease in the amount of lending as a result of the global financial crisis.

𝐿𝑜𝑎𝑛𝑠 𝐶ℎ𝑎𝑛𝑔𝑒_! =!"#$%!!!"#$%!!!

!"#$%!!! (8)

2

–

I

P

Intangible assets are typically not visible in a company’s balance sheet because the General-ly Accepted Accounting Principles (GAAP) considers investment in intangible assets as ex-penses. However, the tangible stock of assets is considered to be the sum of a company’s Property, Plant and Equipment (PP&E) and total current assets. In order to separate our sample of firms into intangible and traditional firms, I am inspired by Peters and Taylor’s (2017) approach and construct a proxy of a company’s stock of intangible capital using a firm’s Research & Developments (R&D) expenses, and Selling, General, and Administra-tive (SG&A) expenses.

I follow Falato, Kadyrzhanova, and Sim’s (2013) claim that a firm’s stock of intan-gible capital is the sum of its knowledge capital and organizational capital. The knowledge capital of a firm is obtained from investing in R&D, and is calculated using the perpetual inventory method.

𝐾𝐶_!" is the end-of-period knowledge capital, 𝛿_!&! is the depreciation rate, and 𝑅&𝐷_!" is the R&D expenditures for the year per firm. The depreciation rate is assumed to be 15%, fol-lowing Hall, Jaffe, and Trajtenberg (2001). This study is not interested in the firms’ invest-ment in intangible capital several years after going public. Therefore, to compute (9), I only use handpicked data for the year of the IPO of each firm, as this study captures the status of the firm as close as possible from its listing in the public stock market.

A firm’s organizational capital is captured from its SG&A expenses, and also meas-ured using the perpetual inventory method.

𝑂𝐶_!" = 1 − 𝛿_!"&! 𝑂𝐶_!"!!+ 0.2 ∗ 𝑆𝐺&𝐴_!" (10) 𝑂𝐶!" is the end-of-period organizational capital, 𝛿!"&! is the depreciation rate, and 𝑆𝐺&𝐴!"

is the SG&A expenses for the year per firm. The depreciation rate is assumed to be 20%, following Lev and Radhakrishnan (2005). Logically, not all the SG&A expenses represents investments in organizational capital. According to Corrado, Hulten, and Sichel (2009), only 20% of SG&A expenses are considered as investments in organizational capital. Therefore, I leave out 80% of SG&A expenses. For similar reasons stated above, I compute (10) using handpicked data for the year of every firm’s IPO.

Choosing a value for the firm’s first missing accounting values on Compustat is a challenge. Hence, I follow Falato, Kadyrzhanova, and Sim (2013) assuming that the firm is investing in knowledge and organizational capital at a constant rate forever: 𝐾𝐶!! =

𝑅&𝐷_!!/𝛿_!&! and 𝑂𝐶_!! = 𝑆𝐺&𝐴_!!/𝛿_!"&!.

The sum of knowledge capital and organizational capital equals a firm’s stock in in-tangible capital, or 𝐼𝑛𝑡𝑎𝑛. This study uses a modified approach in comparison to Peters and Taylor (2017), which is a more reasonable and reflective approach to separate the sample into traditional and intangible firms. The physical stock of asset is considered to be the addi-tion of total current assets and PP&E. Thus, the percentage of intangible assets, %𝐼𝑛𝑡𝑎𝑛, of a company is calculated as follows:

%𝐼𝑛𝑡𝑎𝑛_!" = 𝐼𝑛𝑡𝑎𝑛!"

𝐼𝑛𝑡𝑎𝑛_!"+ 𝑃𝑃&𝐸_!"+ 𝐶𝑢𝑟𝑟𝑒𝑛𝑡𝐴𝑠𝑠𝑒𝑡𝑠_!" (11) This study has one %𝐼𝑛𝑡𝑎𝑛 value for every firm in the sample, and uses the median value to divide the sample into intangible and traditional firms.

3

–

R

I

In order to test the validity of the hypotheses formulated at the end of Section III, three dif-ferent methodologies are used. The first hypothesis is tested using an OLS regression analy-sis with fixed effects, the second one using a logistic regression, and the last one using a dif-ference-in-difference regression model with fixed effects.

Hypothesis 1: In IPOs, founders’ ownership change is more likely to be lower post-crisis than pre-crisis.

Hypothesis 1 is tested using an OLS regression model. As I am studying the financial crisis’ effect on founders’ ownership, the dependent variable is the founders’ ownership change, and the independent variable is the financial crisis dummy. By simply putting these two variables in a regression, there is a major risk of obtaining insignificant and biased re-sults. In order to diminish the omitted variable bias, I include several control variables: (i) venture capital, (ii) underwriter rank, (iii) total assets, (iv) cash ratio, (v) current ratio, (vi) net profit margin, (vii) company age, (viii) return on assets, (ix) CEO ownership, (x) indus-try fixed effect. The detail of every variable is located in Section IV-1. In order to omit out-liers, the founders’ ownership change variable is winsorised at the 1% and 99% levels.

𝐹𝑜𝑢𝑛𝑑𝑒𝑟𝑠!_{𝑂𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝 𝐶ℎ𝑎𝑛𝑔𝑒}

= 𝛼 + 𝛽_!𝐹𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙𝐶𝑟𝑖𝑠𝑖𝑠 + 𝛽_!𝑉𝑒𝑛𝑡𝑢𝑟𝑒𝐶𝑎𝑝𝑖𝑡𝑎𝑙 + 𝛽_!𝑈𝑛𝑑𝑒𝑟𝑤𝑟𝑖𝑡𝑒𝑟𝑅𝑎𝑛𝑘 + 𝛽_!𝐿𝑜𝑔𝑇𝑜𝑡𝑎𝑙𝐴𝑠𝑠𝑒𝑡𝑠 + 𝛽_!𝐿𝑜𝑔𝐶𝑎𝑠ℎ𝑅𝑎𝑡𝑖𝑜 + 𝛽_!𝑁𝑒𝑡𝑃𝑟𝑜𝑓𝑖𝑡𝑀𝑎𝑟𝑔𝑖𝑛

+ 𝛽!𝐿𝑜𝑔𝐶𝑜𝑚𝑝𝑎𝑛𝑦𝐴𝑔𝑒 + 𝛽!𝐿𝑜𝑔𝐶𝐸𝑂𝑜𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝

In the equation above, 𝛼 is constant and 𝜀 is the error term. This is the major regres-sion equation for the testing of the first hypothesis, although the regresregres-sion table will present different models. The return on assets (ROA) and the current ratio are missing from this equation for correlation reasons. Indeed, the cash ratio is highly correlated with the current ratio, and the ROA is highly correlated with the net profit margin (NPM). The two left-out variables are however included in some regression models where the cash ratio and/or the NPM are omitted. Heteroscedasticity is an issue that should be considered in the regression. I correct it by running the Eicker–Huber–White standard errors, or robust standard errors on STATA. In order to accept, or not, this hypothesis, the statistical significance of the results is examined through the t-statistics.

Hypothesis 2: In IPOs, founders are more likely to dilute at least half of their respective ownership post-crisis than pre-crisis.

Hypothesis 2 is tested using a logistic regression. The binary dependent variable is “More Than Half”, and depends on the founders’ ownership change values. It takes the val-ue of 0 if the founders diluted less than half of their ownership in an IPO, and 1 otherwise. Other variables will also be added to control for omitted variable bias, such as (i) Venture Capital, (ii) underwriter rank, (iii) total assets, (iv) cash ratio, (v) net profit margin, (vi) company age, (vii) CEO ownership.

𝑀𝑜𝑟𝑒 𝑇ℎ𝑎𝑛 𝐻𝑎𝑙𝑓

= 𝛼 + 𝛽_!𝐹𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙𝐶𝑟𝑖𝑠𝑖𝑠 + 𝛽_!𝑉𝑒𝑛𝑡𝑢𝑟𝑒𝐶𝑎𝑝𝑖𝑡𝑎𝑙 + 𝛽_!𝑈𝑛𝑑𝑒𝑟𝑤𝑟𝑖𝑡𝑒𝑟𝑅𝑎𝑛𝑘

+ 𝛽_!𝐿𝑜𝑔𝐶𝑜𝑚𝑝𝑎𝑛𝑦𝐴𝑔𝑒 + 𝛽_!𝑁𝑒𝑡𝑃𝑟𝑜𝑓𝑖𝑡𝑀𝑎𝑟𝑔𝑖𝑛 + 𝛽_!𝐿𝑜𝑔𝐶𝑎𝑠ℎ𝑅𝑎𝑡𝑖𝑜

+ 𝛽!𝐿𝑜𝑔𝑇𝑜𝑡𝑎𝑙𝐴𝑠𝑠𝑒𝑡𝑠 + 𝛽!𝐿𝑜𝑔𝐶𝐸𝑂𝑜𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝

+ 𝜀 (13)

Instead of displaying coefficients, this regression shows odds ratio as I use the “lo-gistic” instead of “logit” command on STATA. The odds ratios are simply the exponential value of the coefficients. The Eicker–Huber–White standard errors is used to correct for

het-eroscedasticity. In order to accept, or not, the hypothesis, the statistical significance of the results is be examined through p-values.

Hypothesis 3: In IPOs, the effect of the financial crisis on founders’ ownership change is stronger for intangible firms, compared to traditional firms.

Hypothesis 3 is tested with a difference-in-difference regression model. The goal is to see whether there is a statistically significant difference in founders’ ownership change between intangible and traditional firms, after the crisis. First, I use the %𝐼𝑛𝑡𝑎𝑛_!" proxy var-iable calculated in Section IV-2. Any firm under the %𝐼𝑛𝑡𝑎𝑛_!" median value is considered traditional (control group), and any firm above is considered intangible (treatment group). The financial crisis dummy takes the value of 0 for pre-crisis IPOs, and 1 for post-crisis IP-Os. A dummy variable, “Treated”, is created, that takes the value of 0 for the control group and 1 for the treatment group. In addition, an interaction variable, “DiD”, is generated from the product of the financial crisis and the treated dummies. The dependent variable is the founders’ ownership change, and the independent variables are: (i) financial crisis, (ii) treat-ed (iii), and DiD. In order to rtreat-educe omitttreat-ed variable bias, control variables are addtreat-ed to the regression. In order to omit outliers, the founders’ ownership change variable is winsorised at the 1% and 99% levels.

𝐹𝑜𝑢𝑛𝑑𝑒𝑟𝑠!_{𝑂𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝 𝐶ℎ𝑎𝑛𝑔𝑒} = 𝛼 + 𝛽_!𝐹𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙𝐶𝑟𝑖𝑠𝑖𝑠 + 𝛽_!𝑇𝑟𝑒𝑎𝑡𝑒𝑑 + 𝛽_!𝐷𝑖𝐷 + 𝛽!𝑉𝑒𝑛𝑡𝑢𝑟𝑒𝐶𝑎𝑝𝑖𝑡𝑎𝑙 + 𝛽!𝑈𝑛𝑑𝑒𝑟𝑤𝑟𝑖𝑡𝑒𝑟𝑅𝑎𝑛𝑘 + 𝛽!𝐿𝑜𝑔𝐶𝑜𝑚𝑝𝑎𝑛𝑦𝐴𝑔𝑒 + 𝛽_!𝑁𝑒𝑡𝑃𝑟𝑜𝑓𝑖𝑡𝑀𝑎𝑟𝑔𝑖𝑛 + 𝛽_!𝐿𝑜𝑔𝐶𝑎𝑠ℎ𝑅𝑎𝑡𝑖𝑜 + 𝛽_!𝐿𝑜𝑔𝑇𝑜𝑡𝑎𝑙𝐴𝑠𝑠𝑒𝑡𝑠 + 𝛽_!"𝐿𝑜𝑔𝐶𝐸𝑂𝑜𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝 + 𝜀 (14) Other models of this regression equation above are also presented. The coefficient of interest is DiD, as it is the difference-in-difference estimator. The statistical significances of the results are examined through the t-statistics.

4

–

G

F

Numerous measures are used to assess the goodness-of-fit of a regression model. The squared and adjusted squared are the most common goodness-of-fit measures. The R-squared increases the more explanatory variables are added, and is therefore not used to as-sess the models’ goodness-of-fits. This is an issue that the adjusted R-squared accounts and adjusts for. Indeed, the adjusted R-squared solely increases in value if the added control var-iables increases the model’s materiality. At its turn, the adjusted R-squared has some limita-tions, which can be overcome by the Akaike information criterion (AIC) and the Bayesian information criterion (BIC). The AIC measures the quality of each model in the regression while taking into consideration the other models’ qualities. The BIC adds a penalty for the addition of control variables. Unlike the R-squared and adjusted R-squared, the AIC and BIC are better the lesser their values are.

For each of the three methodologies, the adjusted R-squared, AIC, and BIC values are mainly used to designate the best model in terms of goodness-of-fit. These models are consequently used as the base and support for the economic interpretations of the results.

V

–

D

ATA

This section, divided into two sub-sections, details information concerning the data used in this study. First, the gathering method and the sample periods are showed. Last, a statistical description of the overall sample and subsamples are presented.

1

–

D

G

The sample is constructed using a combination of the EDGAR, SDC Thomson One, Federal Reserve Economic Data (FRED), CRSP, Ritter-Loughran-Rank, and Field-Ritter databases. I am only interested in U.S. IPOs that are/were listed on the New York Stock Exchange (NYSE) and the National Association of Securities Dealers Automated Quotations (NASDAQ).

The sample is divided into two groups; the pre-crisis period is from May 2005 until Au-gust 2007, and the post-crisis period is from February 2009 until April 2012. The dates are selected as follows:

• May 2005: 2 years after the signing of the Jobs and Growth Tax Relief Reconcilia-tion Act of 2003, which is considered to be the end of the dot-com bubble. I do not want to have a direct impact from this bubble on the sample, which is the reason why I start the pre-crisis sample 2 years after the signing of the mentioned act.

• August 2007: As Covitz, Liang, and Suarez (2013) mention, the contraction in ABCP contributes to the broader financial crisis. Indeed, the outstanding ABCP shrink by $190 billion, almost 20%, in August 2007. This date is considered to be the start of the global financial crisis.

• February 2009: It is really hard to say when the 2008 global financial crisis ended because the consequences are still seen until today. However, for the purpose of this study, I consider the end of the financial crisis to be the date of the signing of the American Recovery and Investment Act. Also it is around the time when interest rates are zero, and the U.S. government is involved in the mega bailouts.

• April 2012: The Jumpstart Our Business Startups Act (JOBS) is signed during the start of that month. For instance that act increases the number of shareholders a

com-pany may have before being required to register its common stock with the SEC and become publicly traded. I stop the sample at that date in order to avoid biased results resulting from the signing of this act.

The IPO sample is initially gathered from the SDC Thomson One database, and then adjusted according to several criteria based on previous researches. I exclude: (i) all finan-cials firms (Standard Industrial Classification (SIC) 6000-6999) and non classified estab-lishments (SIC 9999), (ii) all American depositary receipts (ADRs), (iii) all spin-off IPOs, (iv) all limited partnerships, (v) all close-ended funds, (vi) all unit issues IPOs, (vii) all in-vestment trusts (i.e. REITs), (viii) all penny stocks (offer price lower than $5). After these adjustments, the initial sample consists of 420 IPOs. This database is additionally used to see which IPOs are backed by a VC firm.

To this day, there isn’t any easily accessible and downloadable data that covers founders’ ownerships. Therefore, founders’ ownership during an IPO is handpicked from the IPO prospectus 424B forms available in the EDGAR database. I only focus on founders that were/are active in the management team and/or own shares in the company. In the sample, I do not include companies with founders that have sold all of their shares already before the IPO, founders that are no longer part of the company, founders that are deceased, or simply founders that are unknown. These exclusions are thoroughly checked at the moment of every IPO. However, I do keep companies with individuals that have inherited all the shares by being the children or grandchildren of the “true” founder.

The Ritter-Loughran-Rank and the Field-Ritter databases are respectively used to gather information concerning underwriters’ ranking and companies’ founding year. The Ritter-Loughran-Rank, as formed by Ritter and Loughran (2004), is an adjustment of Carter and Manaster’s (1990) rank. The higher the rank, the better the reputation of the underwriter. A list of prestigious underwriters can be found in Appendix – Table A.2.

I do not find a downloaded database with accounting data at the time of the IPO. Therefore, the separation of the firms into two categories, traditional and intangible firms, is created using accounting data handpicked from 10-K forms respective to every company. Initially, I aim to gather accounting data from Compustat. However, I notice a major inaccu-racy on this database concerning SG&A values; the database records the total operating

ex-penses of a company as its SG&A exex-penses. This flaw is mostly observed when the compa-ny separates its SG&A expenses into two expenses, discerning selling and marketing from one side, and general and administrative from the other. Thus, the Compustat database clear-ly lacks precision surrounding SG&A expenses values. In order to assure the values used in the study are correct, I handpick all the data for SG&A expenses of every company.

The CRSP database is used to collect the first day closing stock price of every firm in the sample. This value is helpful in order to calculate the underpricing level of every public offering. Although this computation is not required to the study’s main empirical question, as the underpricing happens after the ownership allocation to founders, I nonetheless decide to take a look at this variable and see if it follows previous empirical findings.

From the FRED database, I gather information on commercial and industrial loans issued during the sample period from U.S commercial banks. This data is solely used during robustness checks.

The final sample consists of 272 U.S. IPOs. 153 IPOs are from the pre-crisis sample period, and 119 IPOs are from the post-crisis sample period.

2

–

D

S

A – GENERAL OVERVIEW

Table I presents an overview of the final sample, consisting of 272 IPOs. Around 56% of the sample is pre-crisis IPOs, and the rest is post-crisis. I consider this fair as one of the two sample periods does not contain a much bigger and significant number of IPOs compared to the other.

When it comes to industry specific, I notice the health care and the information tech-nology industries to cover close to 65% of the sample. The rest is mainly separated between the energy, industrials, and consumer discretionary industries. A distinction between differ-ent industries is important to highlight as some have in general more intangible assets in their balance sheets compared to others. When comparing before and after crisis IPOs in dif-ferent industries, the most flagrant changes come from the energy, the consumer staples, and the telecommunication industries. After the financial crisis, the number of IPOs within the energy and telecommunication industries is reduced by at least half; whilst the consumer staples industry see a rise in its public offerings.

Sample Overview Initial Sample 420 Final Sample 272 Financial Crisis Pre-Crisis Sample 153 Post-Crisis Sample 119 Industries Energy 23 Materials 3 Industrials 27 Consumer Discretionary 31 Consumer Staples 5 Health Care 85 Information Technology 91 Telecommunications 6 Real Estate 1 Venture Capital 174

Lastly, I observe that 64% of the final sample is flagged to have its IPO backed by a VC. It is worth highlighting that the number of VC-backed IPOs after the financial crisis is 22% lower than pre-crisis.

TABLE I

SAMPLE OVERVIEW

This table presents an overview of the sample gathered from the SDC Thomson One database. First, I show the initial sam-ple that is reduced through the exclusions listed in the first part of Section V. Then, I present the amount of IPO that hap-pened during each of the two sample periods. After that, the amount of firms in every industry, as defined by the GICS, is presented. Note that there isn’t any firm from neither the utilities nor the financials sector. Finally, the last variable in this table, “Venture Capital” indicates the number of IPOs in the sample that is backed by a VC. The Industries and the VC backing information are gathered from SDC Thomson One database.

Table II displays a statistical overview of the total sample. I briefly highlight its most relevant parts, before presenting a more elaborative descriptive statistics on subsamples.

Although not a necessity to answer the main empirical question, I calculate the level of underpricing in IPOs and come with the conclusion that IPOs are on average underpriced at 14.6% level. This confirms Ibbotson et al.’s (1994) claim that IPO are on average under-priced by 10-15% after the first day of trading.

Also, I observe the average age of a company before going public to be 13.3 years, compared to a median value of 10 years. I notice a large maximum value of 106 years, which comes from a company where the founder is the child or grandchild of the “true” founder.

For the underwriters’ ranking, a mean value of 8 signifies that the sample of IPOs has on average prestigious underwriters. I define the latter with a Ritter-Loughran-Rank of at least 8.

The pre-IPO offering founders’ ownership is 23.2% on average, compared to the post-IPO offering founders’ ownership value of 16.5%. The founders’ ownership decreases on average by 23.8% during an IPO, although there are some variations seen from the stand-ard deviation value of 30.2%. Since the median is lower than the mean, the sample is skewed to the right.

Concerning the level of intangibility, the sample’s firms’ assets are on average 57.2% intangible. A standard deviation of 26% indicates that there is not much variation from the mean of the percentage of intangible assets. As I intend to have as many intangible as traditional firms in the subsamples, the median of this variable is used.

As for the accounting values and ratios (Balance Sheet Values, Income Statement Values, and Ratios), I observe large differences between the minimum and the maximum value of each variable. This is explained by the different industries present in the sample. Each industry has its own way of doing business, whether it is because of differences in cul-tures, resources, regulations, etc. Also, the question whether a company is a growing or a mature one affects its accounting values. On one hand, and in reference to a cash ratio medi-an value of 0.6, I judge the majority of firms in the sample to have insufficient cash to cover their respective short-term liabilities. On the other hand, and in reference to a current ratio median value of 1.6, the majority of the sample is in a fair position in terms of financial health. As the median is smaller than the mean for both the cash and current ratio, the sam-ple is skewed to the right with large positive outliers. The net profit margin median and mean values of 0.0% and -893.5% respectively show that the sample of firms is not at all profitable, despite some exceptions. The extensive number of new and growing companies present in the sample explains this unprofitability. The net profit margin and the return on asset variables have data skewed to the left.