Two Essays in Finance: Has Momentum Lost its Momentum, and Venture Capital Liquidity Pressure and Exit Choice. Debarati Bhattacharya

Transcription

1 Two Essays in Finance: Has Momentum Lost its Momentum, and Venture Capital Liquidity Pressure and Exit Choice Debarati Bhattacharya Dissertation submitted to the Faculty of the Virginia Polytechnic Institute and State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Business, Finance Raman Kumar, Co-Chair Ozgur S. Ince, Co-Chair Dilip K. Shome Arthur J. Keown March 05, 2014 Blacksburg, Virginia Keywords: Momentum, Venture Capital, IPO Copyright 2014, Debarati Bhattacharya

2 Two Essays in Finance: Has Momentum Lost its Momentum, and Venture Capital Liquidity Pressure and Exit Choice Debarati Bhattacharya ABSTRACT My dissertation consists of two papers, one in the area of investment and the second in the area of corporate finance. The first paper examines robustness of momentum returns in the US stock market over the period 1965 to We find that momentum profits have become insignificant since the late 1990s partially driven by pronounced increase in the volatility of momentum profits in the last 14 years. Investigations of momentum profits in high and low volatility months address the concerns about unprecedented levels of market volatility in this period rendering momentum strategy unprofitable. Past returns, can no longer explain the cross-sectional variation in stock returns, even following up markets. We suggest three possible explanations for the declining momentum profits that involve uncovering of the anomaly by investors, decline in the risk premium on a macroeconomic factor, growth rate in industrial production in particular and relative improvement in market efficiency. We study the impact of venture capital funds (VC) liquidity concerns on the timing and outcome of their portfolio firms exit events. We find that VC funds approaching the end of their lifespan are more likely to exit during cold exit market conditions. Such late exits are also less likely to be via initial public offerings (IPO). A one standard deviation increase in the age of a VC fund at the time of the exit event is associated with a 5 percentage points decline in the probability of an IPO vs. a trade sale from an unconditional probability of roughly 30%. Several tests indicate that the decline in IPOs with VC fund age is not caused by lower portfolio firm quality. Focusing on the aftermath of IPOs, VC-backed firms experience significantly larger trading volume and lower stock returns around lock-up expirations if they are backed by older funds, and this lock-up effect is amplified if there are multiple VC firms approaching the end of their lifespan. Altogether, our results suggest that the exit process is strongly influenced by VCs liquidity considerations.

3 Acknowledgements I didn t realize how difficult it would be to write this part of my dissertation. Now don t get me wrong, it is not for a lack of ability to articulate my emotions and I do have a long list of people without whose support I couldn t do this. It s just that the last five years with the finance department at Tech have been more than writing a dissertation, it has been quite the journey. Leaving a career and family thousands of miles away, coming back to school after a good seven years hiatus wasn t easy. But I loved every second of the challenge despite some days that were more glum than others. Dilip K. Shome admitted me to the program and I am grateful to him for giving me this opportunity. My sincerest appreciation goes out to the co-chairs of my dissertation committee, Raman Kumar and Ozgur Ozzie Ince. I cannot thank them enough for everything that I have learnt from them. Raman s support has gone beyond the scope of an advisor on various projects to that of a compassionate mentor taking care of me through some very tough personal times. Ozzie has always had his doors open for me to walk in and brainstorm ideas. Dilip provided a safe place for me to discuss almost anything, topics ranging from how to get my committee together to decide on the future course of action towards completion of my thesis to the latest movies, albums and books. Arthur J. Keown always had an encouraging thing or two to say about my research and teaching and went out of his way to help me during my job search process. In addition, I am grateful to Vijay Singal for his support. I have often asked him for advice and he always took a sincere interest in helping me. I also appreciate the advice that I got from John Easterwood when I first started teaching at Tech. I also want to thank Gokhan Sonaer who is truly a great friend. We have worked on several co-authored papers and will continue to do so in the future. We have learned a lot together and the contribution of Gokhan in my learning process is invaluable. Terry Goodson, the soft and sincere woman I met five years ago has become one of my closest friends. I do not have words to express my love and gratitude for her friendship. Wei-Hsein Lee, Hong Yang, Jitendra Tayal, Mete Tepe, Nan Qin and Jaideep Chowdhury have also been great friends and support over several years during the program. My deepest appreciation goes out to my family. My parents, Amalendu and Supriya, have given me so much and have expected nothing in return other than my happiness. They iii

4 are my biggest cheerleaders and they have cheered me on no matter what till I reached my goal and for that I will forever be grateful. Debopriyo, my young brother, a man of few words have told the whole world but me how proud he is of his sister. My friend Atish who is almost family has egged me on at times when I went through some of my existential phases (who doesn t get some of those while getting a PhD?) and wanted to throw in the towel. Through the past years I have come to realize that how lucky I am to have this kind of family support. It has made all the difference. iv

5 Attribution I am grateful to my co-authors, Raman Kumar and Gokhan Sonaer in the first paper of my dissertation for their invaluable comments and advice. I am also indebted to my co-author, Ozgur Ince in the second paper of my dissertation. He has been involved in the paper right from the start. He has helped me develop the idea and has been a constant support. v

6 Table of Contents Paper I Has Momentum Lost its Momentum? Introduction Disappearance of momentum profits since Holding period returns: Evidence from subperiods Seasonality and holding period returns Extreme volatility and holding period returns since Holding period return in a 14-year rolling window analysis: Evidence from Market cycles and holding period returns Holding period returns for small firms, large firms, low liquidity, and high liquidity firms Cross sectional variation in returns explained by past returns Cross sectional variation in returns explained by past returns in the intermediate horizon Possible explanations for the disappearance of momentum profits since Uncovering of anomaly by investors Reduced risk premium on macroeconomic variable Relative market efficiency Pre and Post 1999 Periods Conclusion 23 References Paper II Venture Capital Liquidity Pressure and Exit Choice Introduction VC liquidity pressure hypothesis Data and summary statistics Sample selection Variable definitions and summary statistics Timing of VC exits Exit choice Baseline results in exit choice Identification Instrumental variable approach 62 vi

7 Matched sample approach VC age and portfolio firm quality Which funds succumb to liquidity pressure Liquidity pressure at IPO lock-up expirations Conclusion...72 References 2.74 vii

8 List of Tables Table 1.1 Table 1.2 Table 1.3 Table 1.4 Table 1.5 Table 1.6 Table 1.7 Table 1.8 Table 1.9 Table 2.1 Table 2.2 Table 2.3 Table 2.4 Table 2.5 Table 2.6 Table 2.7 Table 2.8 Table 2.9 Momentum portfolios raw returns for 6-month/6-month strategy..31 Fama-French three-factor alphas of momentum portfolios for 6-month/6- month strategy..32 Momentum portfolios returns in times of extreme volatility for the period Momentum portfolios returns following periods of low and high volatility...35 Momentum profits over 14-year rolling window for the period 1965 to Momentum portfolios raw returns following Up and Down markets.38 Momentum portfolios raw returns for 6-month/6-month strategy size and liquidity...39 Fama-MacBeth regressions of stock returns on past 11 months cumulative returns, β, size, and BE/ME.40 Measures of delay for the three sub-periods 41 Summary statistics..79 Number of months between investment and exit, by fund Age at investment 81 OLS analysis of time between VC investment and exit...82 Probit analysis of exit market conditions.83 Exit choice - Probit analysis.84 Exit choice - 2SLS analysis..85 Exit choice - Propensity score matching..86 Liquidity Pressure and Fund Incentives..87 Liquidity pressure at IPO lockup expiration 88 viii

9 List of Figures Figure 1.1 Figure 1.2 Figure 1.3 Figure 2.1 Figure 2.2 Figure 2.3 Figure A.1 Average Winner-Loser Portfolio Returns by Year...26 Comparison of Distribution of Momentum Portfolios Returns following Up Markets...27 Buy and Hold Abnormal Returns of New Entrants to Winner and Loser Portfolios-Event Study...30 Histogram of exits by VC age categorized by exit method.76 Predicted probability of IPO based on observable quality proxies..77 Acquired firm characteristics by VC age at exit..78 VC investment by Washington State Investment Board between December 2002 and December ix

10 Paper I Has Momentum Lost Its Momentum? (Co-authored with Raman Kumar and Gokhan Sonaer) 1.1 Introduction Momentum in stock prices has been shown to be a persistent market anomaly in the past. Jegadeesh and Titman (1993) were the first to document that a trading strategy that longs winner stocks and shorts loser stocks generates significant profits over a holding period of 3-12 months, later labeled in the literature as momentum. Some advocates of market efficiency, however, suspected these observed regularities in returns arise because of data snooping. In a follow up study, Jegadeesh and Titman (2001) respond to such skepticisms by showing that momentum strategy continues to generate abnormal returns in the 1990s. Momentum has grown in its popularity ever since in the finance community that includes both the academics and practitioners. Some of the recent works in the area of market anomalies, such as McLean and Pontiff (2013) asks an interesting question of whether or not academic research could potentially destroy return predictability. 1 In this paper, we investigate whether momentum profits have been driven away or at the very least its pattern altered in the wake of growing knowledge about momentum strategy and competition amongst arbitrageurs who trade on it, if we were to believe momentum profits were caused in the first place due to delayed price reactions to firm-specific information as suggested by Jegadeesh and Titman (1993, 2001). What if momentum is no longer profitable? The answer to this question makes this paper important. It is needless to say that the disappearance of momentum profits, if proven to be true 1 Hwang and Rubesam (2008) build an inter-temporal model that explains momentum returns allowing for structural breaks over an extended sample period They document that momentum profits have slowly started declining in the last two decades of their sample period, a process that began in the early 90 s but delayed by the occurrence of high-technology stock bubble. 1

11 would have an impact over a number of interest groups in the capital market, such as the traders in forming strategies, the investors on how to evaluate their money managers performance, and academics on how they perceive and explain the disappearance of this flagrant affront to the idea of rational, efficient markets. This paper could potentially trigger an entirely different debate on why has momentum disappeared in the context of the rich literature that exists on its persistence and rationale, both behavioral and rational. Our analyses span over the period between 1965 and We divide the entire time period into three subperiods. The first subperiod corresponds to the Jegadeesh and Titman (1993) sample period, 1965 to 1989, the second subperiod covers the Jegadeesh and Titman (2001) out of sample period, 1990 to 1998, the third subperiod corresponds to the period 1999 to In our study, we choose to examine the persistence of momentum profits while avoiding concerns of data dredging by conducting tests in our out-of-sample period that starts at the beginning of 1999 immediately after Jegadeesh and Titman s (2001) out of sample period ends. Using the data over the 1999 to 2012 sample period, we find that Jegadeesh and Titman (1993) momentum strategies fail to yield profits in the more recent times. This period is particularly interesting as it witnessed the dot-com bust after catching the boom by its tail and also the financial crisis followed by the greatest stock market meltdown since the great depression. One of our concerns in dealing with this unique period is what if the recent turbulence in the economy with a series of high-loss episodes in the US stock market and unprecedented levels of market volatility has rendered momentum strategy unprofitable? We employ alternate methodologies to scrutinize whether the rapid decline of momentum profits to insignificant levels in this 14 year period is indeed an outcome of the marked rise in market volatility. For instance, we use controls for the periods of unusual volatilities in the 2

12 capital market, 2007 to 2009 in particular and yet fail to reject the hypothesis that momentum profits have not declined to insignificant levels. Excluding the last financial crisis, 2007 to 2009 serves the additional purpose of excluding spring of 2009 that witnessed the biggest momentum crash in the history of stock market since the summer of 1932 as alluded to by Daniel and Moskowitz (2012). Next, we employ the daily median volatility index, VXO for the period 1986 to 1998 to classify months in the latest subperiod into high and low expected volatility months. 2 If momentum profits have declined because of increased volatility of the market, momentum strategy should be profitable at least in months when the implied volatility is as low as in low volatility months in the period 1986 to1998, a period when momentum is profitable. However, what we document is that while momentum strategy is profitable in the period 1986 to 1998 no matter the implied volatility, it fails to generate profit for the period 1999 to 2012 even in the 60 months classified as low volatility months primarily clustered between November 2003 and July We also investigate whether momentum profits resurface in this period following up markets as documented by Cooper, Gutierezz and Hameed (2004). Not only are these momentum profits insignificant on average following up markets, their distribution also reveal visible and statistical difference from those in the periods 1965 to 1989 and 1990 to 1998, indicating a deeper and more fundamental change in the underlying process of generation of momentum profits, beyond huge market crashes. The distribution of up market momentum profits in this period is extremely volatile interspersed with huge negative returns that suggest that momentum as a strategy has become riskier in the latest subperiod compared to the two earlier subperiods. Further analysis indicates that the idiosyncratic volatility of momentum portfolio returns has increased compared 2 We use VXO instead of VIX since the former that is computed using a different methodology and eventually revised by CBOE provides us with an additional 4 years worth of data. 3

13 to the previous periods. We also examine whether cumulative past returns can explain the crosssectional variation in stock returns. In the presence of return continuation, we expect past stock returns to be positively related to current stock returns, especially following up markets since momentum profits are essentially up market phenomena. As expected in the periods 1965 to 1989 and 1990 to 1998, current stocks returns are positively related to past returns exclusively following up markets. However, in the current subperiod, with decline in momentum profits past returns fail to explain current returns following up markets and show a reliably negative relation following down market. We suggest three possible explanations for the declining momentum profits that involve uncovering of the anomaly by investors, decline in the risk premium on a macroeconomic factor, growth rate in industrial production in particular, and relative improvement in market efficiency. The first explanation proposes that momentum profits decline post 1998 because investors become increasingly aware about the profitability of implementing a relatively simple momentum trading strategy, wherein they identify winner (loser) stocks and buy (sell) them. The growing awareness and competition amongst these investors would lead to an increasingly earlier identification and trading of momentum stocks. This explanation predicts intensified reaction to both winner and loser stocks in the identification period itself, which would result in either exhaustion or, at the least, a substantial reduction in return continuation in the holding period. 3 We find evidence consistent with this prediction. The second explanation is based on the findings of Liu and Zhang (2008) who document that growth rate of industrial production, in various specifications, explains over half of the 3 Reducing underreaction or mispricing may also result in similar patterns of returns from loser and winner portfolios, if we were to believe momentum profits were caused in the first place due to delayed price reactions to firm-specific information as suggested by Jegadeesh and Titman (1993, 2001). The distinction between uncovering of anomaly by investors and reducing undereaction is beyond the scope of this paper. 4

14 momentum profits. We find that in the latest subperiod although the momentum portfolio s returns continue to load on this industrial production factor, this particular risk factor is no longer priced. The third explanation explores the possibility of relative improvement in market efficiency. Following Griffin, Kelley, and Nardari (2010), we compute their DELAY measure, that reflects the degree of response of stock returns to past market returns, and we record a fairly significant reduction in delay in all size portfolios but for the largest one. 1.2 Disappearance of momentum profits since 1999 Our sample is constructed from all common stocks traded on New York Stock Exchange (NYSE), American Stock Exchange (AMEX), and Nasdaq. We obtain the data related to the stock market from the Center for Research in Security Prices (CRSP) database, and accounting data from Standard and Poor s (S&P s) Compustat. We exclude all stocks priced below $5 at the beginning of the holding period and all stocks with market capitalizations smaller than that of the lowest NYSE size decile following Jegadeesh and Titman (2001). Our analyses span over the period between 1965 and We divide the entire time period into three subperiods. The first subperiod corresponds to the Jegadeesh and Titman (1993) sample period, 1965 to 1989, the second subperiod covers the Jegadeesh and Titman (2001) out of sample period, 1990 to 1998, and the third subperiod corresponds to the period 1999 to 2012.We choose our third sample subperiod adhering to standard model validation practice and testing the hypothesis of persistence of momentum profits in our out-of-sample period that starts at the beginning of 1999 immediately after Jegadeesh and Titman s (2001) out of sample period ends. 5

15 1.2.1 Holding period returns: Evidence from subperiods In this section we examine whether momentum strategies continue to be profitable since the late 1990s. Jegadeesh and Titman (2001) document that their out of-sample tests designed to assess persistence of momentum profits in the 1990s performed at least as well as the ones conducted with the original sample in their earlier study in It has been a while since money managers and traders at large have acceded to the claims that momentum strategies generate substantial profits, and we have concurrently seen a phenomenal growth in the size of funds in their hands. Hedge funds managed about $1.64 trillion in 2011 up from $ 200 billion in 1998 and equity mutual funds managed about $13 trillion at year-end 2012 up from $5.5 trillion in These developments raise a fairly obvious question. Has momentum survived this new era of the capital markets? Our tests reveal strong evidence of momentum profits in the first, less strong evidence in the second consistent with the literature, and decline in momentum profits to insignificant levels in the third subperiod. Following Jegadeesh and Titman (1993), we examine the profitability of 16 strategies that select stocks based on the their returns over the past 3, 6, 9, and, 12 (J) months and hold them for either 3, 6, 9, or 12 (K) months in each of our three subperiods. At the end of each month (t), we sort stocks into 10 equally weighted portfolios based on their cumulative returns earned in the past J months (t J + 1 to t). We hold these portfolios for K months (t + 1 to t + K). As a result we have K overlapping portfolios each of which is assigned an equal weight in the portfolio. We also construct a momentum strategy portfolio that buys the winner portfolio (top past return 4 Sourced from McKinsay s Global Institute forecasts, HedgeFundFacts.com and ICIFACTBOOK.ORG. 6

16 decile) and sells the loser portfolio (bottom past return decile). Similar to Jegadeesh and Titman (2001) we compute the portfolio returns using data from the CRSP monthly returns file. Next, we compute the Fama-French three-factor alphas (Fama and French, 1993) earned by the winner, loser and momentum (winner-loser) portfolios for all the 16 (J-month/K-month) strategies. Our investigation reveals that over the periods 1965 to 1989 and 1990 to 1998, the returns for all the momentum strategies are positive and statistically significant confirming the other known results as in Jegadeesh and Titman (1993) and Jegadeesh and Titman (2001). However, for the 1999 to 2012 period none of the 16 momentum strategies delivers any returns different from zero. The risk adjusted profit analysis also confirms that for all the 16 (J-month/K-month) strategies with a few exceptions the alphas of the loser portfolios are negative whereas the alphas of the winner portfolios are positive for the periods 1965 to 1989 and 1990 to Momentum portfolios for all strategies earn statistically significant alphas for these two subperiods. In the period 1999 to 2012, none of the past return deciles earn alphas significantly different from zero and the alpha of momentum portfolio also disappears. 5 Following Jegadeesh and Titman (1993) we now examine the six month formation/ six month holding strategy in more detail. Table 1.1 presents the average monthly raw returns for the 10 past return portfolios. At the end of each month (t), we sort stocks into 10 equally weighted portfolios based on their cumulative returns earned in the past six months (t - 5 to t). We hold these portfolios for the next six months (t + 1 to t + 6). This process presents us with six overlapping portfolios each of which is assigned an equal weight in the portfolio. We also 5 These results are not reported for the sake of brevity, but they are available upon request. 7

17 construct a portfolio following momentum strategy that buys winner (top past return decile) and sells loser (bottom past return decile). Table 1.1 shows that the average returns increase as we go from the lowest to the highest deciles for all the three subperiods. The momentum portfolio (P10-P1) on average earns a 1.10 % per month in the period 1965 to 1989 that continues in the period 1990 to Consistent with the findings of Jegadeesh and Titman (2001), the momentum portfolio in the second subperiod earns 1.37% a month. However, as noted earlier in Table 1, the momentum returns decline to insignificant levels in the period 1999 to Table 1.2 presents the alphas for the 10 past return portfolios. Past losers P1 earn negative alpha and past winners P10 earn positive alpha in the periods 1965 to 1989 and 1990 to The momentum portfolio (P10-P1) on average earns an alpha of 1.27% per month in the period 1965 to 1989 and 1.35% per month in the period 1990 to However, neither the past loser, or past winner or the momentum portfolios earn any alphas in the period 1999 to 2012 that are significantly different from zero Seasonality and holding period returns We examine whether the January effect on momentum profits reported by Jegadeesh and Titman (1993, 2001) have become pronounced in the period 1999 to 2012 so much so that the momentum profits in the non-january months are overshadowed. The momentum profits in January for our sample are no different from zero over the period The momentum profits for the non-january months are, however, positive and significant for the periods 1965 to 6 George and Hwang (2004) find that proximity to the 52-week high predicts the future returns significantly better than past returns. However we find that the 52-week high strategy does not work, exactly as the momentum strategy in the last subperiod. 8

18 1989 and 1990 to 1988 but significantly in the period 1999 to The evidence indicates that there has not been any significant change in the absence of momentum profits Extreme volatility and holding period returns since 1999 The post 1998 period, during which we document significant decline in momentum profits, experienced stretches of extreme stock market volatility as it witnessed the dot -com bust after catching the boom by its tail and also the financial crisis followed by the greatest stock market meltdown since the great depressions. We acknowledge the importance of controlling for these periods of unusual volatilities. Table 1.3 presents the monthly average returns for 10 portfolios formed on the basis of the past 6 months cumulative returns and held for another 6 months, earned in six separate time periods post The first two columns report the returns for the periods 1999 to 2005, and 2006 to 2012, dividing the post 1998 period into two halves. The first two columns of the table reveal that the momentum portfolios (P10-P1) earn no profit in the first as well as the second half of our last subperiod. The third column reports the returns for the period 1999 to 2012 excluding the last financial crisis, 2007 to 2009, a period that also includes spring of 2009, the biggest momentum crash in the history of stock market since the summer of 1932 as alluded to by Daniel and Moskowitz (2012). The fourth column reports the returns for the period 2004 to 2012, excluding the tech boom and bust, 1999 to 2003 as well as the last financial crisis. These columns do not reveal any resurfacing of momentum profits, and it is especially interesting to find no momentum in the period 2004 to 2012 (excluding 2007 to 2009) since the market showed an upward trend in these years, a condition favorable for generating momentum profit. 7 These results are not reported for the sake of brevity, but are available upon request. 9

19 We employ an alternate methodology to scrutinize whether the rapid decline of momentum profits to insignificant levels in this 14 year period is indeed an outcome of the marked rise in market volatility. We obtain daily levels of volatility index, VXO available for the period 1986 to 2012 from the website of Chicago Board of Options Exchange, CBOE. The daily median implied volatility for the period 1999 to 2012 jumps to from in the period 1986 to 1998 consistent with the common knowledge that market volatility in the latest subperiod reached higher levels compared to the previous two subperiods. We classify months in the latest subperiod into high (low) volatility months if the monthly mean volatility, VXO is above (below) the daily median VXO for the period 1986 to months get classified as low volatility months primarily clustered between November 2003 and July 2007 and 108 months get classified as high volatility months. If momentum profits have declined because of increased volatility, momentum strategy should be profitable at least in months when the implied volatility is as low as in low volatility months in the period 1986 to1998, a period when momentum is profitable. However, what we document in Table 1.4 is that while momentum strategy is profitable in the period 1986 to 1998 no matter the implied volatility, it fails to generate profit for the period 1999 to 2012 even in all of the 60 months classified as low volatility months. This evidence suggests it is not the unprecedented levels of market volatility that has rendered momentum strategy unprofitable in the last 14 years Holding period return in a 14-year rolling window analysis: Evidence from Presented with all the initial evidence of disappearing momentum profits, a well-founded question in the reader s mind maybe: Has there been any other 14 year stretch in the past over which the momentum strategy has not been profitable? 10

20 We perform a 14-year rolling window analysis in which we compute the average raw and risk-adjusted momentum returns for every 14 years starting at the beginning of each year from In Table 1.5 we document that starting from 1965 for no other 14 year period until 1996, momentum strategy was ever unprofitable. The momentum profits are not significantly different from zero only over the 14 year periods starting in 1996, 1997, 1998, and Figure 1.1 plots the monthly average returns for each year to the momentum portfolio from 1965 to Post the tech bubble, other than 2002, 2005, and 2007 the momentum return is either negative or close to zero. 8 For those who would still like to ascribe the disappearance of momentum profits to housing crisis of we would like to point out that the period was as good and as bad for momentum strategy, as is evident from the figure, if one were to concentrate only on the highest and lowest return years, 2000 and 2009 respectively. Moreover, as shown in Table 1.4 earlier excluding these years make no difference to our inference that there is no more any momentum effect in stock prices Market cycles and holding period returns Cooper, Gutierezz, and Hameed (2004) document that momentum profits are significant following up market conditions. In this section we examine whether momentum profits reappear once controlled for the up and down market cycles. Following Cooper, Gutierrez, and Hameed (2004), we classify the months following a phase of 36 months of positive (negative) value weighted CRSP index returns as up (down) markets. Table 1.6 presents the monthly average returns for 10 portfolios formed on the basis of the past 6 months cumulative returns and held for another 6 months earned following up and down market conditions. The results indicate that 8 We are aware that momentum returns peaked during 1999 and 2000 riding on the internet bubble. In spite of that we include these years in our last subsample since Jegadeesh and Titman (2001) s out-of-sample period ends in 1998, after which our out-of-sample period begins. 11

21 momentum portfolios (P10-P1) earn significant profits following up markets but they earn no profits reliably different from zero following down markets in the periods 1965 to 1989 and 1990 to 1998 confirming earlier findings. The period 1990 to 1998 experienced no down market conditions and this can partially explain, the larger momentum profit in this period recorded above compared to the period 1965 to However, in the period 1999 to 2012, momentum portfolios do not earn any profit significanlty different from zero, regardless of market conditions. Not only are these momentum profits insignificant on average following up markets, their distribution also turns out of to be visibly and statistically very different from those in the first and the second subperiods indicating a deeper and more fundamental change in the underlying process of generation of momentum profits, beyond huge market crashes. Figure 1.2 plots and compares the distribution of monthly returns of momentum portfolios (winners-losers), following up-markets. The solid line represents a fitted normal distribution and the dashed line represents fitted kernel density, estimated with bandwidth parameter of Panel A plots the distributions of monthly returns of these momentum portfolios in the periods 1965 to 1989 and 1999 to 2012 and Panel B plots the same for the periods 1990 to 1998 and 1999 to Momentum profits in the last subperiod show larger dispersion as compared to the two previous subperiods that may explain the lack of statistical significance of the average momentum returns following up markets in this subperiod. Momentum as a strategy seems to have become riskier in the most recent subperiod. Kuiper two sample tests that are used to assess the uniformity of a set of distributions show that these distributions are significantly different from each other. Panel C plots the distributions of monthly returns of momentum portfolios following up markets in the periods and The distributions look similar indicating comparable riskiness of the momentum strategy in the first two subperiods. The 12

22 Kuiper tests confirm that these two distributions are not significantly different from one another. The idiosyncratic volatility of the momentum portfolio has increased in the latest subperiod compared to the previous two subperiods combined which may be contributing towards the overall rise in volatility of momentum returns. We calculate the variance of the residuals from Fama-French 3-Factor model regression of momentum returns for each sub-period and conduct an F-test to compare the statistical significance of the difference Holding period returns for small firms, large firms, low liquidity, and high liquidity firms It is quite possible that momentum strategy continues to be profitable among smaller and lower liquidity stocks for the simple reason that they are more expensive to trade. To address this possibility, in this subsection we separately examine the momentum returns generated by small and large stocks, and also by high and low liquidity stocks. Following Jegadeesh and Titman (2001), the Small Cap group (Large Cap) comprises of stocks that are smaller (larger) than the median NYSE stock by market capitalization at the beginning of the holding period. 9 Illiquidity is estimated as ratio of absolute one day return to dollar volume in that particular day, a measure proposed by Amihud (2002). Low (High) Liquidity stocks have higher (lower) average illiquidity than the median illiquidity stock in the month preceding the identification period (t - 6). We use the liquidity measure as of the sixth month before the holding period to make liquidity sorting process independent from the past return sorting process. The results in Table 1.7 indicate that the momentum effect that was prevalent in all size and liquidity categories till 1998, decline uniformly across all these groups of stocks in the period 1999 to We repeat our analysis with size subsamples formed on the basis of the market capitalization at the beginning of the identification period to make the size sorting process more independent from the past return sorting process and this has no effect on inferences. 13

23 1.2.7 Cross sectional variation in returns explained by past returns In the subsections above, we have provided evidence that momentum strategies no longer earn significant returns since 1999, and that these results are robust to various controls for seasonality, extreme volatilities and cycles in the capital market. However, financial market anomalies are patterns in security returns not only in time series but also in the cross-section that are not predicted by the central theory of asset pricing. We suspect that with declining return continuation to relative strength portfolios, the past returns can no longer explain cross-sectional variation in stock returns. To investigate whether past returns explain stock returns in the cross section, we adopt the methodology employed by Fama and French (1992). We carry out Fama-MacBeth regressions of monthly returns of individual stocks on its past cumulative returns (t - 12 to t - 2) controlling for post ranking beta, size, and book-to-market equity. The only accounting ratio used in the regressions is the natural logarithm of book-to-market equity, ln(be/me). BE is the book value of common equity plus balance-sheet deferred taxes, and ME is the market equity. BE is obtained for each firm's latest fiscal year ending in calendar year t 1 and BE/ME is computed using market equity (ME) in December of year t - 1. However, firm size, the natural logarithm of market equity ln(me) is measured in June of year t. The explanatory variables for individual stocks are matched with CRSP returns for the months from July of year t to June of year t + 1. The gap between the accounting data and the returns ensures that the accounting data are available prior to the return. Following Fama and French (1996), the cumulative past returns for each stock, each month are computed by cumulating their returns from t - 12 to t - 2 months. Individual stocks are assigned post-ranking β of the size-β portfolio that they are in at the end of June of year t. We compute the post-ranking βs as in Fama and French (1992). Each June all 14

24 NYSE stocks are sorted based on market equity to determine NYSE size decile cut -off points. Then, all NYSE, AMEX and NASDAQ stocks that have data both on CRSP and COMPUSTAT are assigned to these size deciles based on NYSE cut -off points. We sort stocks in each size decile, based on their pre-ranking βs. The pre-ranking βs are estimated using t - 24 to t - 60 monthly stock returns. The equal weighted average monthly returns of the 100 size-β portfolios are computed over 12 months following June of each year and the post-ranking βs for these 100 size-β portfolios are estimated for the full period. We use Fowler and Rorke (1983) correction in estimating the βs. Table 1.8 presents the results of these Fama-MacBeth regressions. These results clearly demonstrate that a positive relation between current and past stocks returns exists for the periods 1965 to 1989 and 1990 to 1998, but is no longer significant in the period 1990 to This confirms our postulate that as momentum returns decline to insignificant levels, past returns can no more explain cross-sectional variation in stock returns. The regressions also show that market β does not help explain average stock returns for the entire sample period confirming the results of Fama and French (1992). The small firm effect prevails through the first two subperiods, though relatively weaker in the post 1989 period. However, it is subsumed by the book-tomarket. The value stocks on the other hand continue to outperform growth stocks over the entire sample period. The results are consistent with the existing literature on widely known stock market anomalies. Momentum profits have been linked to market states in the literature. We earlier presented evidence that momentum profits are insignificant on average following 3-year up markets in the 10 We also include natural logarithm of asset-to-market and asset-to-book ratios as explanatory variables instead of natural logarithm of book-to-market in the regressions and this does not have bearing on our inferences. 15

25 1999 to 2012 period, in contrast to the two previous subperiods. We also examine whether past returns explain stock returns in the cross-section after controlling for market states. We carry out Fama-MacBeth regressions of monthly returns of individual stocks as in Table 8, splitting the subperiods into up and down market states this time. The results confirm all our previous findings. In the periods 1965 to 1989 and 1990 to 1998, past stocks return is positively related to current stocks returns exclusively following up markets. However, in the current subperiod, past returns fail to explain current returns following up markets and show a reliably negative relation following down market. So with decline in momentum profits, past returns do not show the expected positive relation with current stock returns Cross sectional variation in returns explained by past returns in the intermediate horizon Novy-Marx (2012) concludes that the recent past performance does not matter as much as the past performance within the intermediate horizon, in particular the cumulative returns 12 to 7 months prior to formation (t - 12, t - 7). We carry out Fama-MacBeth regressions of monthly returns of individual stocks as in Table 8, only this time using the cumulative returns of stock over the intermediate horizon. In the periods 1965 to 1998, intermediate past stocks return is positively related to current stocks returns. However, in the 1999 to 2012 period, past intermediate returns fail to explain current returns. Hence, with decline in momentum profits, past returns, no matter whether measured over the recent past or the intermediate horizon do not show the expected positive relation with current stock returns These Results not presented for the sake of brevity, but they are available upon request. 12 These results are not tabulated for the sake of brevity, but they are available upon request. 16

26 1.3 Possible explanations for the disappearance of momentum profits since 1999 We suggest three possible explanations for the declining momentum profits that involve uncovering of anomaly by investors, disappearance of the risk premium on industrial production factor, and improvement in relative market efficiency. The first explanation proposes that momentum profits decline post 1998 because investors learn about the benefits of implementing a naive strategy called momentum thereby correcting mispricing if any in the firms identified as winners and losers within the identification or the formation period faster in the last subperiod compared to the earlier subperiods. This explanation predicts intensified reaction to both winner and loser stocks in the identification period itself, which would result in either exhaustion or, at the least, a substantial reduction in return continuation in the holding period, and weakened return reversal (under the scenario of possible overreaction in the holding period perpetrated by behavioral biases) in the post holding period. We find evidence consistent with all these predictions. However, a caveat is order here; reducing underreaction or mispricing may also result in similar patterns of returns from loser and winner stocks, if we were to believe momentum profits were caused in the first place due to delayed price reactions to firm-specific information as suggested by Jegadeesh and Titman (1993, 2001). The distinction between the two is beyond the scope of this paper. The second explanation is based on the findings of Liu and Zhang (2008) who show that macroeconomic factors such as growth rate of industrial production are priced and in various specifications explains over a half of the momentum profits. We however, find that in the latest subperiod the marginal productivity factor is no longer priced. The third explanation explores the possibility of improvement in relative market efficiency. Following Griffin, Kelley, and Nardari (2010), we use the delay in order to assess the 17

27 improvement in market efficiency that measures the degree of response of stock returns to past market returns. We record a fairly significant reduction in delay in all size portfolios but for the largest one that suggests improvement in relative market efficiency Uncovering of anomaly by investors The first explanation proposes that investors simply recognize that momentum strategy is profitable and trade in ways that arbitrage away such profits partially consistent with Schwert (2003) that documents two primary reasons for the disappearance of an anomaly in the behavior of asset prices, first, sample selection bias, and second, uncovering of anomaly by investors who trade in the assets to arbitrage it away. Competition amongst arbitrageurs to buy the winners and short the losers would induce them to try to identify the winners and losers earlier and earlier. Earlier identification and execution of the momentum strategy in the latter part of the identification period itself would reduce, and eventually eliminate the abnormal returns in the holding period. Moreover, the incentive and the competition amongst the arbitrageurs to unwind the long and short trades before any losses due to any possible over-reaction in the holding period would eventually eliminate any systematic over-reaction and subsequent reversals. It is also interesting to note that Brav and Heaton (2002) point out even if irrationality perpetrates financial anomalies, their disappearance hinges on rational learning, an ability of rational arbitrageurs to identify observed price patterns and wipe out any return potential in excess of risk based expectations. This explanation predicts intensified reaction to winner and loser stocks in the identification period itself, exhaustion or, at the least, a substantial reduction in return continuation in the holding period, and weakened return reversal (under possible overreaction in the holding period perpetrated by behavioral biases) in the post holding period. 18

28 To test these implications of growing investor awareness, we compute the buy and hold abnormal returns of new winner and loser stocks during the identification period and in the following 24 months. New winners (losers) are the stocks that enter the winner (loser) portfolio in month t. Abnormal return for each event month is the average of the mean abnormal returns of all stocks with monthly return data for 30 months, t - 5 to t + 24, across all calendar months. Buy and hold abnormal return is the difference between the cumulative raw return and cumulative expected return for each stock for each event month. The expected returns are computed using the loadings on Fama-French three factors over the five year period between t - 71 to t Stocks with less than 24 monthly observations are excluded for the purpose of estimation of the three factor loadings. Figure 1.3 presents the plots of the buy and hold abnormal returns. The buy and hold returns for the winner stocks in the identification period, months t -5 to t show that in the post 1998 period they reach substantially higher levels on average spiraling at a much faster rate compared to the pre 1999 period and they eventually flatten out in the holding and post holding periods, months t + 1 to t Even though the graph for the buy and hold return of winner stocks in the post 1998 period may suggest return continuation for a few months in the post holding periods, months t+3 to t+10 in particular, none of these returns are statistically significant. Very similar pattern is exhibited by the returns of loser stocks. However, front running the traditional momentum traders on the short end seems more difficult to implement. This is not a surprising finding in light of the existing literature that associates higher asymmetry of information, transaction costs and other short trade restrictions We also analyze the risk-adjusted 24 month post holding period returns of the winner and loser portfolios that show substantial reversal consistent with overreaction and subsequent price correction hypothesis until Post 1998, there is no evidence for either return continuation or subsequent reversal. 19

29 1.3.2 Reduced Risk Premium on Macroeconomic Variable As mentioned earlier Liu and Zhang (2008) show that macroeconomic factors such as growth rate of industrial production are priced and in various specifications explains over a half of the momentum profits. If however, in the last subperiod the marginal productivity factor is no more a priced risk factor then that could provide an explanation to the disappearance of momentum profits. To that end, we first compute the loadings of loser, winner, and winner-loser portfolios returns on the growth rate of industrial production. We use monthly regressions of these portfolio returns for estimating the loadings on the Fama-French three factors and the growth rate of industrial production (MP). =log as defined in Liu and Zhang (2008), where is the index of industry production in month t from the Federal Reserve Bank of St. Louis. Momentum portfolio continue to load significantly positive on this factor in the period as in the period. Next, we the estimate of the risk premium of MP from two-stage Fama-MacBeth (1973) cross-sectional regressions. Following Liu and Zhang (2008) in the first stage, we estimate factor loadings using sixty-month rolling-window regressions and extending-window regressions. For the rolling window, the starting month for the estimation is t - 60 and the ending month is t. For the extending window the starting month for the estimation is always January 1965 and the ending month is t. In the first stage, we run regressions of monthly excess returns of 30 testing portfolios on Fama-French three factors and the MP. 30 testing portfolios consist of ten size, ten book-to-market, and ten six/six momentum portfolios. 14 In the second stage, we perform crosssectional regressions of 30 testing portfolios t + 1 month excess returns on the factor loadings estimated in the first stage using information up to month t. We start the second-stage regressions 14 The ten size and ten book-to-market portfolio data are from Kenneth French s web site. 20

30 in January The risk premium of MP is computed by taking the average of the coefficients on the MP loadings from the second-stage cross-sectional regressions. The MP risk premium is positive and significant in the first two subperiods combined. However, neither for the rolling window nor for the extending window analysis is the MP risk premium significantly different from zero indicating that the industrial growth rate factor is no longer priced, a plausible cause for disappearing momentum profits Relative Market Efficiency Pre and Post 1999 Periods Post 1998, neither of the winner or the loser portfolios earn returns that are reliably different from zero in the post identification period. The lack of return continuation and subsequent reversal in the post identification period can be interpreted as an evidence of improvement of market efficiency in the period 1999 to The markets might have become more efficient because information gets impounded into prices faster in this period. Following Griffin, Kelley, and Nardari (2010), we examine improvement in relative market efficiency using the DELAY measure that reflects the degree of response of stock returns to past market returns. DELAY is computed by subtracting the adjusted R 2 of unrestricted market model from the adjusted R 2 of the restricted market model (Delay = ). The unrestricted model uses four lags of weekly market returns:,where is the weekly portfolio (individual stock) return at time t and is the market return. In the restricted model, the coefficients on the lagged market returns are constrained to zero: 15 These results are not reported for the sake of brevity, but they are available upon request. 21

31 Table 1.9 presents the results for DELAY for the 5 size quintiles for our sample of stocks. In Panel A, weekly returns of five size portfolio are the dependent variables in the market model. Weekly returns are the equal weighted portfolio returns for the size quintiles. All stocks in our sample are sorted into quintiles at the end of previous year. DELAY across all size quintiles declines substantially except for the largest portfolio. The smallest size quintile experiences an 88% reduction in delay between the second and the last subperiod. The numbers for the other quintiles are fairly large though they decrease monotonically from the smallest to the largest quintile. The results are not surprising since the larger stocks suffer a lot less from problems of information asymmetry, constitute a big part of the market itself, hence their prices respond to market wide news a lot faster. In Panel B, weekly returns of individual stock are the dependent variables in the market model. For each size quintile, we then compute the average DELAY. We also report the difference between the average DELAY of each subperiods and the corresponding p-values. We record a fairly significant reduction in DELAY in all size quintiles but between the second and the third subperiod in particular other than the third largest and largest portfolios As indicated by Griffin, Kelley and Narrdari (2010), delay measures may be subject to larger estimation error noise for individual firms but in order test the statistical significance of delay measures across the three subperiods we have to use delay measure at the stock level. 22

32 1.4 Conclusion In this paper we ask the question what if momentum which has been shown to be a persistent market anomaly is no longer profitable? The contribution of this paper lies in the answer to this question. It cannot be stressed enough that the disappearance of momentum profits, if proven to be true would have a significant impact over a number of interest groups in the capital market, such as the traders in forming strategies, the investors on how to evaluate their money managers performance, and academics on how they perceive and explain the disappearance of such a persistent market anomaly. This paper evaluates the persistence of momentum or lack thereof over the last half a century. We document that trading strategies, which buy past winners and sell past losers, though remarkably profitable up until 1998, fail to generate significant abnormal returns in the period 1999 to These results are robust across extreme size and liquidity subsamples of stocks, periods of unusual volatilities in the capital market, seasonality, and up and down market conditions. We also document that past returns either in the long run or within the intermediate horizon can no longer explain cross-sectional variation in stock returns in the post 1998 period. We suggest three possible explanations for the declining momentum profits that involve uncovering of the anomaly by investors, decline in the risk premium on a macroeconomic factor, growth rate in industrial production in particular, and relative improvement in market efficiency. In support of these explanations, we conduct an event study, the results of which hinge on investor learning. We document decline in risk premium of industrial growth to insignificant levels, and we also conduct traditional relative market efficiency tests, the results from which suggest that market information gets incorporated faster into stock prices. 23

33 References 1 Amihud, Y. "Illiquidity and stock returns: cross-section and time-series effects." Journal of financial markets 5 (2002), Brav, A., and J. B. Heaton. "Competing Theories of Financial Anomalies." Review of Financial Studies 15 (2002), Cooper, M. J., R. C. Gutierrez, and A. Hameed. "Market States and Momentum." The Journal of Finance 59 (2004), Daniel, K., and T. J. Moskowitz Momentum crashes. Working paper, SSRN elibrary. Fama, E. F., and K. R. French. "The Cross-Section of Expected Stock Returns." The Journal of Finance 47 (1992), Fama, E. F., and K. R. French. "Common risk factors in the returns on stocks and bonds." Journal of Financial Economics 33 (1993), Fama, E. F., and K. R. French. "Multifactor Explanations of Asset Pricing Anomalies." The Journal of Finance 51 (1996), Fowler, D. J., and C. H. Rorke. "Risk measurement when shares are subject to infrequent trading : Comment." Journal of Financial Economics 12 (1983), George, T. J., and C. Y. HWANG. "The 52 Week High and Momentum Investing." The Journal of Finance 59 (2004), Hwang, S., and A. Rubesam. "The Disappearance of Momentum." SSRN elibrary (2008). Jegadeesh, N., and S. Titman. "Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency." The Journal of Finance 48 (1993), Jegadeesh, N., and S. Titman. "Profitability of Momentum Strategies: An Evaluation of Alternative Explanations." The Journal of Finance 56 (2001),

34 Liu, L. X., and L. Zhang. "Momentum profits, factor pricing, and macroeconomic risk." Review of Financial Studies 21 (2008), McLean, R. D., and J. Pontiff. "Does Academic Research Destroy Stock Return Predictability?" In AFFI/EUROFIDAI, Paris December 2012 Finance Meetings Paper (2013). Novy-Marx, R. "Is momentum really momentum?" Journal of Financial Economics 103 (2012), Schwert, G. W. "Chapter 15 Anomalies and market efficiency." In Handbook of the Economics of Finance,Volume 1, Part B, M. H. G.M. Constantinides and R. M. Stulz, eds.: Elsevier (2003). 25

35 Figure 1.1 Average winner-loser portfolio returns by year This figure plots the average monthly returns of winner - loser portfolios for each year during the Winner-loser portfolios are constructed using the methodology as described in Table Average Winner-Loser Portfolio Returns by Year

36 Figure 1.2 Comparison of distribution of momentum portfolios returns following Up markets Panel A and This figure plots the distribution of monthly returns of winner- loser portfolios, constructed as described in Table 1 following up-markets as defined in Table 6 for the first and the most recent subperiods. The solid line represents a fitted normal distribution and the dashed line represents fitted kernel density, estimated with bandwidth parameter of Distribution of Winners-Losers Monthly Returns 27

37 Figure 1.2-Continued Comparison of distribution of momentum portfolios returns following Up markets Panel B and This figure plots the distribution of monthly returns of winner - loser portfolios, constructed as described in Table 1 following up-markets as defined in Table 6 for the second and the most recent subperiods. The solid line represents a fitted normal distribution and the dashed line represents fitted kernel density, estimated with bandwidth parameter of Distribution of Winners-Losers Monthly Returns 28

38 Figure 1.2-Continued Comparison of distribution of momentum portfolios returns following Up markets Panel C and This figure plots the distribution of monthly returns of winner - loser portfolios, constructed as described in Table 1 following up-markets as defined in Table 6 for the first and the second subperiods. The solid line represents a fitted normal distribution and the dashed line represents fitted kernel density, estimated with bandwidth parameter of Distribution of Winners-Losers of Frequency (%) Frequency (%) Monthly Returns 29

39 Buy and Hold Abnormal Returns Buy and Hold Abnormal Returns Figure 1.3 Buy and hold abnormal returns of new entrants to winner and loser portfolios-event study This figure plots the abnormal buy and hold returns of new entrants to winner and loser portfolios (constructed as in Table 1) over t -5 to t Our initial sample includes all NYSE, AMEX and NASDAQ stocks priced above $5 at the beginning of the holding period and with market capitalizations above the cut - off level of lowest NYSE decile. New winners (losers) are the stocks that enter the winner (loser) portfolio in month t and are not included in the winner (loser) portfolios in any of the months t -5 to t -1. Abnormal return for each event month is the average of the mean abnormal returns of all stocks with monthly return data for 30 months, t -5 to t+24, across all calendar months. Buy and hold abnormal return is the difference between the cumulative raw return and cumulative expected return for each stock for each event month. The expected returns are computed using the loadings on Fama-French three factors over the five year period between t -71 to t Stocks with less than 24 monthly observations are excluded for the purpose of estimation of the loadings on the three factors. Winners Losers Event Month Event Month

40 Table 1.1 Momentum portfolios raw returns for 6-month/6-month strategy This table presents the average monthly returns earned by momentum portfolios constructed with all NYSE, AMEX and NASDAQ stocks after excluding stocks priced below $5 at the beginning of the holding period and stocks with market capitalizations less than the cut -off level of lowest NYSE decile. At the end of each month (t) stocks are sorted into 10 equally weighted portfolios based on their cumulative returns earned in the past six months (t-5 to t). This table reports the mean of monthly average returns to these ten portfolios formed on the basis of the past 6 months cumulative returns and held for another 6 months for the three periods, , , and The bottom two rows of this table present the average returns and the corresponding p-values to the winnerloser portfolios that buy winners (highest past return decile) and sells losers (lowest past return decile). All the portfolios are equal weighted P1 (Past Losers) P P P P P P P P P10 (Past Winners) P10-P1 (Winners-Losers) p-value ( ) ( ) ( ) 31

41 Table 1.2 Fama-French three-factor alphas of momentum portfolios for 6-month/6-month strategy This table presents the Fama-French three-factor alphas earned by momentum portfolios constructed with all NYSE, AMEX and NASDAQ stocks after excluding stocks priced below $5 at the beginning of the holding period and stocks with market capitalizations less than the cut -off level of lowest NYSE decile. This table reports the alphas earned by the ten portfolios formed on the basis of the past 6 months returns and held for another 6 months in a Fama-French three-factor OLS regression for the three periods, , , and The bottom two rows of this table present the alphas and the corresponding p-values to the winner-loser portfolios that buys winners (highest past return decile) and sells losers (lowest past return decile). All the portfolios are equal weighted. P-values are in parentheses (SP1) (SP2) (SP3) P1 (Past Losers) ( ) ( ) (0.3726) P ( ) ( ) ( ) P ( ) ( ) ( ) P ( ) ( ) ( ) P ( ) ( ) ( ) P ( ) ( ) ( ) P ( ) ( ) ( ) P ( ) ( ) ( ) P ( ) ( ) ( ) P10 (Past Winners) ( ) ( ) (0.3628) P10-P1 (Winners-Losers) ( ) ( ) (0.3315) 32

42 Table 1.3 Momentum portfolios returns in times of extreme volatility for the period This table presents the average monthly returns earned by momentum portfolios constructed with all NYSE, AMEX and NASDAQ stocks after excluding stocks priced below $5 at the beginning of the holding period and stocks with market capitalizations less than the cut -off level of lowest NYSE decile. At the end of each month (t) stocks are sorted into 10 equally weighted portfolios based on their cumulative returns earned in the past six months (t -5 to t). Panel A reports the monthly average returns for these ten portfolios formed on the basis of the past 6 months cumulative returns and held for another 6 months for the following periods: , , excluding , , excluding The bottom two rows of this table present the average returns and the corresponding p-values to the winner-loser portfolios that buy winners (highest past return decile) and sells losers (lowest past return decile). All the portfolios are equal weighted. Panel B reports the three factor alphas and the corresponding p-values to these ten portfolios formed on the basis of the past 6 months cumulative returns and held for another 6 months for the following periods: , , excluding , , excluding Panel A. Raw Returns (excluding ) (excluding ) P1 (Past Losers) P P P P P P P P P10 (Past Winners) P10-P1 (Winners-Losers) p-value ( ) ( ) ( ) ( ) 33

43 Table 1.3-Continued Momentum portfolios returns in times of extreme volatility for the period Panel B. Fama-French Three-Factor Alphas (excluding ) (excluding ) P1 (Past Losers) ( ) ( ) ( ) ( ) P ( ) ( ) ( ) ( ) P ( ) ( ) ( ) ( ) P ( ) ( ) ( ) ( ) P ( ) ( ) ( ) ( ) P (0.6819) ( ) (0.9618) ( ) P ( ) ( ) (0.5543) ( ) P ( ) ( ) ( ) (0.1489) P ( ) ( ) ( ) ( ) P10 (Past Winners) ( ) ( ) (0.1086) ( ) P10-P1 (Winners-Losers) ( ) ( ) (0.2132) ( ) 34

44 Table 1.4 Momentum portfolios returns following periods of low and high volatility Panel A of this table presents the average monthly returns earned by momentum portfolios constructed with all NYSE, AMEX and NASDAQ stocks after excluding stocks priced below $5 at the beginning of the holding period and stocks with market capitalizations less than the cut-off level of lowest NYSE decile. At the end of each month (t) stocks are sorted into 10 equally weighted portfolios based on their cumulative returns earned in the past six months (t -5 to t). Panel A of this table reports the mean of monthly average returns to the P1 (Losers), P10 (Winners), and P10-P1 (Winners-Losers) portfolios formed on the basis of the past 6 months cumulative returns and held for another 6 months for the two time periods, , and These sub-periods are further segregated into high and low volatility periods based on the median daily VXO of the period (18.35). Panel B of this table presents the Fama-French three-factor alphas earned by the P1 (Losers), P10 (Winners), and P10-P1 (Winners-Losers) portfolios over the low and high liquidity periods for the two time periods, , and All the portfolios are equal weighted. P-values are presented in parentheses. Panel A. Raw Returns Low Volatility High Volatility P1 (Past Losers) P10 (Past Winners) P10-P1 (Winners- Losers) p-value ( ) ( ) ( ) ( ) Panel B. Three-Factor Alphas P1 (Past Losers) P10 (Past Winners) Low Volatility High Volatility ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) P10-P1 (Winners- Losers) ( ) ( ) ( ) ( ) 35

45 Table 1.5 Momentum profits over 14-Year rolling window for the period 1965 to 1999 This table presents the results a 14-year rolling window analysis in which we compute the average raw and riskadjusted momentum returns for every 14 years starting at the beginning of each year from Panel A reports the raw returns and Panel B reports the Fama-French three-factor alphas. P-values are presented in parentheses. Panel A. Raw Returns Starting Starting P1 P10 P10-P1 Year Year P1 P10 P10-P ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (0.0000) ( ) ( ) ( ) ( ) ( ) (0.0000) ( ) (0.0199) ( ) ( ) ( ) (0.0000) ( ) ( ) ( ) ( ) (0.0037) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (0.001) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (0.008) ( ) (0.0012) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (0.0000) ( ) ( ) (0.0123) ( ) ( ) (0.0000) ( ) ( ) ( ) ( ) ( ) (0.0000) ( ) (0.0636) ( ) ( ) ( ) (0.0000) ( ) ( ) ( ) ( ) ( ) (0.0000) ( ) (0.0888) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (0.0005) (0.0000) 36

46 Panel B. Three-Factor Alphas Starting Year P1 P10 P10-P1 Starting Year P1 P10 P10-P ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (0.0012) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (0.0058) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (0.0094) ( ) ( ) ( ) ( ) ( ) ( ) (0.0083) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (0.3726) (0.3628) (0.3315) ( ) ( ) ( ) 37

47 Table 1.6 Momentum portfolios raw returns following Up and Down markets This table presents the average monthly returns earned by momentum portfolios constructed with all NYSE, AMEX and NASDAQ stocks after excluding stocks priced below $5 at the beginning of the holding period and stocks with market capitalizations less than the cut -off level of lowest NYSE decile. At the end of each month (t) stocks are sorted into 10 equally weighted portfolios based on their cumulative returns earned in the past six months (t -5 to t). Positive (negative) returns of the value weighted CRSP index over the past 36 months define UP (DOWN) markets as in Cooper, Gutierrez, and Hameed (2004). Panel A and B report monthly average returns to these ten portfolios formed on the basis of the past 6 months cumulative returns and held for another 6 months for the three periods, , , and following UP and DOWN markets, respectively. The bottom two rows of this table present the average returns and the corresponding p-values to the winner-loser portfolios that buy winners (highest past return decile) and sells losers (lowest past return decile). All the portfolios are equal weighted. Panel A. Up Markets P1 (Past Losers) P P P P P P P P P10 (Past Winners) P10-P1 (Winners-Losers) p-value ( ) ( ) ( ) Panel B. Down Markets P1 (Past Losers) P P P P P P P P P10 (Past Winners) P10-P1 (Winners-Losers) p-value ( ) ( ) 38

48 Table 1.7 Momentum portfolios raw returns for 6-month/6-month strategy size and liquidity This table presents the average monthly returns earned by momentum portfolios for Small Cap, Large Cap, Low Liquidity and High Liquidity stocks. Sample includes all NYSE, AMEX and NASDAQ stocks after excluding stocks priced below $5 at the beginning of the holding period and stocks with market capitalizations less than the cut -off level of lowest NYSE decile. At the end of each month (t) stocks are sorted into 10 equally weighted portfolios based on their cumulative returns earned in the past six months (t -5 to t). This table reports the mean of monthly average returns to these ten portfolios formed on the basis of the past 6 months cumulative returns and held for another 6 months for the and periods. The bottom two rows of this table present the average returns and the corresponding p-values to the winner-loser portfolios that buy winners (highest past return decile) and sells losers (lowest past return decile). All the portfolios are equal weighted. Small Cap (Large Cap) comprises of stocks that have market cap smaller (larger) than median market cap NYSE stock. Illiquidity is estimated as ratio of absolute one day return to dollar volume in that particular day. Low (High) Liquidity stocks have higher (lower) average illiquidity than the median illiquidity stock in the month t Small Cap Large Cap Low High Low High Small Cap Large Cap Liquidity Liquidity Liquidity Liquidity P1 (Past Losers) P P P P P P P P P10 (Past Winners) P10-P1 (Winners-Losers) p-value ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) 39

49 Table 1.8 Fama-MacBeth regressions of stock returns on past 11 months cumulative returns, β, size, and BE/ME This table presents the average slopes from month-by-month regressions of stock returns on cumulative past returns, beta, size, and book-to-market for each subperiod. We consider all NYSE, AMEX and NASDAQ stocks that have data available both on CRSP and COMPUSTAT. Following Fama and French (1996), the cumulative past returns for each stock, each month are computed by cumulating their returns from t - 12 to t - 2 months. Stocks are assigned post -ranking β of the size-β portfolio they are in at the end of June of year t. BE is the book value of common equity plus balance-sheet deferred taxes. BE is obtained for each firm's latest fiscal year ending in calendar year t - 1. The accounting ratio is computed using market equity ME in December of year t - 1. Firm size ln(me) is measured in June of year t. In the regressions, these values of the explanatory variables for individual stocks are matched with CRSP returns for the months from July of year t to June of year t + 1. The gap between the accounting data and the returns ensures that the accounting data are available prior to the returns. LNBM is natural logarithm of BE/ME. P-values are in parentheses. *, **, *** indicate significance at the 10%, 5%, and 1% levels, respectively CUM_RETURN *** *** *** ** ** ** ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) POST BETA ( ) ( ) ( ) ( ) ( ) ( ) LNME *** *** ** * *** * ( ) ( ) ( ) ( ) ( ) ( ) LNBM *** ** * ( ) ( ) ( ) Number of observations

50 Table 1.9 Measures of delay for the three sub-periods This table presents the delay for the 5 size quintiles (individual stocks) for our sample of stocks. Delay is computed by subtracting the adjusted R 2 of unrestricted market model from the adjusted R 2 of the restricted market model (Delay = R ). The unrestricted model uses four lags of weekly market returns:,where is the weekly portfolio (individual stock) return at time t and is the market return. In the restricted model, the coefficients on the lagged market returns are constrained to zero: In Panel A, weekly returns of five size portfolio are the dependent variables in the market model. Weekly returns are the equal weighted portfolio returns for the size quintiles. All stocks in our sample are sorted into quintiles at the end of previous year. In Panel B, weekly returns of individual stock are the dependent variables in the market model. For each size quintile, we then compute the average delay. We also report the difference between the average delays of each subperiods and the corresponding p-values. Panel A. Portfolio Small Large Panel B. Individual Stocks Small Large (SP1) (SP2) (SP3) Diff. SP1 and SP p-value ( ) ( ) ( ) ( ) ( ) Diff. SP1 and SP p-value ( ) (0.4901) ( ) ( ) ( ) Diff. SP2 and SP p-value ( ) ( ) ( ) ( ) ( ) 41

51 Paper II Venture Capital Liquidity Pressure and Exit Choice (Co-authored with Ozgur Ince) 2.1 Introduction A lot of partnerships are 10 years, so many are looking for exits now. The IPO market has calcified, so M&A is the only exit for many. It takes a long time to do these deals, so they better get on it. Ideally, people would like to take their time, but the reality is they can't afford to now. Venture capital firms turn to M&A more for exits, MarketWatch, Dick Kramlich, senior partner and co-founder of New Enterprise Associates In this study we examine the impact of venture capital funds limited lifespan on the timing and the outcome of their portfolio companies liquidity events. The majority of venture capital funds (VCs) are structured as limited partnerships with a limited life span of 10 years. 1 At the end of the funds lifespan, the general partners are contractually obliged to dissolve the fund by liquidating the remaining equity holdings in the portfolio and return the proceeds to their limited partners. The limited lifespan of venture capital funds is a standard contractual feature designed to protect the limited partners from general partners conflicts of interest (Sahlman, 1990). While this mandatory liquidation requirement may protect limited partners from expropriation, we posit that it can also affect various other aspects of the venture capital process 1 VC firms use their investors capital to acquire large minority stakes in young and high-risk private start-ups that offer the potential of high returns. In independent limited partnerships with a limited life span, the venture capitalists serve as general partners and the investors as limited partners. An independent limited partnership VC funds lifespan can be extended to 12 or 13 years in one- or two-year increments with the consent of the funds board of advisors or at the discretion of the general partners (Sahlman, 1990). 72% of venture capital funds raised in number and 76% in dollars between 1985 and 2012 were structured as independent limited partnerships. The rest were mostly subsidiaries of industrial and financial corporations and university endowments. 42

52 in material ways. In particular, we investigate whether the obligation to dissolve the fund imposes a constraint on venture capitalists and influences their investment and exit decisions. The limited lifespan of independent limited partnership venture capital funds might act as a binding constraint for two main reasons. First, venture capital investments are inherently illiquid and venture capitalists rely on major liquidity events (e.g., initial public offerings (IPOs) and trade sales) to generate the high returns expected by their limited partners. The start-up firms financed by the VCs typically require multiple financing rounds over many years to reach the maturity required for successful exits (usually 5-10 years; average of 7.9 years in our sample). Second, IPO and M&A markets are inherently cyclical, frequently going through cold periods with low deal volume and low valuations. 2 This cyclicality can pose a significant challenge for VCs since turning down a profitable exit opportunity today in favor of a potentially better but uncertain exit in the future could be costly if the window closes. Consequently, the limited lifespan of VC funds is likely to influence general partners decisions long before the funds actually mature and are dissolved. We empirically test this VC liquidity pressure hypothesis by conditioning on VC funds age and comparing the exit outcomes of VC funds that are under liquidity pressure with those that are not. 3 Our sample includes 6,966 successful exits via initial public offerings and trade sales of companies backed by independent limited partnership VCs between 1985 and The mean age of independent venture capital funds at the time of the exits is 6.96 years, with 42% of exits occurring on or after VC funds eighth year and 30% occurring on or after their ninth year. 2 See, for instance, Lowry and Schwert (2002) for IPO market cycles, Harford (2005) for M&A waves, and Dittmar and Dittmar (2008) for a comprehensive examination of corporate financing waves. 3 Note that it is the age of the VC fund rather than that of the VC firm that is relevant for the liquidity hypothesis. VC firms do not have a limited life span and can manage multiple overlapping funds. In our analysis we use the age of the VC firms as a proxy for VC reputation and skill following earlier studies (see, e.g., Gompers (1996)). 43

53 Focusing on independent VCs with limited lifespans, we find a significantly negative relation between the age of the VC funds at the time of the investment and the time until exit. In univariate results, we find that entrepreneurial firms backed by VC funds that are five years old at the time of the first VC financing round have, on average, seven months less until exit compared to firms that receive a first financing round from a VC fund that is in its first year after inception. Controlling for portfolio firm and VC characteristics in a multivariate framework, we find that older VC funds are associated with significantly quicker exits especially when they are the lead VC and have greater influence on the portfolio firm, suggesting that the relation between fund maturity and exit timing is primarily due to the funds influence on their portfolio firms (the influence channel) rather than the funds strategic choice of portfolio firms (the sorting channel). We also find that older funds are more likely to exit their portfolio companies during cold markets, providing evidence that increasing liquidity pressure lowers VC funds ability to time the market. Next, we investigate the effect of VC funds liquidity pressure on the method of exit from their portfolio companies. We hypothesize that the longer time commitment, illiquidity, and uncertainty associated with IPO exits might lead older VC funds to prefer a sure gain from an immediate trade sale to a potentially more lucrative but uncertain future IPO. We find that entrepreneurial firms backed by VC funds that are older at the time of the exit are indeed significantly more likely to be acquired than go public. Focusing on successful exits (i.e., IPOs vs trade sales), a one standard deviation increase in the age of the VC fund at the time of the exit from the mean of 7 to 9.65 is associated with a 5.0 percentage point decline in the likelihood of an IPO from an unconditional probability of 30%. Furthermore, we find that the negative relation 44

54 between fund age and the likelihood of an IPO is stronger for lead VCs, suggesting that liquidity pressure works through the influence channel. A potential concern with the exit method analysis is that if VC fund age at exit is correlated with portfolio firm quality, this might cause a spurious relation between fund age and exit choice if firm quality is not adequately controlled for. More specifically, if higher quality firms are exited earlier, the likelihood of an IPO might drop with VC fund age due to a decline in the quality of the remaining portfolio of firms rather than an increase in the fund s liquidity pressure. We address this potential endogeneity concern in three ways. First, we repeat our exit method analysis after excluding trade sales with low (or undisclosed) transaction values based on the idea that these portfolio firms are likely to be of lower quality and unlikely to have a realistic choice between an IPO and a trade sale. Second, we exploit time-varying capital market conditions as a source of exogeneous variation in VC funds liquidity considerations and conduct a two-stage analysis with past IPO market conditions as the instrument. And third, we use propensity-score matching to estimate the impact of VC fund age on exit method in a subsample of firms with similar characteristics along several dimensions. With all three methods, we find that the relation between VC fund age at exit and the likelihood of an IPO is negative and both statistically and economically significant. Next, we turn our attention to VC-backed portfolio firms that go public and examine their lock-up expiration. Several studies report an abnormally high trading volume and a permanent stock price decline around lockup expirations, especially for firms with venture capital backing. 4 For instance, Field and Hanka (2001) document that venture capital backed firms experience 4 Lockup agreements are voluntary but standard agreements between the issuing firms shareholders and the IPO underwriters that restrict the insiders and pre-ipo shareholders from selling any of their shares for a pre-specified period of time after the IPO. Lockups usually last for 180 days and cover most of the shares that are not sold at IPO. For empirical analyses of IPO lockups, see for instance Bradley et al. (2001), Field and Hanka (2001), and Brav and Gompers (2003). 45

55 abnormal returns that are almost three times larger and abnormal trading volume that is five times larger compared to firms without VC backing during the three days surrounding the lockup expiration, and interpret this as evidence of aggressive selling by venture capital funds. Consistent with the VC liquidity pressure hypothesis, we find that portfolio firms backed by VC funds that are closer to liquidation experience significantly lower stock returns and larger abnormal trading volume around their lockup expirations. Moreover, both the trading volume and stock return effects are more pronounced when there are multiple independent VCs under liquidity pressure, whereas the number of independent VCs that are not under liquidity pressure does not matter. To briefly preview our results, we document that portfolio firms backed by VC funds that are at the tail-end of their limited lifespan experience earlier exits, are more likely to be sold off than taken public, are more likely to be exited during colder markets, and are more likely to experience insider selling at the time of lock-up expiration following IPOs. Our results indicate that the limited lifespan of independent VC funds has real consequences for the timing and the outcome of their portfolio firms exit events. Two recent studies also focus on the limited lifespan of venture capital funds. Theoretical work by Kandel, Leshchinskii, and Yuklea (2011) shows that funds limited life horizon and general partners informational advantage over the limited partners lead to inefficient decisions during the investment cycle. However, they do not investigate the consequences of the funds limited lifespan on the exit cycle. Masulis and Nahata (2011) investigate the effects of VC backing on the profitability of private firm acquisitions. They report that portfolio firms backed by VC funds nearing maturity earn a lower acquisition premium over the book value of their assets; however, they do not investigate the impact of fund maturity on the timing of exits and 46

56 the IPO process. Our study is also related to Gompers (1996) who shows that young venture capital firms take their portfolio companies public earlier than the more established firms in order to establish a track record quickly and raise capital for a new fund. Our results show that the limited lifespan of VC funds cause a similar exit timing behavior regardless of the VC firm s reputation and future fundraising concerns. This article also relates to the broader literature that examines the impact of VCs incentives and the structure of VC contracts on exit outcomes. Cumming (2008) finds that the use of convertible securities in VC investments is associated with a higher frequency of acquisitions and fewer IPOs. There is also evidence that companies that share a common VC are more likely to engage in strategic alliances (Lindsey, 2008) and successful acquisitions (Gompers and Xuan, 2008). Ince (2012) finds that IPO firms are more likely to grant the underwriting mandate to investment banks with a strong relationship with the firms leading VCs, and such repeated pairings between the investment banks and VC firms are associated with better IPO outcomes. 5 Finally, several other articles also examine trading volume and stock returns around IPO lockup expirations. Field and Hanka (2001) and Brav and Gompers (2003) attribute the permanent price decline around the lockup expiration to a combination of downward sloping demand curves, limited arbitrage in the form of restricted short-selling, and systematically biased prior beliefs about the extent of insider selling. We contribute to this literature by documenting that VC funds liquidity pressure is an important factor in lockup expirations. While our results shed new light on this enduring market anomaly, they also add a new piece to the puzzle given that VC funds time to maturity is observable and the predictable selling of shares by VC funds should not raise any adverse selection concerns. 5 For a recent comprehensive survey of venture capital research, see Rin, Hellmann, and Puri (2011) and Metrick and Yasuda (2011). 47

57 2.2 VC liquidity pressure hypothesis VC funds are typically organized as limited partnership and VC firms act as general partners (GPs) for them. The limited partners (LPs) of VC funds are mostly institutional investors who commit to provide a certain amount of capital during initial fundraising. Independent VC funds typically have a finite life of 10 years, with an option to extend their life up to 12 or 13 years. During the ten years of the fund s typical lifetime, GPs select, monitor, mentor and provide a variety of other services for their portfolio companies. At the end of this period the fund needs to dissolve and distribute its profits to its LPs. In this paper we investigate whether the limited lifespan of VCs acts as a constraint in general partners investment and exit decisions. In particular, we examine whether VC-backing by funds nearing maturity influence their portfolio firms (i) exit timing, (ii) exit method (IPO or trade sale), and (iii) the lock-up expiration following IPOs. We label this as the VC liquidity pressure hypothesis. Venture capital funds usually hold large equity stakes and obtain significant control rights in their portfolio firms. Most notably, venture capitalists hold multiple board seats, maintain veto rights that grant them control over potential exit events, and retain the right to put their investment back in the portfolio firm at original cost plus the cumulative dividends accrued. These latter redemption rights provide the general partners with leverage over the entrepreneur based on the credible threat of withdrawal in addition to allowing them to extract their original investment from portfolio firms that are unlikely to succeed. 6 There is growing empirical evidence that such control rights effectively grant VCs influence over their portfolio firms exits. Cumming (2008) finds that the use of convertible securities in 6 See Sahlman (1990), Lerner (1994), and Smith (2005) for the properties of contracts between general partners and limited partners of venture capital funds. 48

58 VC investments is associated with a higher likelihood of trade sales, consistent with the theories of Bascha and Walz (2001) and Hellmann (2006). There is also evidence that companies that share a common VC are more likely to engage in strategic alliances (Lindsey, 2008) and successful acquisitions (Gompers and Xuan, 2008). Ince (2012) finds that VCs prior relationships with investment banks influence their portfolio firms choice of IPO underwrites. Gompers (1996) documents that young VC firms are associated with quicker IPOs at lower valuations, and interprets this as evidence that venture capitalists lacking a strong track record force their portfolio firms to early exits in order to facilitate future fundraising. Our empirical tests focus on how VC funds time to maturity affects their portfolio firms (i) exit timing, (ii) exit method, and (iii) abnormal trading volume and stock prices around the expiration of the lock-up periods. We use the age of the VC fund at the time of exit as our primary proxy for the liquidity pressure faced by venture capitalists, based on the notion that VC funds which are closer to maturity are more likely to be under pressure to exit their portfolio firms. In addition, in several tests we pay special attention to the liquidity pressure faced by the lead VC firm, which typically has the most control rights and influence over the portfolio company. Since VCs ownership stakes and control rights are not reported by most commonly used commercial databases, we follow earlier studies (e.g., Masulis and Nahata, 2011 and Lee and Wahal, 2004) and designate a VC as the lead on the basis of VC firms pre-exit financing rounds as reported by VentureXpert. More specifically, we classify a VC fund as the lead VC for a portfolio firm if it participates in the firm s first VC financing round and its VC firm makes the largest total investment in the firm across all pre-exit investment rounds. The lead VC designation allows us to investigate whether the liquidity pressure of the VCs with larger 49

59 influence over the portfolio firm has a larger impact on the portfolio firm s exit timing and method. 2.3 Data and summary statistics Sample selection Our primary sample includes all VC investments made by U.S. based independent VC firms in private entrepreneurial companies headquartered in the U.S. with a successful exit (via an IPO or a trade sale) between 1985 and The venture capital investment sample is drawn from Thomson Financial s VentureXpert and includes data on investment dates, investment amounts, identities and characteristics of venture capital firms and their funds, and the exit outcomes of VC-backed portfolio firms. We supplement VentureXpert with IPO data from Thomson Financial s Global New Issues and acquisition data from Merger and Acquisitions databases. Our focus on VC funds liquidity pressure requires complete data on VC funds identities and the dates for inception, investment, and exit from portfolio companies. We exclude investments by angel investors and subsidiary VC firms (i.e., venture capital operations of corporations, insurance companies, and financial institutions), which do not typically have limited lifespans. 7 We obtain monthly return data from the Center for Research in Security Prices (CRSP) database to calculate industry returns. We collect data on patents granted to the entrepreneurial companies in our sample from the US Patent and Trademark Office using fuzzy name and headquarter location matching. 7 Funds with Investment type PRIV and VC firms that have firm type Private Equity Firm are classified as independent VCs. Corporations, insurance companies, and financial institutions are classified as subsidiary VCs. Investments by funds with incomplete identification and missing inception dates are excluded. We also exclude investments that occur more than ten years after VC funds inception since the majority of such investments are erroneously attributed to an earlier fund in the VC organization due to unknown identity information. 50

60 Lock-up expiration data for VC-backed IPOs is from the Global New Issues Database in SDC Platinum (Securities Data Corporation) provided by Thomson Reuters between 1985 and We apply the conventional filters and exclude firms that issue a security other than common equity, financial firms (SIC codes ), spinoffs and carve-outs, reverse LBOs, ADRs, foreign listings, and those with an offer price less that $5. For firms with multiple share classes, we calculate total shares outstanding by summing up shares outstanding across all classes. IPO firms with multiple share classes are obtained from Jay R. Ritter s website. 8 We obtain daily returns, daily trading volume and shares outstanding from CRSP database for portfolio firms with successful initial public offerings. IPOs with missing lockup dates are excluded from tests that require an exact date for the lockup expiration Variable definitions and summary statistics Table 2.1 presents descriptive statistics for the variables used in our empirical tests. Panel A reports variables that are measured at the individual VC investment level. VC fund age at investment is calculated as the number of years between a fund s inception and its financing round in a portfolio firm. Time until exit is calculated as the number of months between an investment and the VC s exit from the portfolio company via an IPO or trade sale. First VC round dummy equals one if the investment round marks the first time an entrepreneurial company received capital from a venture capital fund, and zero otherwise. 19% of all venture capital investments are made at the first VC financing round. Syndicate size is the number of For firms with multiple lockup expiration days, we pick the earliest date reported by SDC if the percentage of the shares released on that date is larger than 15%, otherwise we choose the date with the greatest percentage of shares released. We exclude firms with multiple lockup expiration dates if data on the percentage of shares released is not reported. 51

61 distinct VC funds participating in the financing round. On average, each financing round has participation by 3.14 VC funds. Panel B reports variables that are measured at the portfolio firm VC fund level, and thus do not vary across multiple investments of the same VC fund in the same entrepreneurial firm. Fund age at exit is the number of years between a fund s inception and the portfolio firm s exit event via an IPO or trade sale. We define the lead VC firm as the one that makes the largest total investment across all rounds of funding after participating in the first VC financing round (see also Nahata (2008) among others). Following Nahata (2008), we measure VC capitalization share as the cumulative market value of all companies taken public by the VC firm over the five years prior to the VC s first investment in the portfolio firm, normalized by the aggregate market value of all VC-backed companies that went public during the same time period. Following Hochberg et al. (2007), we measure VC connectedness as the number of unique VCs each VC has syndicated with during the five years prior to the VC s first investment in the portfolio firm, normalized by the number of all possible combinations during the same time period. Following Chen et al. (2010), VC center dummy equals one if the VC firm is located in the Combined Statistical Areas of San Francisco, New York, or Boston, and zero otherwise. Chen et al. (2010) find that both VCs and their portfolio companies concentrate in these three geographic regions and VC firm located in these VC centers exhibit better performance. Panel C reports variables that are measured at the portfolio firm level. IPO dummy equals one if the portfolio firm has an IPO, and zero if it is exited via a trade sale. 29% of the successful exits in our sample are via an IPO. # of VC rounds is the number of distinct VC financing rounds received by the portfolio firm prior to an IPO or trade sale. Each portfolio firm receives an average of 3.42 VC financing rounds prior to a successful exit. We collect the number of patents 52

62 granted to the entrepreneurial by the United States Patent and Trademark Office with an application date that falls between the first VC financing round and the exit date. We measure the number of IPOs in prior quarter as the number of completed VC-backed IPOs in the same industry during the three months prior to the month of the exit. Lagged # of IPOs (qtrs. -2:-9) is the number of completed VC-backed IPOs during the two year period ending three months prior to the month of the exit. Market returns in prior quarter is the equally-weighted stock returns of public firms belonging to the high-tech industries (three-digit SIC codes of 283, 481, , , 357, and 737) during the three months prior to the month of the exit. 2.4 Timing of VC exits According to the VC liquidity pressure hypothesis, independent VCs face a pressure to exit their investments as their funds approach maturity. In this section we examine the empirical relation between VC funds time to maturity and the timing of their portfolio firms exit events, and investigate if funds limited lifespan acts as a binding constraint. First, we split the time between a fund s closing date and its exit from a portfolio company into two periods: (i) the time between the closing of the fund and the date of the fund s investment in the portfolio company, and (ii) the time between the investment and the fund s exit from the portfolio company. If the timing of exit is unrelated to the VC funds liquidity considerations, solely dictated by the start-ups characteristics (e.g., growth rate, profitability etc.) and market conditions instead, we should not observe a significant relation between the funds age at the time of investment and the time until exit. On the contrary, VC liquidity pressure hypothesis predicts a negative relation between the two: a start-up backed by a VC fund closer to maturity will experience a quicker exit. 53

63 In Table 2.2 we sort VC funds by their age at the time of an investment in a portfolio company and report summary statistics for the number of months between the investment and the portfolio company s exit event. The first two columns report the mean and median time until exit for all VC investments. We find a monotonic negative relation between the age of the fund at the time of the investment and the number of months until exit. The mean (median) number of months between investment and exit is 54.9 (47) months for portfolio companies that receive financing from a VC that is one year old at the time of the investment. In comparison, the mean (median) number of months until exit is 37.4 (29) months for VC funds that are 10 years old. The mean and median difference of and 18 months, respectively, are statistically highly significant. The significantly negative relation between the age of the VC fund and the time until exit suggests that VC funds limited lifespan is a binding constraint on the timing of their portfolio firms exit. There are two mutually non-exclusive possible explanations for this relation. First, VC funds might choose their portfolio firms strategically and avoid investing in start-ups that are expected to take too long to mature when the fund is nearing maturity the sorting channel. Second, VC funds might exercise their control rights and influence their portfolio firms towards earlier exit events when they are under liquidity pressure the influence channel. The distinction is important: if VC funds liquidity pressure works through the `influence channel, it might impose externalities on the portfolio companies, whereas through the `sorting channel, it would not. One way to distinguish between the sorting and influence channels is to focus on VC-backed portfolio firms first VC financing round. We posit that while sorting might play an important role in later stages when the portfolio firm is close to an exit, sorting is unlikely to be a factor in 54

64 early stages. To that end, Table 2 also reports summary statistics for the number of months until exit for the subsample of first round VC investments. We limit this analysis to VC funds that are five years old or younger at the time of the first round of investment given the standard covenant in VC partnership agreements that restricts initial investments in new portfolio companies to the first five years of funds lives. According to Table 2.2, first round investments also exhibit a significantly negative relation between VC fund age and time until exit. Portfolio firms that receive their initial VC financing from a fund that is five years old have on average seven fewer months until exit compared to those financed by a fund at its first year. Moreover, the difference is observed primarily on the right tail of the distribution, which is consistent with the idea that the primary effect of liquidity pressure is on portfolio companies that take relatively longer to exit. In Table 2.3 we explore the determinants of exit timing in a multivariate regression framework and further distinguish between the sorting and the influence channels. The dependent variable is the natural logarithm of the number of months between the first round of VC financing received by a portfolio firm and its exit date. We relate the time until exit to the age of the VC fund at the time of the investment (in years), the exit method (IPO vs. trade sale), the size of the VC syndicate in the financing round, the natural logarithm of the adjusted number of patents granted to the portfolio company from applications prior to its exit 10, VC capitalization share, VC connectedness, VC center dummy, the natural logarithm of the number of IPOs in the same industry during the prior three months, the average stock return of public companies in the high-tech industries during the prior three months, and industry fixed effects. In the first column the sample includes all investments made by independent VC funds in portfolio companies that 10 We calculate the natural logarithm of the adjusted number of patents as the residual from an OLS regression of the natural logarithm of one plus the number of patents on the number of years between the firm s first VC financing round and exit, the number of years squared, and year- and industry-fixed effects. 55

65 are subsequently exited via an IPO or trade sale. 11 In columns 2 and 3 we limit the sample to portfolio firms initial VC financing round only. In specification 1, we find a very significantly negative relation between the age of the fund at the time of an investment and the time until exit, confirming the univariate results from Table 2.2. In column 2 with the subsample of first round VC investments, the relation remains significantly negative. Column 3 adds an interaction between the age of the fund at the time of the first round investment and an indicator that equals one if the VC fund is the lead VC for that investment and zero otherwise. The interaction variable is intended to capture the marginal impact of the liquidity considerations of VC funds with larger influence over the management of their portfolio companies. According to the `sorting channel, the coefficient on the interaction term should be insignificant since all VCs are expected to have similar strategic motives in choosing their portfolio firms regardless of the amount of influence they have over the portfolio firm. On the other hand, the `influence channel predicts a significantly negative coefficient on the interaction term since lead VCs are expected to have a greater influence on their portfolio firms exit decisions. We indeed find that the interaction variable has a significantly negative coefficient whereas the coefficient on the stand alone fund age variable becomes only marginally statistically significant with a t-statistic of Overall, the multivariate results in Table 2.3 confirm the univariate results, and provide support for the argument that VC funds liquidity pressure affects the timing of their portfolio firms exit events via the influence channel. Several other factors affect the timing of exits. We find that IPOs are associated with quicker exits after investment. After controlling for the method of exit, proxies for the quality of the 11 Therefore, each exit event is represented multiple times since each portfolio firm typically receives multiple rounds of financing from multiple VC funds. 56

66 portfolio firm and the VCs appear to be positively related to the time until exit. We find that start-ups with more patents and those backed by a larger number of VCs at the financing round and by VCs with larger market shares have longer time until exit. Finally, we find that exits that occur during more favorable IPO conditions tend to be quicker exits, consistent with the idea that VCs are eager to take advantage of better market conditions before the window of opportunity closes (see also Giot and Schwerenbacher, 2007). The results in Tables 2.2 and 2.3 are consistent with the idea that independent VC funds liquidity considerations impose a binding constraint on their exit policy. In Table 2.4 we investigate if this binding constraint causes a loss of flexibility in the VCs ability to time the exit market. The dependent variable is a dummy that equals one if the exit occurs during cold IPO market conditions. We classify an exit as one occurring during a cold market if the number of IPOs during the prior three months is below the median for all successful exits. We relate the market conditions at the time of the exit to the age of the fund at the time of the investment and exit, along with the control variables from Table 2.3 with the exception of proxies related to market conditions. In Table 2.4 we report the marginal effects of the independent variables. In addition, we standardize the continuous independent variables such that they have a mean of zero and a standard deviation of one. As a result, the reported marginal effects capture the effect of a one standard deviation change in the regressor on the probability of the exit occurring during cold exit market conditions. Column 1 of Table 2.4 reveals a significantly positive relation between the age of the VC fund at the time of the investment and the probability of exit during a cold market. The marginal effect is 0.024, indicating that a one standard deviation increase in the age of the VC fund is associated with a 2.4 percentage point increase in the likelihood of exit during a cold market. 57

67 Column 2 adds the age of the fund at the time of the exit. We observe that the age of the fund at both the time of investment as well as the exit have significantly positive coefficients when included together. The marginal effect of age at exit is 0.023, indicating a further increase in the likelihood of exit during a cold market of 2.3 percentage points as a result of a one standard deviation increase in the age of the fund at the time of the exit. Altogether, the results in Table 2.4 indicate that the liquidity pressure documented in Table 3 is also associated with a decline in the flexibility to time the exit market. To the extent that conducting an exit during colder markets is less desirable, these results suggest that VC liquidity pressure is associated with a deviation from the optimal exit policy. 2.5 Exit Choice In this section, we investigate the impact of VCs liquidity pressure on the method of exit. More specifically, we relate the age of the VC fund at the time of the exit to the decision to exit via an IPO or trade sale. The liquidity pressure hypothesis predicts a negative relation between fund age at exit and the probability of IPO for two reasons. First, the results in Table 2.4 show that later exits are more likely to occur during colder IPO markets. Given the well-documented positive relation between IPO market conditions and the likelihood of an IPO over a trade sale (see, e.g., Nahata 2008), later exits should also be less likely to be via an IPO due to the reduced flexibility of aging funds to time the market. Second, the liquidity pressure should be more severe with IPOs due to the increased time commitment and illiquidity associated with a prolonged exit process, and the associated increase in the sensitivity of deal success to uncertain future market conditions. In other words, a trade sale might be preferable to an IPO on an uncertainty- and illiquidity-adjusted basis for a VC fund that is under liquidity pressure, even if an IPO might generate larger exit proceeds conditional on success. 58

68 Figure 2.1 presents a histogram of the age of VC funds at the time of successful exits separately for IPOs and trade sales. Exits appear to reach a peak when VC funds are around 6 or 7 years old. Notably, exits are relatively more likely to be via IPOs in VC funds early years and less so as funds age. The decreasing likelihood of IPOs in funds later years is surprising in light of the fact that it typically takes firms a considerably longer time to prepare for and execute an IPO compared to a trade sale, and that prospective firms are generally expected to have reached a certain level of maturity before becoming a publicly listed company. 12 On the other hand, the trend in Figure 2.1 is consistent with the negative influence of liquidity pressure on the likelihood of IPOs. In the remainder of section 5, we undertake a thorough examination of the relation between VC fund age and the exit method after controlling for other factors that might affect the choice between an IPO and trade sale Baseline results in exit choice Table 2.5 reports probit regressions of the exit method on fund age at exit and investment, along with several proxies for portfolio firm quality, VC reputation, market conditions, and industry and year fixed effects. Marginal effects with standardized coefficients reflecting a one standard deviation change from the mean are reported along with robust standard errors clustered at the portfolio firm level in parentheses. Since our primary variable of interests--vc fund age at exit and the exit method--do not vary across multiple investments by the same VC fund in a portfolio firm, we conduct the regressions at the portfolio firm VC fund level by aggregating VC 12 After deciding to go public, prospective IPO firms prepare for the offering by appointing independent board members, creating an audit committee, evaluating corporate governance practices, hiring investment bankers, a law firm, accounting advisors, and an independent auditor, registering the offer with the SEC, preparing the IPO prospectus, and marketing the company to investors in road shows (PWC, 2011). Boehmer and Ljungqvist (2004) analyze the duration between the date firms announce their intention to go public and the IPO date for a sample of German IPOs and find an average waiting time of more than two years. It is difficult to conduct a similar duration analysis for U.S. IPOs since intentions to go public are not systematically announced and recorded. 59

69 funds multiple investments in each portfolio firm. As a result, our main sample includes 20,860 total observations belonging to 6,966 successful exits (2,010 IPOs and 4,956 trade sales), with each portfolio company backed by three unique VC funds on average. Column 1 reports that the likelihood of an IPO is positively related to the size of the syndicate at the first VC financing round, the number of patents assigned to the portfolio firm, the reputation of the VC firm as measured by its IPO capitalization share during the five years prior to its initial investment in the portfolio firm (VC capitalization share), whether the VC firm is headquartered in one of the three VC centers (VC center dummy), and recent market conditions. The two control variables with the largest economic significance are number of patents and recent IPO market conditions, with a one standard deviation change from the mean causing a 8.3 and 11.5 percentage points increase in the likelihood of an IPO from a baseline probability of 28.9%. In column 2 we add the age of the VC fund at the time of the exit to the probit regression. We find that a one standard deviation change in VC fund age at exit (from the mean of 6.96 to 9.65) is associated with a 5.0 percentage points decline in the probability of an IPO, with a t-statistic of In column 3 we include an interaction between the fund age at exit and a dummy for lead VC to explore whether the age of the VC funds with more influence over their portfolio firms has a stronger relation to the exit choice. We find that the coefficient on the interaction term is significantly negative, consistent with a larger impact of the liquidity pressure of the more influential VCs. In columns 4 and 5, we run the baseline specification from column 2 in two subsamples. In column 4, we exclude early exits that occur when the VC fund is 4 years or younger, which are disproportionately less likely to be trade sales and thus may not represent a realistic choice 60

70 between an IPO and trade sale. 13 The marginal effect of the VC fund age at exit is -3.7 percentage points with a t-statistic of -4.3 after excluding early exits, indicating that the earlier results from the full sample are not driven by a higher likelihood of IPOs in early years. In column 5, we exclude trade sales with low (or missing) transaction values under the assumption that these portfolio firms were less likely to have had a realistic IPO option. 14 The marginal effect of the VC fund age at exit is -3.6 percentage points with a t-statistic of -3.8 after excluding trade sales with low or missing transaction values. The results from the subsample analyses indicate that a higher likelihood of IPOs in early years or an excess of low quality trade sales in later years does not drive the full sample results Identification In this subsection we address the possibility that the negative relation between the age of the VC fund at exit and the likelihood of an IPO documented in Table 5 might be spurious. The primary concern is that portfolio firm quality might not be fully captured by the control variables included in our tests. If omitted portfolio firm quality is correlated with the age of the VC fund at exit, this could cause a spurious relation between fund age and exit choice. In particular, if portfolio firms exited late are of lower quality, their propensity to be sold off instead of taken public might be due to low portfolio firm quality rather than VC funds liquidity pressure. We control for such potential endogeneity in VC fund age using three approaches. First, we conduct two-stage least-squares regressions using lagged market conditions as the instrumental variable. Second, we conduct a propensity score matching analysis to identify portfolio firms in 13 The fraction of observations that are IPOs is 62% in VC funds first year, 49% in their second year, and 39% in their third and fourth years. 14 More specifically, we limit the sample to trade sales with a non-missing transaction value at least as large as the market capitalization of a VC-backed IPO in the same industry during the same time period. We split the full sample period to , , , and This filter leaves 1,080 trade sales with 3,340 total observations. 61

71 the treatment group (late exits) that are as similar as possible to the firms in the control group (earlier exits) in terms of observable measures of quality. Finally, we investigate the relation between portfolio firm quality and the age of the VC fund at exit directly to explore if later exits are more likely to be via trade sales due to declining portfolio firm quality Instrumental variable approach The primary motive behind the instrumental variable approach is to decompose VC fund age into an exogeneous component uncorrelated with portfolio firm quality and an endogeneous component potentially correlated with portfolio firm quality. To that end, we need an exogeneous variable that is correlated with fund age at exit for reasons unrelated to the quality of the firm being exited. Our strategy is to exploit past exit market conditions as a source of exogeneous variation in VC liquidity considerations. For example, an abnormally cold IPO market in the past is likely to cause a delay in exits for market-wide reasons unrelated to the quality of a particular portfolio firm. In contrast, exit choice for late exits that follow favorable market conditions is more likely to be dictated primarily by firm quality. Therefore, we use the natural logarithm of the lagged number of IPOs in the industry during the two years ending three months prior to the exit as our instrument. 15 Table 2.6 presents the two-stage least squares results. Column 1 reports the first-stage OLS regression with the age of the VC fund at exit as the dependent variable. The coefficient on the instrumental variable is negative and statistically very significant with a t-statistic of -28.9, indicating that the instrumental variable satisfies the inclusion restriction. As expected, unfavorable IPO market conditions in the past are associated with later exits from portfolio 15 We exclude the number of IPOs in the most recent three-month period from the instrumental variable and instead separately control for recent market conditions in the second-stage to ensure that the instrument does not pick up any variation in market conditions correlated with firm quality through a short-term demand channel. 62

72 companies. Columns 2-4 report the second-stage probit regressions of exit method. The dependent variable is a dummy that equals one for IPOs and zero for trade sales. Marginal effects with standardized coefficients reflecting a one standard deviation change from the mean are reported along with robust t-statistics clustered at the portfolio firm level. Column 2 adds the predicted VC fund age at exit from the first-stage regression. The marginal effect of predicted VC fund age at exit is with a t-statistic of -5.2, indicating an economically very significant 13.7 percentage points drop in the likelihood of an IPO associated with a one standard deviation increase in the age of the VC fund at exit. In column 3, we include the residual from the first-stage regression to explore the relation between exit method and the endogeneous component of VC fund age that is potentially correlated with portfolio firm quality. The coefficient on the residual component is also significantly negative with a t-statistic of However, the economic significance is considerably less than the predicted component with a one standard deviation increase from the mean causing a -2.8 percentage points drop in the likelihood of an IPO. Finally, in column 4, we interact both the predicted and residual components with an indicator for lead VCs. Notably, the interaction with the predicted component is significantly negative with a t-statistic of -5.8, indicating that lead VCs liquidity pressure associated with past market conditions has a larger impact on exit choice. In contrast, the interaction with the residual component is statistically insignificant Matched sample approach In this section, we further explore the relation between VC funds age at exit and the method of exit using a treatment effect method. The primary purpose of this approach is to ensure that the treatment effect (the impact of a late exit on the likelihood of an IPO) is estimated by 63

73 comparing treated subjects (late exits) with control subjects (earlier exits) that are as similar as possible across various observable characteristics considered important in explaining the outcome (IPO vs. trade sale). This is achieved by estimating the counterfactual unobserved outcomes of treated subjects using the observed outcomes from a subsample of similar subjects from the control group. We use the propensity score matching method to construct the subsample of control subjects. Roberts and Whited (2011) proposes propensity score matching as a useful robustness test for regression based analysis. In particular, matching avoids the functional form restrictions imposed by linear regressions. Table 2.7 presents exit choice results using propensity score matching. The treated group consists of the exits of VC funds that are nine years or older at the time of the exit. The control group consists of the exits of VC funds that are eight years or younger. For each observation in the treated group, we locate an observation from the control group with the closest propensity score. Propensity scores are estimated using a probit regression of a dummy indicating a late exit (VC fund age at exit >=9) on the age of the VC fund at 1 st investment, the size of the initial syndicate, the natural logarithm of the adjusted number of patents, VC capitalization share, VC connectedness, and VC center dummy. In Panel A, the treatment effect is reported for the unmatched and matched samples. The unmatched treatment effect of indicates that late exits are -5.2 percentage points less likely to be IPOs in the full sample. The matched treatment effect of indicates that late exits are percentage points less likely to be IPOs compared to matching early exits with the closest propensity scores. The more negative treatment effect estimate from the matched sample indicates that the late exit group consists of observations associated with a higher than average propensity to conduct an IPO if not for the liquidity pressure. This is consistent with the results from propensity score matching that 64

74 assignment to the treated group is positively related to the size of the VC syndicate, the number of patent assignments, and the VC s capitalization share (untabulated), which are all significantly positively related to the propensity of an IPO (see, e.g., column 1 in Table 2.5). Panel B reports the results of a full-specification probit regression of exit choice using the subsample of matched observations, and compares them to results from the full, unmatched sample. Consistent with the results in Panel A, we find that the treatment effect is more negative in the matched sample. The coefficient on Dummy (Age at exit>=9) has a marginal effect of vs in the unmatched sample. The results in Panel B confirm that the negative impact of late exits on the likelihood of an IPO is larger in matched samples after controlling for market conditions and including industry and year fixed effects in a regression framework VC age and portfolio firm quality Finally, we directly examine the relation between portfolio firm quality and the age of the VC fund at exit to investigate whether later exits are associated with a decline in portfolio firm quality. First, we repeat the probit regressions of exit choice in Table 5 using only proxies for the quality of the portfolio firm and its investors as independent regressors and excluding all other variables associated with the VCs liquidity and market timing considerations. We posit that if the decline in the likelihood of IPOs as VCs age is driven by early exits of higher quality portfolio firms and the associated decline in the quality of the remaining firms in the portfolio, then the predicted probability of an IPO based on observed measures of quality should also decline with VC fund age. Figure 2.2 plots the predicted probability of an IPO vs. a trade sale by VC fund age. We find that the likelihood of an IPO increases slightly over the first five years from 30% to 35% and 65

75 remains roughly flat thereafter. In other words, there does not appear to be a decline in portfolio firm quality with increasing VC fund age based on observable quality proxies. In Figure 2.3, we conduct a closer examination of important characteristics of acquired portfolio firms. According to the results in Table 2.5, the number of patents granted to VCbacked companies is a statistically and economically important predictor of exit choice, and thus is likely to be a useful proxy for firm quality. In Panel A of Figure 2.3, we investigate if the patent intensity of VC-backed firms is negatively related to the age of their VC at the time of the exit. Since the number of patents granted to a firm is likely to increase over time with firm age, we control for this time effect by focusing on the number of patents granted per year of VC backing. More specifically, we measure patent intensity as the number of patents granted to the portfolio firm with an application date prior to the exit date as recorded by the U.S. Patent and Trademark Office divided by the number of years between initial VC financing and the exit date. Furthermore, we scale patent intensity with the average patent intensity of IPO firms in the same VenturExpert ten-industry classification in order to account for industry effects. We find that the patent intensity of acquired portfolio firms actually increase with VC fund age and reach a maximum of 56% of the patent intensity of IPO firms in the same industry by the end of VC funds life cycle. In Panel B of Figure 2.3, we examine the observed valuations of a subsample of acquired portfolio firms at the time of the exit event for which the transaction values are reported by the SDC (available for approximately 45% of the trade sales, distributed sporadically over the sample period). For each trade sale, we adjust the transaction value for inflation using the Consumer Price Index and scale it by the average market capitalization of IPOs in the same 66

76 industry at the time of the offering during the same time period. 16 We find that the valuation of trade sales relative to IPOs declines at first with VC fund age reaching a minimum of 23% in year 7, and starts to increase thereafter to 33% by year 12. The increase in valuations at the tail end of funds is inconsistent with a decline in portfolio firm quality and is observed despite a likely decline in the bargaining power of the acquired firms vis-à-vis the acquirers. Altogether, the evidence from figures 2.2 and 2.3 is inconsistent with the notion that the quality of portfolio firms declines with VC fund age. This presents further evidence against the notion that the negative relation between fund age and IPO likelihood documented in tables 2.5, 2.6, and 2.7 is driven by an omitted variable bias caused by unobserved systematic variation in portfolio firm quality. 2.6 Which funds succumb to liquidity pressure? In this section, we investigate which VC funds are more likely to modify their exit strategy due to liquidity considerations. We consider two proxies for VCs incentives to engage in such liquidity management. First, we examine whether liquidity considerations are more important for younger VC firms with limited track record. Gompers (1996) documents that young venture capital firms take companies public earlier at less favorable terms, and attributes this to young VC firms desire to establish a reputation quickly and raise capital for new funds even at the expense of greater initial IPO underpricing. 17 While Gompers (1996) focuses only on IPOs, it is possible that the grandstanding incentive of young VC firms might influence the timing and method of exits more generally. 16 We group exits by the following four time periods: , , , Lee and Wahal (2004) document a positive relation between IPO underpricing and young VC s future fundraising success, consistent with the idea that VCs that lack a track record benefit from grandstanding. 67

77 Second, we investigate if fund performance affects liquidity management. On the one hand, aging funds that have not had successful exits might be more incentivized to accelerate their exits in an effort to return capital to their investors and earn performance-based compensation without further delay. On the other hand, the lower likelihood of an IPO associated with liquidity pressure documented in section 5 might be considered less costly for VC funds that have already had successful IPOs from the same portfolio. Table 2.8 presents the results. The first two columns examine how fund sequence affects the relation between fund age at investment and exit timing (column 1), and fund age at exit and exit method (column 2). Column 1 repeats the OLS regression from column 2 of Table 2.3 after including a dummy that equals one if the VC fund is the parent firm s first fund as an interaction with fund age at first-round investment and as a stand-alone regressor. The dependent variable is the natural logarithm of the number of months between the first VC round of investment and the exit event. We find that both the stand-alone regressor and the interaction term have statistically insignificant coefficients, indicating that first funds are not more prone to accelerating their exits. Column 2 repeats the probit regression from column 2 of Table 2.5 after including the first-fund dummy. The dependent variable equals one for IPOs and zero for trade sales. Once again, both the stand-alone regressor and the interaction term have statistically insignificant coefficients, indicating that aging funds tendency to favor trade sales over IPOs is not related to fund sequence. In columns 3 and 4, we repeat the exit timing and exit method analyses from the first two columns using the number of prior IPOs in the fund as the incentive proxy. We find that the negative relation between fund age at first round and time until exit is greater for funds with a larger number of IPOs prior to the exit. It appears that portfolio firms of aging VC funds are 68

78 exited more quickly if the VC fund has already established a track record of IPOs. In column 4, we find evidence of intra-portfolio performance persistence: the coefficient on the number of prior IPOs is positive and statistically highly significant. However, despite performance persistence, the likelihood of an IPO declines more with fund age for funds with a greater number of prior IPOs. Altogether, the evidence in Table 2.8 suggests that liquidity considerations have a greater influence on exit strategies of VC funds that have exhibited better performance. We interpret this as evidence that the costs of liquidity management (e.g., exiting companies earlier and via trade sales instead of IPOs) are lower for VC funds that have already established a track record of IPOs from the portfolio. In contrast, we do not find any evidence that first time funds are any more likely to engage in liquidity management compared to more experienced VC funds. We conclude that the influence of liquidity considerations on the VC exit cycle is distinct from the grandstanding behavior documented by Gompers (1996). 2.7 Liquidity pressure at IPO lock-up expirations Field and Hanka (2001) report that the lockup expiration phenomenon a permanent decline in stock prices and abnormally high trading volume around the IPO lock-up expiration is stronger for newly public firms with venture capital backing and attribute this phenomenon to a particularly large amount of selling by VCs following the expiration. If VC funds limited lifespan causes liquidity pressure, we expect older funds approaching their liquidation date to be more likely to sell shares at the lock-up expiration. Specifically, we investigate whether the age of the VC fund at the time of expiration is positively related to trading volume and negatively 69

79 related to stock returns around lock-up expirations of their portfolio firms that recently went public. We measure abnormal trading volume (AVOL) relative to each firm s mean daily trading volume during the 45 trading days ending six trading days prior to lock-up expiration: AVOL 1 45 V it, 6 V t 50 it, 1 where is the average daily trading volume for firm i surrounding the lock-up expiration window beginning at day 0 and ending at day +5. Following Field and Hanka (2001), we compute cumulative abnormal returns surrounding lock-up expirations as follows: CAR i 1 1 R 1 R it, t 5 mt, 1 where CAR is the cumulative abnormal return of firm i,, is the daily stock return on day t i R it relative to the expiration date, and is the CRSP equal-weighted market index return. If the lock-up expiration falls on a non-trading day we take the next trading day as the date of expiration. The estimation window for CAR begins at day -5 and ends at day +1, capturing price changes both in anticipation of future insider sales as well as simultaneously with actual sales upon expiration. We use two proxies to capture VC funds liquidity pressure: i) the age of the oldest fund at the time of the lock-up expiration, and ii) the number of independent VC firms that are nine years or older at the time of the expiration. The second proxy is motivated by the idea that multiple VC funds facing liquidity pressure is likely to cause a more pronounced effect around 70

80 lock-up expirations. Control variables include the percentage of the firm s shares that were locked up prior to the expiration, the cumulative abnormal stock returns during the 45 trading days ending six days prior to the expiration, a dummy that equals one if pre-ipo shareholders sold any shares at the IPO, the natural logarithm of IPO proceeds, and year fixed effects. Robust t-statistics clustered at the industry level are reported in parentheses. The first three columns in Table 2.9 present regression results for the average abnormal volume observed around lock-up expirations. In column 1, the coefficient on the age of the oldest fund at the time of lock-up expiration is positive and statistically highly significant with a t-statistic of 3.5, revealing evidence consistent with pronounced selling around lock-up expirations by VC funds that are closer to liquidation. We also find a larger abnormal volume for firms with a larger fraction of shares released at the expiration and smaller abnormal volume following larger IPOs. Column 2 adds the number of independent VC firms that own shares of the IPO firm. The coefficent on the number of independent VCs is positive and statistically highly significant with a t-statistic of 2.5, suggesting that a larger number of VCs is associated with more selling at the lock-up expiration. Next, we split independent VCs into two groups by their age at expiration, and include their numbers separatly in column 3. We classify VCs that are 9 years or older at the time of the expiration as under liquidity pressure. We find that the number of VCs under liquidity pressure is significantly positively related to abnormal volume with a t- statistic of 2.3, whereas the number of VCs that are not yet under liquidity pressure is only marginally significantly positive with a t-statistic of 1.7. The results in column 3 indicate that the positive relation between the number of VCs and abnormal volume documented in column 2 is driven primarily by older VCs under liquidity pressure. 71

81 In columns 4 through 6, we investigate the relation between VCs liquidity pressure and abnormal stock returns around lock-up expirations. In column 4, the coefficient on the age of the oldest VC fund at the time of the expiration is negative and statistically significant with a t- statistic of -2.2, providing evidence that the increased trading volume documented in column 1 is associated with a significant decline in stock prices around lock-up expirations. Column 5 includes the number of independent VC firms, which turns out to be significantly negatively related to abnormal stock returns. Finally, column 6 splits the number of independent VCs into two groups by liquidity pressure. We find that the number of VCs that are nine years or older at the time of the lock-up expiration is significantly negatively related to abnorman returns with a t- statistic of -4.3, whereas the number of VCs that are not under liquidity pressure does not have a statistically significant coefficient. The coefficient of on the number of VCs that are nine years or older indicates that each additional VC firm under liquidity pressure is associated with a 40 basis points decline in stock returns around lock-up expirations. A one standard deviation increase in the number of VCs under liquidity pressure (from a mean of 2 to 4.9) is associated with a 1.16 percentage points decline in stock returns, which is economically significant compared to an unconditional average CAR of 4.25% for all VC-backed IPOs in our sample. Overall, the results from abnormal trading volume and abnormal stock returns analyses are consistent with each other and provide evidence consistent with the VC liquidity pressure hypothesis. 2.8 Conclusion In this paper, we investigate whether independent venture capital funds limited lifespan imposes a constraint on the general partners by subjecting the fund to liquidity pressure at the tail-end of the funds lifecycle. 72

82 We find that portfolio firms backed by independent VC funds approaching maturity are associated with quicker initial public offerings and selloffs, consistent with the idea that the venture capital exit cycle is influenced by liquidity pressure faced by older funds. These portfolio firms are also more likely to have a liquidity event during unfavorable market conditions and are more likely to be sold off rather than taken public. These findings raise a concern that the liquidity pressure faced by VC funds might lead to suboptimal exit outcomes. Turning our attention to initial public offerings of VC-backed firms, we find that IPO firms backed by VCs under liquidity pressure experience significantly larger trading volume and lower stock returns around their lockup expirations, and this lockup effect increases with the number of independent VC funds under liquidity pressure. Our results suggest that VC funds liquidity constraints impose externalities and influence the IPO process. Our evidence is consistent with the presence of agency conflicts between venture capitalists and their portfolio firms, as several key exit-related choices appear to be made in the VCs self interest. In addition, our finding that the significant stock price decline observed around lockup expirations is related to the VCs liquidity pressure supports the view that this enduring market anomaly is caused by downward sloping demand curves. 73

83 References 2 Bascha, A., Walz, U., Convertible securities and optimal exit decisions in venture capital finance. Journal of Corporate Finance 7, Bradley, D.J., Jordan, B.D., Ha-Chin, Y., Roten, I.C., VENTURE CAPITAL AND IPO LOCKUP EXPIRATION: AN EMPIRICAL ANALYSIS. Journal of Financial Research 24, Brav, A., Gompers, P.A., The Role of Lockups in Initial Public Offerings. Review of Financial Studies 16, 1-29 Chen, H., Gompers, P., Kovner, A., Lerner, J., Buy local? The geography of venture capital. Journal of Urban Economics 67, Cumming, D., Contracts and Exits in Venture Capital Finance. Review of Financial Studies 21, Da Rin, M., Hellmann, T.F., Puri, M., A survey of venture capital research. National Bureau of Economic Research Dittmar, A.K., Dittmar, R.F., The timing of financing decisions: An examination of the correlation in financing waves. Journal of Financial Economics 90, Field, L.C., Hanka, G., The Expiration of IPO Share Lockups. The Journal of Finance 56, Giot, P., Schwienbacher, A., IPOs, trade sales and liquidations: Modelling venture capital exits using survival analysis. Journal of Banking & Finance 31, Gompers, P., Xuan, Y., Bridge building in venture capital-backed acquisitions. In: AFA 2009 San Francisco Meetings Paper Gompers, P.A.q., Xuan, Y., Harvard Business, S., Bridge building in venture capitalbacked acquisitions. Harvard Business School, [Boston] Harford, J., What drives merger waves? Journal of Financial Economics 77, Hellmann, T., IPOs, acquisitions, and the use of convertible securities in venture capital. Journal of Financial Economics 81, Hochberg, Y.V., Ljungqvist, A., Lu, Y., Whom You Know Matters: Venture Capital Networks and Investment Performance. The Journal of Finance 62, Ince, O., Double Intermediation in IPOs:Double the trouble? Kandel, E., Leshchinskii, D., Yuklea, H., VC Funds: Aging Brings Myopia. Journal of Financial & Quantitative Analysis 46, Lee, P.M., Wahal, S., Grandstanding, certification and the underpricing of venture capital backed IPOs. Journal of Financial Economics 73, Lerner, J., Venture capitalists and the decision to go public. Journal of Financial Economics 35, Lindsey, L., Blurring Firm Boundaries: The Role of Venture Capital in Strategic Alliances. The Journal of Finance 63, Lowry, M., Why does IPO volume fluctuate so much? Journal of Financial Economics 67, 3-40 Lowry, M., Schwert, G.W., IPO Market Cycles: Bubbles or Sequential Learning? The Journal of Finance 57, Masulis, R.W., Nahata, R., Venture Capital Conflicts of Interest: Evidence from Acquisitions of Venture-Backed Firms. Journal of Financial & Quantitative Analysis 46,

84 Metrick, A., Yasuda, A., Venture Capital and Other Private Equity: A Survey. SSRN elibrary Nahata, R., Venture capital reputation and investment performance. Journal of Financial Economics 90, Roberts, M., Whited, T., Endogeneity in empirical corporate finance. Sahlman, W.A., The structure and governance of venture-capital organizations. Journal of Financial Economics 27, Smith, G., The Exit Structure of Venture Capital. UCLA Law review 75

85 Frequency Figure 2.1. Histogram of exits by VC age categorized by exit method This chart depicts the frequency of VC-backed exits by the age of the VC fund at the time of the exit, categorized by the method of exit. The sample includes VC investments by independent VC firms in companies with successful exits (IPO or trade sale) between 1985 and Fund age at exit is the age (in years) of the VC fund at the time of the exit. 16.0% 14.0% 12.0% 10.0% 8.0% 6.0% 4.0% 2.0% 0.0% VC Fund Age at Exit Series1 Series2 76