Endowment and Entrepreneurial Holding of Private Equity

Transcription

1 Endowment and Entrepreneurial Holding of Private Equity Hai Huang Duke University, Durham, NC USA Abstract We study how an entrepreneur s endowment portfolio affects the proportion of his personal wealth held in the equity of his firm. Two types of entrepreneurs are studied. Endowed entrepreneurs gained ownership of their firms through inheritance or gifting. Non-endowed entrepreneurs either founded or purchased their firms. We find that, controlling for firm and entrepreneur characteristics, endowed entrepreneurs hold approximately 5% more of their personal wealth in the private equity of their firms than do non-endowed entrepreneurs. This difference in private equity holdings is highly persistent over time. Endowed entrepreneurs do not actively seek to reduce the amounts of private equity in their portfolios by withholding new equity injections, applying for new bank loans, or limiting physical investments in their firms. Endowed entrepreneurs do not take on more debt than non-endowed entrepreneurs. Instead, they finance the large investments in their firms through investing less in other types of assets, such as real estates. We explore possible explanations for the observed difference in private equity holdings between endowed and non-endowed entrepreneurs, including agency theory, difference in degrees of risk aversion, transaction costs, family ownership, status quo bias, and differences in overconfidence about their own managerial skills. Our results suggest that the status quo bias explanation is the most consistent with the data.

2 I. Introduction Small businesses constitute a vital part of the U.S. economy. There are more than five million small businesses in the United States. They employ more than 50% of all non-farm workers and account for approximately half of the private, non-farm GDP. 1 As of 1998, the total market value of all private equity in the United States was approximately $5,737 billion, about 80% of the total market value of all public equity (Moskowitz and Vissing-Jørgensen (2002)). Despite the important role private equity plays in the economy, the finance literature on this topic is relatively small. A major finding of this literature is that investments in private equity are highly concentrated. Entrepreneurs on average hold approximately 40% of their personal wealth in private equity (Moskowitz and Vissing-Jørgensen (2002), Gentry and Hubbard (2004)). Moreover, for a typical household that owns private equity, approximately 80% of this private equity is concentrated in one single firm in which the household has an active management role (Moskowitz and Vissing-Jørgensen (2002)). Thus entrepreneurs suffer from two types of under diversification. First, large fractions of their personal wealth are devoted to the equity of their firms, exposing these entrepreneurs to high idiosyncratic risks associated with the returns of their firms. Second, the returns of their human capital are intimately tied to the fortunes of their businesses. Under diversification can be very costly. Brennan and Torous (1999) estimate that, for an investor with a relative risk aversion coefficient of 2, over a ten-year horizon, investing in a single public firm results in a 64% wealth loss compared to investing in a well diversified portfolio. Meulbroek (2002) finds that, for employees with 25% of their wealth invested in company stock, over fifteen years, approximately 50% of the stock s market value will be sacrificed due to under diversification. Despite such high under diversification costs, there is no evidence that returns to private equity are on average higher than returns to public equity, nor are the returns to private equity less volatile than returns to public equity (Hamilton (2000), Moskowitz and Vissing- Jørgensen (2002)). Moreover, aggregate private equity returns are highly correlated with the aggregate returns of public equity. Thus small business owners do not appear to be adequately compensated for holding such highly concentrated portfolios. The motivating factors for entrepreneurial investments in private equity remain a subject of active research. In this paper, we study the cross-sectional variation of private equity holdings across entrepreneurs. In particular, we explore how an entrepreneur s endowment portfolio affects the 1 See the website of the United States Small Business Administration, Office of Advocacy for detailed statistics related to small businesses in the United States. 1

3 proportion of his personal wealth held in the equity of his firm. We study two types of entrepreneurs: those who obtained ownership of their firms through inheritance or gifting; and those who founded or purchased their firms. For the rest of this paper, we call the first (second) type of entrepreneurs endowed (non-endowed) entrepreneurs, and their respective firms endowed (non-endowed) firms. We find that, controlling for firm and entrepreneur characteristics, endowed entrepreneurs hold approximately 5% more of their personal wealth in the private equity of their own firms than do non-endowed entrepreneurs. This difference in private equity holdings is highly persistent. In our data, the average entrepreneur has owned his firm for more than 15 years, and there is no indication that the private equity holdings of endowed and non-endowed entrepreneurs converge over this time frame. Furthermore, there is no evidence that endowed entrepreneurs actively seek to reduce the amounts of private equity in their portfolios by withholding new equity injections, applying for new loans from financial institutions, or limiting physical investments in their firms. Endowed entrepreneurs do not take on more debt than nonendowed entrepreneurs to accommodate their relatively large own firm equity holdings. Instead, they invest less in other types of assets such as real estates. We explore six potential explanations for the difference in private equity holdings between endowed and non-endowed entrepreneurs: agency theory, differences in risk aversion, transaction costs, family ownership, status quo bias, and differences in overconfidence about their own managerial skills. Out of the six possible explanations, the status quo bias explanation is the most consistent with all the empirical facts, while the other explanations all have some difficulties explaining certain aspects of the data. The contribution of our paper is two folded. First, we show that inheritance plays an important role in explaining the cross section of entrepreneurial investments in private equity. Second, while the analysis of this paper focuses on private equity, our results have broad implications. The evidence in this paper suggests that, due to status quo bias, i.e., inertial in portfolio rebalancing, the portfolio composition of an individual s inherited wealth is an important determinant of his eventual portfolio allocation. This is a general result that should apply to all types of assets, including public and private equity, bonds, and real estates. Among other things, this paper contributes to our understanding of the prevalence and persistence of under diversification, and the household decision to participate in certain asset classes. The rest of the paper is organized as follows. Section II describes each of the six potential explanations for our results, and develops strategies to test them. Section III describes the data. 2

4 Section IV studies the effect of endowment on entrepreneurial holding of private equity, and tests the potential explanations. Section V concludes. II. Potential explanations of differences in own-firm private equity ownership Our data suggest that, controlling for firm and entrepreneur characteristics, endowed entrepreneurs hold approximately 5% more of their personal wealth in the private equity of their firms than non-endowed entrepreneurs. There exist six potential explanations for this phenomenon, including agency theory, differences in risk aversion, transaction costs of portfolio rebalancing, family ownership, status quo bias, and differences in entrepreneurs overconfidence about their own managerial skills. 2 Below we discuss each of the potential explanations and develop strategies to test them. II. A. Agency theory Models based on agency theory argue that a large entrepreneurial ownership stake can be explained by the need to align the incentives of the risk-averse owner/manager to that of outside equity investors (Holmstrom (1979), Bitler, Moskowitz and Vissing-Jørgensen (2005)). Since entrepreneurs are more risk averse than outside equity investors, the amount of equity an entrepreneur holds in his firm decreases with firm risk. In our study, endowed firms have been established for a long time, these firms are likely to have lower risks than non-endowed firms. For instance, Popkin and Kirchhoff (1991) show that, in 1976, a small business that s less than two years old has ten-year survival rate of 34.4%; while a small business that is between ten and nineteen years old has ten-year survival rate of 46.9%. The difference in firm risk may help explain endowed entrepreneurs high equity holdings in their firms. This explanation can be tested using two approaches. First, we use a proxy for firm risk developed by Bitler, Moskowitz and Vissing-Jørgensen (2005). Agency theory predicts that the equity holding difference between endowed and non-endowed entrepreneurs should disappear after properly controlling for firm risk. Second, we consider a sub-sample of entrepreneurs with 2 Another influential theory on entrepreneurial holding of private equity is the signaling theory of Leland and Pyle (1977). This theory suggests that an entrepreneur holds equity in his own firm to signal the quality of his firm to outside investors. In our study, endowed firms have been in business much longer than non-endowed firms. Outside investors have greater amount of time to learn about the quality of endowed firms. Therefore, compared to a non-endowed entrepreneur, there is much less need for an endowed entrepreneur to signal the quality of his firm through holding firm equity. This is contrary to what we found in the data. Thus, it is unlikely for the signaling theory to drive our results. 3

5 full ownership of their firms. For this sub-sample, there is no agency conflict between the owner/manager and outside equity investors. Therefore, if the difference between endowed and non-endowed entrepreneurs persists within this sub-sample, it will cast doubt on agency theory as the driving force of our main result. II. B. Differences in risk aversion Endowed and non-endowed entrepreneurs may also differ in their degrees of risk aversion. It is not obvious, however, which group of entrepreneurs is less risk averse. On the one hand, endowed entrepreneurs, having grown up in households whose fortunes are highly volatile and under diversified, may develop lower degrees of risk aversion than non-endowed entrepreneurs. On the other hand, all individuals in our sample have already self-selected to become entrepreneurs. Thus, they should all have fairly low degrees of risk aversion. Moreover, compared to endowed entrepreneurs, who inherit established firms, non-endowed entrepreneurs choose to start from scratch, an endeavor that requires high risk tolerance. Therefore, it is not clear ex ante which group of entrepreneurs is less risk averse. If endowed entrepreneurs are indeed less risk averse than non-endowed entrepreneurs, they would be willing to hold larger amounts of equity in their firms. We test this explanation by studying entrepreneurs personal portfolios and their borrowing behavior. In rational portfolio choice models, under the assumption of normal asset returns, an individual s degree of risk aversion only affects the amount of total wealth he allocates to risky assets. It does not affect the relative weights of each of the risky assets. Thus if endowed entrepreneurs hold greater amounts of private equity due to their lower degrees of risk aversion, one would expect endowed entrepreneurs to hold greater amounts of public equity and real estates as well. We would also expect endowed entrepreneurs to borrow more aggressively to finance their asset holdings. II. C. Transaction costs A third possible explanation for our results is transaction costs. When an entrepreneur inherits equity stake in a small business, it is likely to account for a disproportionally large part of his endowment portfolio. In other words, an endowed entrepreneur s inheritance portfolio is likely to be much more concentrated in private equity than his parents portfolio is. Several factors contribute to this phenomenon. First, parents tend to leave their firms to sons rather than to daughters. In the Survey of Small Business Finances data, 76.7% of endowed entrepreneurs 4

6 are male, only 23.3% are female. Moreover, recent studies on family firms have shown that family characteristics, including the number of sons and the son-daughter ratio, are important determinants of firm decisions such as family CEO successions (Bertrand et. al. (2004), Bennedsen et. al. (2005)). These evidences suggest that sons, rather than daughters, are groomed to be next generation firm owners and managers. Second, because of control issues, parents are more likely to leave their equity in a firm to one or two children rather than to split it evenly among all children. Or, even if the ownership stake was originally split evenly among all children, control considerations could prompt one child to buy out his siblings shares, which results in consolidation of private equity ownership. Finally, when the inherited wealth is sufficiently large, an endowed entrepreneur will face estate tax liabilities. The entrepreneur may use his more liquid assets to pay off the estate tax, rather than trying to sell his equity stake in the firm, which is much less liquid. Therefore, after the inheritance process, the next-generation entrepreneur s endowment portfolio tends to be more concentrated in private equity than the parents portfolio. An endowed entrepreneur may attempt to rebalance his portfolio to achieve better diversification. However, because of the illiquid nature of private equity, this rebalancing can be a very lengthy process. Therefore, the inheritance process, coupled with high transaction costs of portfolio rebalancing, can lead to the observed high private equity holdings of the endowed entrepreneurs. The transaction costs story generates several testable predictions. First, the inheritance process is influenced by an entrepreneur s family structure. The more siblings an endowed entrepreneur has, the more his inheritance portfolio will be concentrated in the equity of his firm. With transaction costs, we should observe this pattern to persist even after the endowed entrepreneur has taken over his firm for a period of time. The second prediction is that the private equity holdings of endowed and non-endowed entrepreneurs should converge over time. In other words, the more time an endowed entrepreneur has to rebalance his portfolio, the smaller the difference should be between his private equity holdings and that of a non-endowed entrepreneur. We can also test the transaction costs story by studying the effect of outside financing. Small firms have less access to outside financing, due to greater asymmetric information problem. Entrepreneurs who inherit such firms have greater difficulty rebalancing their portfolios. Therefore, the transaction costs story would predict that the difference in private equity holdings between endowed and non-endowed entrepreneurs is greater for owners of small firms. 5

7 Besides the costs associated with searching for a suitable outside investor, another type of transaction costs is tax. An endowed entrepreneur may be discouraged from selling part of his firm because of the tax liabilities associated with the realized capital gains. However, there exist several channels through which an endowed entrepreneur could reduce the amount of wealth invested in his firm without incurring capital gains tax. For example, he could withhold new equity injections into the firm, apply for new bank loans to finance capital investments, or reduce the amount of capital investments in the firm. Furthermore, these actions are not subject to the transaction costs associated with searching for an outside investor. Therefore, if an endowed entrepreneur would like to reduce his private equity holdings but does not want to incur capital gains tax, or cannot locate suitable outside investors due to the illiquidity of private equity, then we should observe the endowed firms to receive less new equity injections, less capital investments, and be more likely to apply for bank loans. II. D. Family ownership An endowed entrepreneur may be driven by the non-pecuniary benefits of owning a family business. If a company has been owned by a family for generations, an entrepreneur may feel great pride in carrying on the family tradition. Such an endowed entrepreneur may be reluctant to sell part of his ownership stake in the firm, for fear that he may lose his control rights in the firm. Similar to the discussions in Section II.C., an entrepreneur could reduce his private equity holdings as a fraction of his total wealth without losing the control rights of his firm. For example, he could withhold new equity injections, apply for new bank loans, or limit capital investments in the firm. Therefore, similar to the predictions of the transaction costs explanation, if an endowed entrepreneur would like to reduce his private equity holdings but does not want to lose the control rights in the firm, then we should observe the endowed firms to receive less new equity injections, less capital investments, and be more likely to apply for bank loans. II. E. Status quo bias The difference in private equity holdings between endowed and non-endowed entrepreneurs could also be driven by status quo bias. Status quo bias refers to the following tendency in individual decision making: when faced with choices between status quo and an alternative, individuals choose status quo over the alternative much more frequently than what would be 6

8 predicted by rational choice models (Samuelson and Zeckhauser (1988)). 3 In other words, there exist tremendous amount of inertia in individual decision making. Status quo bias can be attributed to loss aversion (Kahneman and Tversky (1979), Tversky and Kahneman (1991)). 4 These authors find that, when an individual evaluates risky choices, he is mainly concerned with changes relative to a reference point. The individual s value function is more sensitive to losses than to gains. In other words, a loss has greater impact on the individual s happiness than an equal sized gain. Due to loss aversion, people are reluctant to move away from their reference points. Status quo bias can be demonstrated using an experiment from Samuelson and Zeckhauser (1988). In this experiment, two alternatives are presented to the subjects. In the first part of the experiment, the two alternatives are presented under equal footing as follows: You are a serious reader of the financial pages, but until recently have had few funds to invest. That is when you inherited a large sum of money from your great uncle. You are considering different portfolios. Your choices are: a) Invest in moderate-risk Company A. Over a year s time, the stock has.5 chance of increasing 30% in value, a.2 chance of being unchanged, and a.3 chance of declining 20% in value. b) Invest in high-risk Company B. Over a year s time, the stock has.4 chance of doubling in value, a.3 chance of being unchanged, and a.3 chance of declining 40% in value. In the second part of the experiment, one of the two alternatives are presented as the status quo allocation: You are a serious reader of the financial pages, but until recently have had few funds to invest. That is when you inherited a large sum of money from your great uncle. A significant portion of this portfolio is invested in high-risk Company B. You are deliberating whether to leave the portfolio intact or to change it by investing in other securities. Your choices are Company A and Company B. When the two choices are presented under equal footing, 40% of the subjects choose to allocate the funds into the high-risk Company B. However, when the high-risk choice is 3 A special case of status quo bias is the endowment effect, which refers to people s tendency to demand more to part with an object than what they would be willing to pay to obtain it (Thaler (1980), Knetsch and Sinden (1984), Knetsch (1989), Kahneman, Knetsch, and Thaler (1990), Boyce et. al. (1992)). 4 For more discussions on the link between status quo bias and loss aversion, see Kahneman, Knetsch, and Thaler (1991) and Rabin (1998). 7

9 presented as status quo, 56% of the subjects choose to retain the investment in Company B. In other words, being presented as status quo greatly increases the likelihood of the high risk allocation being chosen by the subjects. Many studies have demonstrated the importance of status quo bias in making investment decisions. Using Swedish data, Cronqvist and Thaler (2004) show that, without significant advertising, more than 90% of new participants of individual retirement plans choose to invest in the default fund. Moreover, once a fund is chosen, participants rarely change their allocations. In fact, less than 10% of participants made any changes to their portfolios in the first three years after joining the plan. Similar inertia is also evident in the U.S. retirement plans (Samuelson and Zeckhauser (1988), Madrian and Shea (2001), Ameriks and Zeldes (2004)). Beyond its effect on investment decisions, status quo bias has also been shown to influence a wide range of consumer decisions, such as choice of residential electronic services (Hartman, Doane, and Woo (1991)), choice of medical treatments (Halpern, Berns, and Israni (2004)), and decision to become organ donors (Johnson and Goldstein (2003)). As discussed in Section II.C., when an entrepreneur inherits equity in a small business, it is likely to account for a disproportionally large part of his endowment portfolio. If he is subject to status quo bias, he would be reluctant to rebalance this portfolio. Therefore, status quo bias can drive an endowed entrepreneur to hold a larger fraction of his personal wealth in the equity of his firm than a non-endowed entrepreneur. The status quo bias explanation generates several predictions. Similar to the transaction costs argument, the status quo bias explanation is based on the concentration mechanism embedded in the inheritance process. Therefore, the status quo bias explanation predicts that the more siblings an endowed entrepreneur has, the more his portfolio will be concentrated in the equity of his firm. However, unlike the transaction costs story, the status quo bias explanation predicts that the private equity holding difference between endowed and non-endowed entrepreneurs does not converge to zero over time, nor is it affected by a firm s access to outside financing. Finally, contrary to the predictions of the capital gains tax story and the family ownership explanation, the status quo bias explanation predicts that endowed entrepreneurs do not actively reduce their private equity holdings through withholding new equity injections, applying for new bank loans, or limiting capital investments in their firms. 8

10 II. F. Differences in overconfidence Entrepreneurs may also differ in degrees of overconfidence about their own managerial skills. Survey evidence has shown that all entrepreneurs tend to be overconfident in their firms likelihood to achieve success (Cooper, Woo, and Dunkelberg (1988)). However, it is not clear ex ante which type of entrepreneurs is more overconfident. Arguments could be made both ways. On the one hand, an endowed entrepreneur may be more overconfident about his own skills, because he has extensive exposure to the daily operations of his small business before becoming the actual owner of the firm; while a non-endowed entrepreneur may have less experience in running a small business. On the other hand, a non-endowed entrepreneur chooses to build a brand new firm from ground up, knowing that the failure rates of small businesses are very high. So a typical non-endowed entrepreneur could have very high overconfidence. If an endowed entrepreneur is in fact more overconfident about his own managerial ability, he would invest a larger fraction of personal wealth in his firm than a non-endowed entrepreneur would. There are two methods to test this explanation. First, since both types of entrepreneurs can learn about their own abilities over time, their degrees of overconfidence should converge over time. Therefore, similar to the transaction costs story discussed in Section II.C., the overconfidence story predicts that the private equity holdings of endowed and non-endowed entrepreneurs converge over time. Second, we could study the likelihood for a firm to hire an outside manager. If endowed entrepreneurs are more overconfident about their own managerial skills than non-endowed entrepreneurs, they should be less likely to hire outside managers to run their firms. However, if endowed entrepreneurs hold greater amounts of equity in their firms due to status quo bias, there shouldn t be any systematic difference between endowed and non-endowed firms in their likelihood to hire outside managers. III. Data and summary statistics The data used in this paper are the Survey of Consumer Finances (SCF, ), and the Survey of Small Business Finances (SSBF, 1998). The Survey of Consumer Finances is conducted every three years. It provides detailed information on the household balance sheet, asset holdings, income, and other demographic characteristics of a representative sample of approximately 4,000 U.S. households. Therefore, this data allow us to study entrepreneurs 9

11 family structure, borrowing behavior, and their portfolio allocations in assets other than private equity. The SSBF gathers data from small businesses that are for-profit, non-financial, nonagricultural, non-subsidiary, and have less than 500 employees. The survey sample is designed to represent the population of about five million such small businesses in the United States. This survey provides detailed information on owner demographics and firm operations, especially the firm s financing practices. 5 In the SSBF, if a small business has multiple owners, only the characteristics of the principal owner are reported. Thus the entrepreneurs studied in this paper include only the principal owners of small businesses. The SCF differs from the SSBF in several aspects. First, the SCF reports an entrepreneur s self-estimated market value of his firm equity, while the SSBF reports book value of firm equity. Second, in the SCF, a household could have several businesses, and the method of firm acquisition is reported for actively managed firms only. In our analysis below, whenever the SCF data is used, we focus on an entrepreneur s largest actively managed firm only. An endowed entrepreneur is thus defined as an entrepreneur whose largest actively managed private business is inherited or gifted. Finally, the entrepreneurs reported in the SCF include both principal and minority firm owners. To make the SCF sample comparable to the SSBF, we limit our attention to non-farm business owners with at least 50% of the ownership shares of their firms. We exclude from our sample those observations with firm equity value less than $1000, or profit margin, i.e., profit / sales ratio, less than -100 or greater than 100. We also exclude observations with the entrepreneur s total wealth less than $10,000. Furthermore, for the SSBF data, we exclude firms with equity / assets ratio less than Our SCF sample consists of 3,171 entrepreneurs, among which 131 are endowed and 3040 are non-endowed. The SSBF sample consists of 2571 observations, with 161 endowed and 2410 non-endowed entrepreneurs. Table 1 summarizes the characteristics of the entrepreneurs and their firms. For the SCF sample, the median firm generates annual sales of $422,000. The median number of employees is six. The sample is heavily right-skewed. The average sales are $15,904,100, and the average number of employees is The average entrepreneur in the SCF sample is 52.4 years old, and has owned his firm for 15.6 years. Typically, these entrepreneurs are male. Only 4.6% of them are female. The median net worth of an entrepreneur is $1,880,600. Endowed entrepreneurs are on average 56.3 years old, four years older than non-endowed entrepreneurs. On average, 5 For more information about the sample design, please refer to the survey documentations, available at Federal Reserve s website. Also see Cole and Wolken (1995) and Bitler, Robb and Wolken (2001). 10

12 endowed entrepreneurs have owned their current firms for 21.2 years, 6.1 years longer than their non-endowed counterparts. Endowed firms are much larger than non-endowed firms. For instance, endowed firms have median annual sales of $1,300,000, while the median annual sales of non-endowed firms are $400,000. Endowed entrepreneurs are also wealthier. Their median net worth is $6,065,500, compared to $1,783,200 for non-endowed entrepreneurs. Compared to the SCF sample, the firms in the SSBF sample are generally smaller. The median firm in the SSBF sample has five employees, and generates annual sales of $300,000. The average endowed entrepreneur in the SSBF sample is 52.4 years and has owned their firms for 19.5 years. Next we study the proportion of personal wealth that an entrepreneur holds in the equity of his firm. In the SCF sample, an average endowed entrepreneur holds 40.2% of his personal wealth in the equity of his firm. This is 6.1% higher than the average non-endowed entrepreneur. Similarly, in the SSBF sample, private equity in his firm accounts for 35.9% of an average endowed entrepreneur s total wealth, compared to 25.6% for an average non-endowed entrepreneur. In both samples, the differences in private equity holdings are statistically significant with p-values less than Endowed entrepreneurs tend to have higher wealth than non-endowed entrepreneurs. To make a fair comparison between the two groups, we construct matching samples. For each endowed entrepreneur, we find a non-endowed one with the closest total wealth. In Figure 1, we plot histograms of private equity holdings of endowed entrepreneurs and that of the matching samples. As shown in Figure 1, compared to the matching samples, endowed entrepreneurs still hold greater proportions of personal wealth in their firms. Holding a highly concentrated portfolio is very costly to an endowed entrepreneur. To illustrate its economic significance, we estimate the Sharpe ratios of three portfolios, each composed of the equity of a single small firm and the S&P 500 index. The first portfolio invests 25% in the equity of a small firm; while the second (third) portfolio invests 30% (35%) in the equity of a small firm. Since the data for private equity is not available, the Sharpe ratios are estimated using public firms. To mimic a small private firm, we use public firms with beginningof-the-year total assets less than $26 million in 1998 dollars, i.e., the 90 th percentile of the total assets of the firms in our SSBF sample. From 1950 to 1998, the average monthly excess stock return of these public small firms is 1.17%; while the average monthly excess return of the S&P 500 index is 0.72%. The variances of the portfolios are estimated using monthly stock returns from 1994 to The annual Sharpe ratio for the first portfolio with 25% of assets invested in 11

13 a single small firm is When the fraction of assets invested in the small firm increases to 30% (35%), the portfolio s annual Sharpe ratio is reduced to 0.43 (0.38). The reason for such significant decline in Sharpe ratio is the high idiosyncratic risk of a small firm. When a portfolio is concentrated in one firm, the idiosyncratic risk of this small firm contributes significantly to the portfolio s total variance. When the portfolio becomes more concentrated in the small firm, the variance of the portfolio increases much faster than does the excess return of the portfolio. Thus a 5% change in the portfolio allocation has significant impact on its Sharpe ratio. IV. Empirical analysis Our univariate analysis in Section III suggests that endowed entrepreneurs tend to hold larger fractions of their personal wealth in equity of their firms than do non-endowed entrepreneurs. Now we perform multivariate analysis. Since an entrepreneur s private equity holdings, i.e., the fraction of his wealth held in private equity, is always positive, it could not be used directly as a response variable for OLS regressions. Instead, we use the natural log of the private equity holdings as the response variable. In Table 2, Panel A, columns (1) and (5), we regress the natural log of an entrepreneur s private equity holdings on Endowed, an indicator variable for endowment, controlling for firm and entrepreneur characteristics, including firm size, firm age (measured as the number of years that the firm is under current ownership), profitability, owner age, education, gender, whether or not the owner has high personal wealth, and industry dummies. 6 The SCF data do not report a firm s total assets. Therefore, in the SCF regressions, profitability is measured as profit margin, i.e., profit / sales. In the SSBF regressions, profitability is measured as ROA. In the SCF sample, as shown in column (1), the coefficient estimate for Endowed is 0.202, which suggests that, for an average endowed entrepreneur with 40.2% of his wealth devoted to private equity, if he were non-endowed, his private equity holdings would decrease by 7.4% to 32.8%. Similarly, in the SSBF sample, our results in column (5) suggest that, for an average endowed entrepreneur with private equity holdings of 35.9%, if he were non-endowed, his private equity holdings would decrease by 5.3% to 30.6%. 6 In the SCF data, firms are classified into the following industries: (1) contracting, construction, mining, oil and gas; (2) manufacturing, arts and crafts; (3) restaurants, direct sales, gas stations, food/liquor stores, other retail/wholesale; (4) auto repair, real estate, insurance, entertainment, various business services, banking and financing; (5) professional practices, beauty shops, trucking, repairs, personal services, management and consulting services, communications, writing services, transportation, educational services; and (6) other. The SSBF reports a firm s two-digit SIC industry. 12

14 It is possible that a firm s profitability can be influenced by the amount of firm equity held by the entrepreneur (Bitler, Moskowitz and Vissing-Jørgensen (2005)). Entrepreneurs with larger amounts of personal wealth tied up in their firms may put in greater efforts to manage their firms, thus resulting in higher firm profitability. To address this endogeneity issue, we instrument for profitability using Endowed, industry average profitability, owner age, owner age squared, whether or not the owner has high outside personal wealth, whether or not an entrepreneur uses his personal assets as collateral for the business (SCF only), and Rural, an indicator variable for firm location (SSBF only). The first-stage instrumented regressions are reported in Table 2, Panel B. In columns (2) and (6) of Table 2, Panel A, we use profitability estimated from the firststage instrumented regressions as a control variables. The coefficient estimates for Endowed are very similar to the results reported in columns (1) and (5). 7 IV. A. Agency theory We now explore possible explanations for why endowed entrepreneurs tend to have higher private equity holdings than their non-endowed counterparts. As discussed in Section II, there exist six possible explanations. Agency theory based models suggest that the private equity holding of an entrepreneur is determined by the trade-off between optimal risk allocation and incentive considerations (Holmstrom (1979), Bitler, Moskowitz and Vissing-Jørgensen (2005)). Since entrepreneurs are more risk averse than outside equity investors, the amount of equity an entrepreneur holds in his firm decreases with firm risk. Therefore, if the equity of endowed firms is less risky than that of non-endowed firms, endowed entrepreneurs would hold larger fractions of their personal wealth in equity of their firms. This explanation is tested using two approaches. The first approach is to control for firm risk. If agency theory and difference in firm risk drive our results, one would expect the equity holding difference between endowed and non-endowed entrepreneurs disappear after properly controlling for firm risk. To measure firm risk, we use a methodology described in Bitler, Moskowitz and Vissing-Jørgensen (2005). Using the SCF (SSBF) sample, we first run a cross-sectional regression of firm ROE (ROA) on a constant, log of total number of employees, log of total firm equity, number of years the firm is under current ownership and its squared term, industry dummies, whether or not the entrepreneur uses personal assets as collateral for the business, 7 In the regressions reported in Tables 2, 4, 5 and 6, ROA or profit margin estimated from the first-stage instrumented regression is included as an explanatory variable. We also repeated the regressions in these tables using ROA or profit margin directly as an explanatory variable. Our results are qualitatively unchanged. 13

15 (whether or not the firm s sales are concentrated in local region, whether or not the firm is an exporter), and entrepreneur demographics, including gender, education, race, and length of entrepreneur s business experience. The absolute value of the residual of this regression is our measure for firm risk. 8 In Table 2, Panel A, columns (3) and (7), we show that controlling for firm risk does not take away the effect of endowment. The coefficient estimates for Endowed are very similar to the results reported in columns (2) and (6). The second approach to test the agency theory explanation is to consider the sub-samples of entrepreneurs with full ownership of their firms. For these sub-samples, there is no agency conflict between the entrepreneur and outside equity investors. The regression results using these sub-samples are reported in columns (4) and (8) of Table 2, Panel A. Our results suggest that, within these sub-samples, the magnitudes of the coefficient estimates for Endowed are actually greater than the coefficient estimates obtained from the full sample. For example, the result in column (4) suggests that, for the SCF sub-sample of entrepreneurs with full ownership of their firms, changing an average endowed entrepreneur to a non-endowed one would reduce his private equity holdings by 9.4%, from 40.2% to 30.8%. Similarly, for the SSBF sub-sample, changing the endowment status of an endowed entrepreneur would lead to a 7.9% reduction in private equity holdings. Taken together, these results suggest that the difference in private equity holdings between endowed and non-endowed entrepreneurs cannot be attributed to agency costs and the differences in firm risks. IV. B. Differences in risk aversion The second explanation for our main result is differences in risk aversion. Endowed entrepreneurs, with extensive prior exposure to the risks associated with owning small businesses, may become less risk averse than non-endowed entrepreneurs. This difference in risk aversion could drive endowed entrepreneurs to hold greater amounts of private equity. We test this explanation by studying entrepreneurs personal portfolio allocations. If endowed entrepreneurs are indeed less risk averse, they should hold greater amounts of all risky assets, including private equity, public equity, and real estates. One would also expect endowed entrepreneurs to take on more debt to finance their asset holdings. 8 In Bitler, Moskowitz and Vissing-Jørgensen (2005), the cross-sectional regression of ROE (ROA) includes Endowed as an explanatory variable. In our construction of Risk, Endowed is excluded from the list of explanatory variables. Thus if endowed and non-endowed firms differ in firm risk, this would be captured by the Risk variable. 14

16 We use data from the SCF to study entrepreneurs personal portfolios. As shown in Table 3, Panel A, more than 70% of an average entrepreneur s household wealth is invested in private equity and real estates. Contrary to the predictions of the risk aversion explanation, public equity accounts for 10.1% of an average endowed entrepreneur s household wealth, compared to 11.7% for an average non-endowed entrepreneur. Endowed entrepreneurs on average invest 30.9% of their personal wealth in real estates, 16.0% less than non-endowed entrepreneurs. Moreover, endowed entrepreneurs do not borrow more aggressively than non-endowed entrepreneurs. Their average debt / wealth ratio is 11.0%, substantially less than 25.0%, the average debt / wealth ratio of non-endowed entrepreneurs. In Table 3, Panel B, we perform regression analysis to study the effect of endowment on an entrepreneur s real estates holdings and debt / wealth ratio, controlling for entrepreneur and firm characteristics, such as the entrepreneur s age, education, gender, wealth, firm size, firm age, profit margin, industry dummies, and survey year dummies. Since 4.2% of the entrepreneurs in our sample do not own any real estates, and 17.9% of the observations have zero debt, we perform Tobit regressions in Table 3, Panel B. In columns (1) and (2), the response variable is log(1+realestates/wealth). The log transformation is performed to deal with the outliers in the sample. For the regression reported in column (2), with firm and entrepreneur characteristics at their mean values, the marginal effect for Endowed is In other words, holding all other variables at their sample means, an endowed entrepreneur holds approximately 4% less of his total wealth in real estates than a non-endowed entrepreneur. In columns (3) and (4), the response variable is log(1+debt/wealth). For the regression reported in column (4), with firm and entrepreneur characteristics at their mean values, the marginal effect for Endowed is In other words, holding all other variables at their sample means, the debt / wealth ratio of an endowed entrepreneur is approximately 4% less than that of a non-endowed one. Taken together, the results in Table 3 are not consistent with the predictions of the risk aversion explanation. IV. C. Transaction costs The third possible explanation for our results is transaction costs. When an entrepreneur inherits a small business, his inheritance portfolio is likely to be much more concentrated in the equity of his firm than his parents portfolio was. Because of the illiquid nature of private equity, it takes a long time for an endowed entrepreneur to reduce the amount of private equity in his portfolio. 15

17 The inheritance process is influenced by an entrepreneur s family structure. The more siblings an endowed entrepreneur has, the more his inheritance portfolio will be concentrated in the equity of his firm. With transaction costs, we should observe that, even after the endowed entrepreneur has taken over his firm for a period of time, the degree to which his portfolio is concentrated in the equity of his firm should increase with the number of siblings he has. We test this prediction using data from the Survey of Consumer Finances, which reports an entrepreneur s number of living siblings at the time of the survey. Since an entrepreneur s actual number of siblings at the time of inheritance could differ from the number of living siblings at the time of the survey, we limit the sample to entrepreneurs less than 55 years old. Our results are reported in Table 4. In columns (1) and (2) of Table 4, we introduce the variable Siblings, i.e., an entrepreneur s number of living siblings, and the interaction term between Endowed and Siblings. The coefficient estimates for Endowed*Siblings are positive and statistically significant, suggesting that the difference between an endowed and a non-endowed entrepreneur increases with the number of siblings an entrepreneur has. It is possible that there exists non-linearity in the relation between an entrepreneur s private equity holdings and the number of siblings. To deal with this possibility, in columns (3) and (4) of Table 4, we measure the number of living siblings using the variable Siblings2, which is equal to Siblings when its value is less than 3, and set to 3 when the entrepreneur has at least 3 living siblings. The results are similar to those reported in columns (1) and (2). The transaction costs story is not consistent with all aspects of the data, however. It predicts that the more time an endowed entrepreneur has to rebalance his portfolio, the smaller the difference should be between his private equity holdings and that of a non-endowed entrepreneur. In our data, the endowed entrepreneurs on average have already held their firms for more than 19 years. In other words, an average endowed entrepreneur has close to two decades to rebalance his personal portfolio. In Table 5, we study whether the difference in private equity holdings between the two groups of entrepreneurs changes over time. In columns (1) and (3), we interact Endowed with FirmAge, i.e., the number of years that the firm is under current ownership. In columns (2) and (4), firms are classified into four age categories: FirmAge less than 10 years; from 10 to 20 years; from 20 to 30 years; and greater than 30 years. We then interact Endowed with FirmAge_Cat, the firm s age category. In Table 5, none of the coefficient estimates for Endowed*FirmAge or Endowed*FirmAge_Cat is statistically significant. These results suggest 16

18 that, contrary to the prediction of the transaction costs story, the private equity holdings of endowed and non-endowed entrepreneurs do not converge over time. We can also test the transaction costs story by studying the effect of outside financing. An endowed entrepreneur may search for suitable outside investors to unload some of his private equity holdings. Small firms have more severe asymmetric information problems when it comes to attracting outside investors. Therefore, the transaction costs story would predict that the difference in private equity holdings between endowed and non-endowed entrepreneurs is greater for smaller firms. To test this prediction, we introduce the indicator variable SmallSize, set to 1 if and only if a firm s sales are less than the median values, and the interaction term between Endowed and SmallSize as explanatory variables for private equity holdings. The results are presented in Table 6. The transaction costs story predicts that the coefficient estimates of the interaction term Endowed*SmallSize to be positive and statistically significant. The results in Table 6 do not support this prediction. In all of the regressions reported in Table 6, the coefficient estimates for the interaction term Endowed*SmallSize are negative and statistically indistinguishable from zero. These results suggest that a firm s access to outside financing, as proxied by firm size, does not affect the difference of private equity holdings between endowed and non-endowed entrepreneurs. If an entrepreneur intends to reduce the amount of private equity in his portfolio, there are several ways to do so. He could try to sell part of his equity holdings to outside investors, pay himself higher compensation or take out more dividends, or withhold new equity injections and seek bank loans to finance his firm. He could also control the firm size by withholding new capital investments. Among these approaches, selling firm equity to outside investors incurs the highest amount of transaction costs, both in terms of the search costs of locating an outside equity investor, and in terms of capital gains taxes, while the other approaches involve less transaction costs. Therefore, if an endowed entrepreneur wants to reduce the amount of private equity in his portfolio, but face high transaction costs, he may withhold new equity injections, apply for new bank loans, or limit new investments in his firm, leading to slower firm growth. In the first three columns of Table 7, we study the likelihood for a firm to receive new equity injections in the three years prior to the survey. The data are from the SSBF. Since the SSBF does not report whether the new equity injections are from the principal owners or from the minority owners, we limit the sample to those firms that are fully owned by their principal owners. The likelihood of new equity injection can be influenced by both supply and demand for new equity. Entrepreneurs with greater wealth outside of the firm are more capable of providing 17

19 new equity (Holtz-Eakin et. al. (1994), Blanchflower and Oswald (1998), Dunn and Holtz-Eakin (2000)); while firms with lower profitability are in greater demand for new equity injection to remain afloat. An entrepreneur s decision to inject new equity can also be influenced by the availability of bank financing. Earlier studies show that small businesses tend to receive more favorable treatments from those financial institutions with which they have longer business relationships and those financial institutions that are located in closer geographical proximity (Petersen and Rajan (1994), Berger and Udell (1995)). More recently, Petersen and Rajan (2002) show that, due to the developments in the financial sector and the greater availability of information, the physical distance between financial institutions and firms are less important. In the regressions reported in Table 7, we study the relation between endowment and the likelihood of new equity injections, controlling for the entrepreneur s outside wealth, ROA, firm location (measured by the minimum distance between the firm and those financial institutions that the firm has business relationships with), and the maximum length of the firm s relationship with financial institutions, as well as other owner and firm characteristics. The results in Table 7, columns (1) to (3) suggest that entrepreneurs tend to inject new equity into firms that have poor ROA. Probability of new equity injection is negatively associated with the number of years that a firm is under current ownership. In other words, firms that have been recently founded or acquired, are more likely to receive new equity injections. The coefficient estimates for Endowed are not statistically significant. In other words, endowed entrepreneurs do not hold back new equity injections in their firms. In columns (4) to (6) of Table 7, we study the relation between endowment and the likelihood for a firm to apply for new bank loans in the three years prior to the survey, controlling for the entrepreneur s outside wealth, ROA, firm location (measured by the minimum distance between the firm and the financial institutions that the firm has business relationships with), and the maximum length of the firm s relationship with financial institutions, as well as other owner and firm characteristics. We focus on applications for new bank loans, which do not include renewal of existing loans or extension of existing line of credits. The data are also from the SSBF. The coefficient estimates for Endowed are not statistically significant in these regressions. In other words, compared to the non-endowed firms, the endowed firms are no more likely to apply for new bank loans. In unreported regressions, we also show that the two types of firms do not differ in the amounts of new loans applied for, or the amounts of new loans approved. Firm growth rates are studied in Table 8. We regress firms sales growth rates on Endowed, controlling for other entrepreneur and firm characteristics. We find no evidence that endowed 18