SOMMARIO:

Abstract - Introduction - 1. The lock-up period - 2. The data set - 3. Methodology - 4. Results - 5. Conclusions - References - 6. Appendix

Abstract

In connection to a company’s initial public offering (“IPO”), insider shareholders are often subject to lock-up agreements, prohibiting these shareholdersfrom trading in a specified time period after the IPO. Since the period and shareholders subject to lock-ups are clearly stated in the prospectus, being public information, any price reaction should already be predicted and there should be abnormal returns connected to the lock-up expiration, according to the weak form of the efficient market hypothesis. In the literature, there is a recent debate about the truth of the last sentence: some studies claim against market efficiency trying to demonstrate the presence of statistically significant abnormal returns, while other studies affirm either that abnormal returns are not significant or that abnormal returns depend on other effects different from lock-up agreements. We use the event study methodology toinvestigate the statistical significance of abnormal returns and of abnormal trading volume in a time window centered at the end of lock-up period (usually 180 days), applied to different IPOs within American markets, between 2003 and 2007. We found significant abnormal returns and abnormaltrading volumes in the proximity of the end of lock-up, important correlation between cumulative abnormal returns and stock performances and also important correlation between cumulative abnormal returns and the presence of venture capitalist.

Introduction

Lock-up agreements, within the initial public offering topic, often prohibit insider shareholders from trade their securities for a specified time period, typically 180 days, after the IPO operations. Lockups are not mandated by regulators but are contractual agreements betweenunderwriters and issuers and the terms of the lock-up are explicitly disclosed in the IPO prospectus, that is public before the IPO first day. Financial economists have offered three distinct rationales for the existence of lock-ups: • a signaling solution to the adverse problem; • a contracting solution to the moral hazard problem; • rent extraction mechanism by powerful underwriters. Empirical evidence regarding the signaling hypothesis is mixed: (Brav and Gompers, 2003)[4] reject it, while (Brau, Lambson et al., 2005)[3] find empirical support for it. (Brav and Gompers, 2003)[4] suggest that lock-ups serve as a commitment mechanism for the moral hazard problem. They do not find any evidence to support the rent extraction hypothesis. Most researches, however, agree that these agreements are intended to reduce information asymmetry with a signaling function so that they reassure the market about the permanence for at least some months of the key shareholders and employees; signal that insiders are nottrying to cash out at an early stage, by trading on private information and temporarily protect the stock price constraining the supply ofshares. Empirical evidence regarding effects of lock-up expiration has been documented by (Field and Hanka, 2001)[6] and (Bradley, Jordan et al., 2001)[1]. There is overwhelming evidence of a significant negative stock price reaction at lock-up expiration. These researchers also documented a concomitant increase in trading volume after the unlock day. The finding of negative stock price reaction challenges the more extreme versions of the efficient markets hypothesis. The efficient market hypothesis (EMH) (Fama, 1970)[5] claims that asset prices fully reflect all available information. A direct implication is that it is impossible to “beat the market” [continua ..]

1. The lock-up period

At an initial public offering, the existing shareholders rarely sell the entire company. Instead, approximately 15-20% of the shares areissued to the public. Though not a legal requirement, it is a standard arrangement for the underwriters to insist upon the shares of the remaining 80-85% shareholders to be restricted from sale for a certain period of time. This period, the so-called lock-up period, is one wayof aligning the incentives of the current owners and new owners, at least during the initial stages of the company being public. There are norules regarding the length of the lock-up period; however, the majority of lock-up periods last 180 days, or approximately 6 months. As an illustration of the lock-up language, in (Ofek and P. Richardson, 2000)[9] the authors report the following excerpt from a typical IPO prospectus: holders of at least approximately 96% of all outstanding Common Stock, as well as the Company, have agreed that, during a period of 180 days from the date of this Prospectus, the Company and such holders will not, without the prior written consent of the U.S. and International Representatives, directly or indirectly, sell, offer to sell, grant any option for the sale of or otherwise dispose of any Common Stock or any security convertible or exchangeable into or exercisable for Common Stock (except for Common Stock issued as partof the Offerings) or, in the case of any holder of Common Stock, exercise any right to have securities of the Company registered by the Company under the Securities Act. From an economic viewpoint, this period is interesting for several reasons. First, after the lock-up period, it is reasonable to expect certain characteristics of stocks trading. In particular, due to the shift in the supply of available shares, we would expect trading volume to rise. This is because the quantity of liquidity trading (i.e., non information-based) increases in [continua ..]

2. The data set

Our study is based on a sample of 50 selected stocks listed in the U.S. markets (at New York Stock Exchange and Nasdaq). The lock-up date has been set between December 2003 to June 2007 and the lock-up period has been set with a length equal to 180 days, or about 6 months. The initial share price was not less than 10 $ (usd). We decided for this period in order to analyze a period rich of IPOs and before the 2007 financial crisis. All the data have been taken, courtesy, from “Kenney, Martin and Donald Patton. 2010. Firm Database of Initial Public Offerings(Version B)”. This database is comprised of all “de novo” initial public offerings (IPOs) on American stock exchanges and filed with the Securities and Exchange Commission (SEC). In addition, we have also used data taken from Bloomberg database. For every stock, we have reported and considered in our analysis the daily close price from 90 days to + 90 days, with respect to lock-up expiration day; the corresponding daily index price related to the stock; the corresponding stock traded volume; the total equity, as stated within the IPO prospectus; the percentage of locked-up stock, as stated in IPO prospectus and the Venture Capitalist/Equity Fund presence, as stated in IPO prospectus.

3. Methodology

In order to detect and search significant abnormal returns, we have implemented the event study methodology as follows: 1. computation of the daily returns for every stock and indexreturns; 2. definitionofthree time windows: the event window, at the middle of which the event happens, the estimation window and the post-event window; 3. estimation of the model followed by the stock in the estimation window/post-eventwindow; 4. comparison,forevery stock, of the daily returns in the event window with the expected returns from the previous determined model:we obtained the abnormal returns (AR) and, through the sum, the cumulative abnormal return (CAR). 5. average through the differentstocks; 6. check of the significance of the averagedvalues. By looking at the time window, we assume that the event-window lasts 10 days before and 10 days after the lock-up expiration day (4 working weeks). The estimation-window lasts 80 days: this number is a trade-off between the necessity of enough data for the market model estimation and the limited length of the lock-up period (it is necessary also to discard the first days after IPO to make the price stabilized). The post event-window lasts as the estimation-window. This kind of time window partition has been adopted by some authors, for more see (Bradley, Jordan et al., 2001)[1]. The following picture explains the time window partition. Figure 1 – Time window definition 3.1. Returns and the market model Stock returns ( ps ) have been computed as: Equity Index returns ( pm ) have been computed as: The market model is a linear regression between market and stock returns given by: By defining: that is substituting the coefficients which fit the equation “best”, in the sense of solving the quadratic minimization problem, we are able to define the Abnormal Returns (AR) as: By assuming that in the linear regression model the errors εs(t) have expectation zero and are uncorrelated: E[εs] =0 theyare considered to be homoscedastic: V ar(εs) = σ2 < ∞ distinct error terms are uncorrelated: that is equal to assume that Gauss-Markow assumptions [continua ..]

4. Results

After computing CAR(t) and AV(t) for every stock included in sample, an average through the 50 stocks has been performed, getting the final estimations (as described early). Regarding the statistics of the final CAR, we get that the average through the time is very close to zero (0.000254 less than 1/10 of standard deviation), as expected. About the standard deviation, we implement two method of computation: – A priori method (σ = 0.00366): from the variance of AR, we have derived the variance of CAR (5 times AR variance) for every stock.Then, we average the CAR-variance through the stocks. This method assumes the independence of stock return, that is true if they are considered in different period of time. In this article, it is not true for some couples of stocks, so the final variance can be underestimated; – A posteriori method (σ = 0.00399): we extract the variance from the final CAR, using the common estimator. In this case, the elements of the sample are not independent (e.g. CAR(t) is not independent from CAR(t-1)) so the estimator can be biased, but the very small average makes this effect rather negligible [see appendix]. Regarding the Abnormal Volume, the distribution of the AV for one stock is difficult to determine. The final AV is an average of 50 stock AV: if we suppose the stock AVs are independent and identically distributed (i.i.d.), it is possible to apply the central limit theorem and to claim that the final AV distribution is near a normal distribution. This claim is not rigorous (the stock AVs are not i.i.d.) but can help to understand final AV behavior. The final AV average, through the time, is near 1 (1.061), as expected, and the standard deviation is 0.226. Next graphs explain these concepts. The significance of the estimations, for the event window day, has been studied using the t-statistic (given the small sample size and population standard deviation unknown). Figure 2 – Market Model from estimation window; a priori computed vaiance. We can note a confidence near 95% for the days before Lock-up(especially for -3 day). Figure 3 – Market Model from estimation window; a posteriori computed variance. The variance is [continua ..]

5. Conclusions

Talking about hypothesis checking, we have found that Cumulative Abnormal Returns are significantly correlated with stockperformances and this kind of correlation is to be considered as negative. It seems to suggest that if the stock well performs during estimationwindow, the market could believe that after lock-up day the insiders want to sell their stocks to cash in the earnings; for this reason the stockprices could go down. Cumulative Abnormal Returns are also significantly correlated with the presence of Venture Capitalists and/or Private Equity Funds; this kind of correlation is to be considered, again, as negative. In this case, the statistical significance is close to 100% and thiscould imply that Venture Capitalists are expected to be less likely than firm insiders to hold long-term their shares and more likely to salethem at the first opportunity to do so. Cumulative Abnormal Returns seem to be negatively correlated with the amount of locked-up shares,but without a high significance. In addition, we have found partial evidence that our explanatory hypothesis (supply shock and insiders willingness to cash-in) could be true, but to obtain stronger evidences we should adopt a very bigger sample and we should know better the insiders behaviors and the possible presence of spurious events near lock-up expiration day that can bias our conclusions (for example, in (Ofek and P. Richardson, 2000)[9] the authors use more than 1000 stocks in their sample). The value of R2 in explaining the correlation between Abnormal Volume and CAR is very low, this suggests that other phenomena can cause the Abnormal Volume and, a bigger sample, could help to better understand the phenomenon. Summing up, in our work we have applied event study methodology to 50 stocks during their IPO lock-up [continua ..]

References

[1] Daniel J. Bradley, Bradford D. Jordan, Ha-Chin Yi, and Ivan C. Roten. Venture capital and ipo lockup expiration: An empirical analysis. Journal of Financial Research, 24(4):465-493, 2001. [2] James C. Brau, David A. Carter, Stephen E. Christophe, and Kimberly G. Key. Market reaction to the expiration of ipo lockup provisions. Managerial Finance, 30(1):75-91, 2004. [3] James C. Brau, Val E. Lambson, and Grant McQueen. Lockups revisited. Journal of Financial and Quantitative Analysis, 40(3):519-530, 2005. [4] Alon Brav and Paul Gompers. The role of lockups in initial public offerings. Review of Financial Studies, 16(1):1-29, 2003. [5] Eugene F. Fama. Efficient capital markets: A review of theory and empirical work. The Journal of Finance, 25(2):383-417, 1970. [6] Laura Casares Field and Gordon Hanka. The expiration of ipo share lockups. The Journal of Finance, 56(2):471-500, 2001. [7] H. Geijer L.Beck-Friis. Ipo share lockups unlock day effect : An empirical study on nordic data. University essay from Stockholm/Institutionen for finansiell eko­nomi, 2010. [8] A. Craig MacKinlay. Event studies in economics and finance. Journal of Economic Literature, 35(1):13-39, 1997. [9] Eli Ofek and Matthew P. Richardson. The ipo lock-up period: Implications for market efficiency and downward sloping demand curves. New York University Working Paper No. FIN-99-054 2000, 2000.

6. Appendix