Liquidity and IPO performance in the last decade

Transcription

1 Liquidity and IPO performance in the last decade Saurav Roychoudhury Associate Professor School of Management and Leadership Capital University Abstract It is well documented by that if long run IPO underperformance is measured by cumulative average buy and hold returns, the IPOs underperform a sample of size matched firms. However, if the multifactor framework is considered, some results are mixed as some studies have found underperformance and some have reported no underperformance either in identical sample periods or different sample periods. My paper uses the multifactor framework and finds underperformance depends on the liquidity profile of the IPOs. By separating IPOs into sub-samples based on their liquidity, I find that IPO performance during 2000 to 2010 differs significantly based on the liquidity profile of an IPO. The IPOs with high initial liquidity tend to underperform their peers who start with low initial liquidity if they cannot sustain their high liquidity levels. Keywords: Underperformance, Liquidity, IPO, Multifactor model 1. Introduction The decade ( ) started with a bang in the IPO market in the United States but quickly fizzled out only to regain some ground during 2004 to 2007 but again falling sharply in The period also experienced high and growing liquidity in the US market that was dramatic (Chordia et. al., 2011). It is interesting to see how the IPOs which debuted in the past decade fared and I provide some evidence and explanation that a part of their performance was due to liquidity effects. 2. Literature Review It is well established in literature that if IPO underperformance is measured by three year cumulative average buy and hold returns, the IPOs underperform a sample of size matched firms (see Ritter (1991)). Welch and Ritter (2003) find that, from 1980 to 2001, an average IPO firm underperformed the CRSP value-weighted market index by 23.4% and underperformed seasoned companies with the same market capitalization and book-to-market ratio by 5.1%. Although there appears to be evidence on the long run underperformance of IPOs, the source of this underperformance is not completely clear. Shiller (1990) emphasizes the influence of behavioral fads in the market leading to long run IPO underperformance. Ritter (1991) finds that younger firms and companies going public in heavy-volume years tend to underperform more than other firms. Teoh, Welch, and Wong (1998) attribute some of the poor post-ipo stock performance to optimistic accounting early in the life of a firm, suggesting that at least a part of the poor long-run performance is because of a market that is overly optimistic and unable to properly forecast rough times. Previously, when using a multi factor framework, some studies have found underperformance, such

2 as Loughran and Ritter, (2000) while some have reported no underperformance, such as Brav and Gompers (1997) or Eckbo and Norli (2005). Our explanation for this can is that the extent of long run performance of IPOs in a sample period is determined by the liquidity profile of IPOs. Amihud and Mendelson (1986, 1989) present the concept that more liquid firms have lower expected returns than other firms. An investor will only be willing to purchase an illiquid stock if there is a premium to buying the stock. The liquidity hypothesis suggests that holders of less liquid stocks will demand higher expected returns as a result of bearing more liquidity risk. Brennan and Subrahmanyam (1996), Datar, Naik, and Radcliffe (1998), Chordia and Subrah manyam (2001), all find a negative relationship between liquidity and subsequent returns. I argue that if liquidity is not priced appropriately for some IPOs then when the IPO starts to trade then that could lead to either underperformance or outperformance depending on its liquidity profile. If the level of liquidity observed in the IPO after-market in the short run cannot be sustained in the long run, then as the liquidity declines and the expected rate of return increases to its long run equilibrium value. This causes underperformance in the long run. 3. Methodology Our sample consists of monthly observations spanning all listed securities for three equity markets (NYSE, AMEX, and NASDAQ) over the time period January, 2000 to December, The new IPO listings were obtained from Jay Ritter s website at which contains founding dates and CRSP permanent IDs (permno). The corresponding closing price, trading volume, returns, shares outstanding, market capitalization, exchange code are from the CRSP database. The securities ineligible for continued listing (priced at less than one dollar per share, less than 750,000 shares outstanding, or number of shareholders fewer than 300), were deleted from the sample in order to avoid any delisting bias. Also excluded were closed-end funds, ADRs (American Depository Receipts, issues by foreign firms which have listings in at least one other market outside the US), REITs (real estate investment trusts), units, mutual-to-stock conversions, preferred stocks, penny stocks (with offer price less than $5), and financial companies and IPOs which had less than 150 days of trading activity in the 250 trading days following the offer. Our IPO sample consists of 1,310 firms. Figure 1 shows the number of IPOs issued during each of the years from 2000 to Figure IPO listings in the United States during

3 To measure liquidity I use monthly share turnover, which is the ratio of trading average monthly volume to the number of shares outstanding. This gives a direct measure of the asset trading frequency. Datar, et al. (1998) finds that measures based on bid-ask spreads and price impacts may not capture liquidity to the extent that large shareholders wish to make transactions. Turnover is a widely used proxy for liquidity (Chordia et al., 2000). Following Gao and Ritter (2010) I make adjustments to the trading volume for Nasdaq-listed IPOs. The trading volume is divided by 2 for January 2000-January 2001, by 1.8 for February December 2001, and by 1.6 for to allow more meaningful comparisons with NYSE and Amex-listed IPOs. During 2002, securities firms began to charge institutional investors commissions on Nasdaq trades, rather than the prior practice of merely marking up or down the net price, resulting in a further reduction in reported volume of approximately 10%. Thus, for 2002 and 2003, I divide Nasdaq volume by 1.6. For 2004 and later years, in which much of the volume of Nasdaq (an d NYSE)stocks has been occurring on crossing networks and other venues, we use a divisor of 1.0, reflecting the fact that there are no longer important differences in the reporting of Nasdaq and NYSE volumes. Chordia et. al. (2011) finds that the share turnover has increased dramatically over the last decade especially during Figure 2 shows a similar pattern for mean and median values of IPO monthly turnover over years 2000 to The average monthly turnover was 2.07 and the median monthly turnover was But both the mean and median values follow similar patterns. Fitting the monthly IPO turnover data with a linear trend gives a positive and statistically significant coefficient of 0.19 with an Adjusted R-squared of 72%. A positive significant trend creates a problem in creating turnover and size breakpoints for the entire sample period. Rather I need to create those breakpoints at the beginning of each year and adjust the turnover breakpoints accordingly. Size, as measured by the market capitalization, is a standard criterion for sorting stocks in empirical investment studies. Data for each month is sorted based on the combined ranking by the size of the NYSE, AMEX and NASDAQ stocks. In order to distinguish IPOs based on their liquidity profile, one option is to simply sort them by stock turnover. However, I find smaller stocks have low turnover compared to larger stocks but shows a monotonic increase with the size of the firm, which

4 is also reported by Pastor and Stambaugh (2003) among others. If turnover and size have a positive relationship, it would be difficult to separate the size effect of the IPOs if they are ranked by turnover alone. To create the three size quintile breakpoints for each year only the non-ipo stocks were included but as IPO firms tend to be small firms I set the maximum value of the market capitalization for the non-ipos to be no more than the market capitalization of the largest IPO firm. I also create the three turnover quintiles each year from the truncated non-ipo dataset that was used to compute the size quintiles. Pairing the size quintiles with turnover quintiles gives us a 3x3 matrix using turnover and size quintiles for each of the 11 years. The breakpoints are created using low turnover (LT), medium turnover (MT) and high turnover (HT) and small size (S C), medium size (MC) and large size (L C). The nine pairs were ranked in descending order of turnover as 1.(HT,LC), 2.(HT,MC), 3.(HT,SC),4.(MT/LC), 5.(MT/MC),6.(MT/SC), 7.(LT, LC),8.(LT, MC), and 9. (LT, SC). Any new IPO in a given year is assigned a rank depending on its size and turnover quintiles. To categorize IPOs with low turnover I consider rankings from 7 to 9. Table 1 panel A, shows the median monthly turnover for values of the monthly turnover for the quintiles LT, MT and HT. Panel B reports the median values of the three size quintiles. Table1 Panel A: Median Monthly Turnover Year Low (LT) Medium (MT) High (HT) Panel B: Median Market Capitalization (in $000s) Year Small Cap (SC) Medium Cap (MC) Large Cap (LC) , ,221 1,275, , ,701 1,098, , ,852 1,023, , ,613 1,019, , ,922 1,035, , ,758 1,069, , ,155 1,151, , ,806 1,207, , ,546 1,171, , ,530 1,138, , ,512 1,228,625

5 I discard the IPOs from the sample which did not undergo a change in rankings between their first and third anniversary and were ranked 4 to 6 (middle turnover and size quintiles). To differentiate between the short run and the long run, most IPO studies have categorized long run returns as returns for more than one year. Therefore, any firm from a portfolio is excluded if it has any of the intermediate months missing between the 24 th and 36 th month, or if the life of the IPO is less than 36 months. I ended up with 541 IPOs which is approximately 41% of our original sample of 1310 IPOs. Among the 541 IPOs, there were 359 IPOs that moved down the rank implying relatively lower liquidity than their initial value. The long run performance is measured in a multi-factor asset pricing framework for the whole IPO sample as well as the groups created with low and high turnover IPOs. Time-series regressions have an advantage over traditional event-study procedures in that the regressions can explicitly control for temporal dependence in returns. (see Appendix A for a discussion). In each calendar month over the entire sample period, a portfolio is constructed comprising all IPO firms experiencing the event within the previous 36 months. Since the number of event firms is not uniformly distributed over the sample period, the number of firms included in a portfolio is not constant through time. As a result, some new firms are added each month and some firms exit each month. Accordingly, the portfolios are rebalanced each month and an equal-weighted portfolio excess return is calculated. The resulting time series of monthly excess returns is regressed on the three Fama-French (1993) factors and the Carhart (1997) four-factor model. The regression equation is therefore: (1) where R IPO, t is the return to IPO portfolio for month t, R ft is the one month Treasury-Bill return for month t, and MKTt is the excess market return to the CRSP value-weighted index for month t. The factor loading b on the market premium MKT gives the CAPM beta of the IPO portfolio. SMBt is the realization on a size-based portfolio that buys small cap stocks and sells large cap stocks. Similarly, HMLt is the realization on a factor portfolio that buys high Book-to-Market stocks and sells low Book-to-Market stocks. The s and h coefficients measure the sensitivity of the portfolio's return to the small-minus-big and high-minus-low factors respectively. Portfolios of value stocks should have a high positive value for h, while growth portfolios should have a high negative h. The largest capitalization portfolio should have a negative value for s, and small capitalization portfolios should have a large positive factor loading on SMB (s should be positive). UMD is the momentum factor as in Carhart (1997). The alpha ( ) in the regression is the abnormal return for that investment portfolio. 4. Findings R IPO, t R f t bmkt t ssmb Table 2 reports time series multifactor regressions on equal-weighted excess returns of a portfolio of all IPO firms in the sample which have issued equity for cash during the prior 36 months. The excess return is calculated each month by subtracting the 1-month Treasury bill rate from the calendar time IPO portfolio return for that month. I run regressions (equation 1) using the fu ll sample from year 2000 to 2010 and another by dropping the IPOs which debuted in year Column (1) and (2) of table 2 refers to the Fama French and the 4 factor model results for the truncated sample. Columns (3) and (4) report the results for the entire sample. The factor loading on the market premium in the Fama French model for both sample periods is above 1.3 and highly significant. This suggests that IPO firms have high market risk as measured by the beta. The coefficient on the size factor is less than 1 and is also highly significant. The loading on the value factor, HML, is significantly negative which is expected as IPO firms are usually growth firms. Newey West standard errors are in parentheses and *,**,*** denote significance at t hhml t u UMD t t

6 10%,5% & 1% respectively. The Jensen s alpha which measures abnormal performance is negative at -0.19% and significant at 10% for the entire if regressed against the Fama French model. This translates to an underperformance of about 2.3% per year. Loughran and Ritter (1999) used a similar model for the time period from 1973 to 1996, and found Jensen s alpha to be -0.40% and significant at 1%. For the time period 1990 to 2000, Welch and Ritter (2002) find the underperformance of -0.48% with significance at 5%. Table 2 IPO performance in a Multifactor Framework Years: Years: (1) (2) (3) (4) FF-3 factor Carhart(1997) FF-3 factor Carhart(1997) * (0.0052) (0.0051) (0.0011) (0.0021) MKT *** *** *** *** (0.1502) (0.1766) (0.1972) (0.1389) SMB *** *** *** 1.077*** (0.1776) (0.1562) (0.1806) (0.1547) HML ** *** *** *** (0.0369) (0.0634) (0.1266) (0.1828) UMD *** *** (0.0816) (0.0924) Adj. R-squared No. of Observations Column (2) and (4) reports the regression results during the same sample period for the Carhart four factor model which includes the momentum factor UMD in addition to the Fama French factors. The factor loadings for the market, size, and value factors have the same sign and significance as in three the factor Fama French model. The UMD factor has negative and highly significant factor loading. The Jensen s alpha is negative but not significant for the entire sample. To compare Jensen s alpha coefficient over each year, we repeat the regression for the Fama French factors only for each of the years starting from This is because IPOs which have less than 3 years of post- issue performance were not included. The early 2000 saw one of the greatest stock market increases in history with NASDAQ (reaching above 5000 points in March, 2000), and a record number of new issues came into the market in the first half of With the terrorist attack in U.S. soil in September, 2001, the market declined a bit but the sharpest decline started in early 2002 with the fall of Enron. Within a few months, NASDAQ had fallen to levels of Table 3 reports the results for different years for the Fama French 3 factor regression. Years 2003, 2005, 2006, 2009 and 2010 saw negative alphas which were statistically significant. However, the only positive and significant alpha among those years came in 2008, the year which started the great recession. Table 3

7 IPO Performance for Different Years Year Alpha Adj. R-squared * (0.0026) ** (0.0026) (0.0014) ** (0.0037) (0.0003) * (0.0032) ** (0.0027) ** (0.0032) The results from table 2 show that the extent, to which IPOs underperform, as measured by Jensen s alpha, varies significantly across years. This is consistent with the findings of Loughran and Ritter (2000) and Welch and Ritter (2002). Baker and Wurgler (2006) find that in periods of high investor sentiment, stocks that are attractive to optimists and speculators and at the same time unattractive to arbitrageurs like IPOs, such as small stocks and extreme growth stocks, tend to earn relatively low subsequent returns. The Table 4 Long run Performance of Low Turnover and High Turnover portfolios Low Turnover High Turnover (1) (2) (3) (4) FF-3 factor Carhart(1997) FF-3 factor Carhart(1997) ** (0.0014) (0.0018) (0.001) (0.0015) MKT 1.239*** 1.017*** 1.425*** 1.186*** (0.256) (0.1591) (0.2068) (0.1089) SMB.8996***.9575*** 1.003*** 1.174*** (0.183) (0.1765) (0.2662) (0.2753) HML * * ** ** (0.182) (0.1795) (0.1829) (0.2613) UMD *** *** (0.0818) (0.2387) R-squared No. of Observations

8 focus will now turn to the two sub sample portfolios of IPOs created using the low and high turnover portfolios. The results will first show how the two portfolios performed in calendar time for a holding period of three years. It should be noted that the portfolio can only be formed by looking at the liquidity of the first year of IPOs, so the results are ex-post and only indicate how the two portfolios performed overall. The results presented here can only give some idea of how the factor loadings have changed from the portfolio formation period to the actual evaluation period. Table 4 reports the multi factor regression results. Columns (1) and (3) give the Fama French regression results for low and high turnover portfolios respectively. The monthly alpha of 0.12% for low turnover IPO is positive but insignificant; the corresponding number for the high turnover portfolio is negative and significant with a value of -0.22% which translates to about 2.64% annual underperformance. Additionally, the loading on the market factor is higher for the high turnover portfolio, possibly indicating that those stocks are more exposed to market risk. The loading on the size factor is not markedly different, though the low turnover portfolio has lower coefficients. The value factor both liquidity portfolios is negative. The momentum factor is about twice higher in absolute magnitude for the high turnover firms than the low turnover firm which is expected. In the sample period the abnormal return was always negative for the high turnover portfolio (though was insignificant in the Carhart model) but positive and insignificant in the low turnover portfolio suggesting that any underperformance that was seen in IPOs was confined to the high liquidity IPOs. 5. Conclusion Though a positive contemporaneous relationship exists between liquidity and returns the relationship between current liquidity and expected returns is negative. This has significant implications about the role of observed liquidity as an indicator of future returns. Early high turnover seems to be an indicator of IPO underperformance in the long run. Specifically, if there are more IPO firms characterized by high liquidity in a given year, then the subsequent future returns could show underperformance. If there is high liquidity to begin with, the expected return on the IPO is lower in the short run because of low illiquidity premium. If the level of liquidity observed at the IPO after-market in the short run cannot be sustained over the long run, then as the liquidity declines, the expected rate of return will rise towards its long run equilibrium value causing underperformance. Two sub-samples of IPOs based on their low and high turnover show that high turnover IPOs underperform compared to its low turnover IPO peers in the long run after the initial one year portfolio formation period following the IPO. Empirical evidence on the long term performance of IPOs from 2000 to 2010 using a multifactor asset pricing framework show that there is underperformance when the market, size and value factors are considered. However, the underperformance disappears if I include the momentum factor in the framework or if we drop IPOs which debuted in year 2000 from our sample. The underperformance measured in the factor framework is sensitive to the sample period which supports the findings of Loughran and Ritter (2000). If the multi- factor framework is considered, some studies find underperformance and some have reported no underperformance either in identical sample periods or different sample periods. A potential explanation for this can be the fact that the extent of long run performance of IPOs in a

9 sample period is determined by the liquidity profile of the IPOs. More specifically, if there are more IPO firms which are characterized by positive excess liquidity, the subsequent future returns would show underperformance if measured above a period of one to three years. 6. References Amihud, Y., and H. Mendelson Asset Pricing and the Bid-Ask Spread, Journal of Financial Economics 17, Amihud, Y. and H. Mendelson The Effects of Beta, Bid-Ask Spread, Residual Risk, and Size on Stock Returns, Journal of Finance 44, Baker, M., and J. Wurgler Investor Sentiment and the Cross-Section of Stock Returns. Journal of Finance 61, Brav, A. and P. A. Gompers Myth or Reality? Long-Run Underperformance of Initial Public Offerings: Evidence from Venture Capital and Non-Venture Capital Backed IPOs, Journal of Finance 52, Carhart, M On Persistence in Mutual Fund Performance, Journal of Finance 52, Chordia, T., R. Roll, and A. Subrahmanyam Market liquidity and trading activity, Journal of Finance 20, Chordia, T., and A. Subrahmanyam Trading Activity and Expected stock returns, Journal of Financial Economics 59, 3-32 Chordia, T., and A. Subrahmanyam, and V. R. Anshuman Recent Trends in Trading Activity and Market Quality, Journal of Financial Economics 101, 2, Datar, V.T., N. Naik, and R. Radcliffe Liquidity and Stock Returns: An Alternative Test, Journal of Financial Markets, Eckbo, E. B., and Ø. Norli Liquidity Risk, Leverage and Long-Run IPO Returns, Journal of Corporate Finance 11, Fama, E. F., and K. French The Cross-Section of Expected Stock Returns, Journal of Finance 47, Fama, E. F., and K. French Common Risk Factors in the Returns on Stock and Bonds, Journal of Financial Economics 33, 1, Gao, X., and J. Ritter The Marketing of Seasoned Equity Offerings. Journal of Financial Economics, vol. 97, no. 1, Loughran T. and J. R. Ritter Uniformly Least Powerful Tests of Market Efficiency, Journal of Financial Economics 55, Pastor, L., and R. Stambaugh Liquidity Risk and Stock Returns, Journal of Political Economy 11,

10 Ritter J.R The Long-Run Performance of Initial Public Offerings, Journal of Finance 47, Shiller, R.J Speculative Prices and Popular Models, Journal of Economic Perspectives 4, Teoh, S. H., I. Welch, and T. J. Wong Earnings Management and the Long-Run Market Performance of Initial Public Offerings, Journal of Finance 53, Welch I., and J. Ritter A review of IPO activity, pricing and allocations. Journal of Finance 57,