IPO Lockup Agreements in the Netherlands
Evidence from Firms Listing on the Amsterdam Stock Exchange
DUCO S.W. MERKENS
May 2007
UNIVERSITY OF GRONINGEN
Faculty of Management and Organization
MSc Finance
Under the supervision of
P. Smid
IPO Lockup Agreements in the Netherlands
Evidence from Firms Listing on the Amsterdam Stock Exchange
Abstract
With a sample of 71 initial public offerings (IPOs), this study investigates the influence of firm and shareholder characteristics on shareholder lockup agreements in the Netherlands. The results show no support for several hypotheses that may explain the duration of lockups. Event studies show a statistically significant abnormal return at the lockup expiration date of -0.8 percent and a statistically prominent three-day abnormal return of -2.0 percent. This result is not in line with the Efficient Market Hypothesis since the parameters of the lockup; the length and the number of shares, are well specified in the offering prospectus. Minor explanations are found for the anomalous abnormal returns with a multivariate analysis.
Keywords: Initial public offerings, lockup agreements, asymmetric information, market efficiency
Table of Contents
1
Introduction ... 4
2
Theoretical Background... 8
2.1
The Influence of Firm and Shareholder Characteristics on Lockup Contracts... 8
2.1.1
Information Asymmetry... 8
2.1.2
The Influence of Underwriters and Venture Capital... 11
2.2
Price Reactions at Lockup Expirations ... 12
2.2.1
Reasons for Observed Price Reactions at the Lockup Expiration ... 13
2.2.2
Magnitude of the Price Reaction at the Lockup Expiration... 14
2.3
Empirical Evidence on the Influence of Firm and Shareholder Characteristics
on Lockup Contracts ... 17
2.3.1
Information Asymmetry... 17
2.3.2
The Influence of Underwriters and Venture Capital... 19
2.4
Empirical Evidence on Price Reactions at Lockup Expirations ... 21
2.4.1
Reasons for Observed Price Reactions at the Lockup Expiration ... 23
2.4.2
Magnitude of the Price Reaction at the Lockup Expiration... 24
3
Hypothesis Development ... 27
4
Data and Methodology ... 31
4.1
Lockup Length using a Regression Analysis... 32
4.2
Event Study... 34
4.2.1
Multivariate Regression ... 37
5
Empirical Results... 39
5.1
Descriptive Statistics... 39
5.2
Regression Analysis... 42
5.3
Expiration Date Abnormal Return ... 45
5.3.1
Multivariate Analysis... 47
6
Conclusion ... 51
1
Introduction
An Initial Public Offering (IPO) represents the first opportunity for a firm to raise capital on a public stock exchange. The chance to sell common shares to investors in an IPO is very often a critical step in the life cycle of a firm. The foremost reason for an IPO is to raise capital for future growth. Ritter [1998] states on the advantages of going public (p.5): “Once the stock is publicly traded, this enhanced liquidity allows the company to raise capital on more favorable terms than if it had to compensate investors for the lack of liquidity associated with a privately-held company. Existing shareholders can sell their shares in open-market transactions.” Obviously there are also disadvantages, besides the direct cost of listing, also indirect costs like management effort. Listed firms also become subject to more scrutiny from regulators and investors.
An IPO also represents the first opportunity for insiders to capitalize the value of their ownership stake in a firm. In other words, flotations are an opportunity for insiders to share the firm’s risk with well-diversified investors. Insiders often decide to diversify in order to reduce asset risk, since their shares in the firm often represent a major part of their private wealth. Most corporate insiders issue about 15 to 20 percent of the entire company to the public in a flotation.
In order to align the interest of new shareholders with the interest of corporate insiders after the issue, (standard) contracts are created that restrict corporate insiders from selling a specific percentage of their shares for a certain period of time (i.e. it regulates the ability of insiders to cash out on their stock holdings). This (standard) contract is called a lockup contract or a lockup agreement. Requiring insiders to maintain a significant economic interest after the equity issue through a lockup agreement will reduce asymmetric information not only between insiders and outsiders, but also between insiders and the underwriter(s).
Euronext Amsterdam mentions1 that the rationale behind a lockup is to promote public confidence in the issuer by ensuring that founders, directors, and supervisory board members retain a financial interest and also stay financially involved with the offering.
Lockup contracts, found in the IPO prospectus, are typically phrased as follows in the Netherlands: Each of the Selling Shareholders has agreed with the Underwriters that it will not sell or otherwise transfer, other than in the Global Offer, any of our ordinary shares or securities exchangeable or convertible into, or exercisable for, our ordinary shares for a period of 360 days from the Settlement Date, except with the prior joint written consent of the Joint Global Coordinators and Joint Bookrunners, subject to certain exceptions.
The lockup agreement2 is a legally enforceable contract. Shareholders need to receive permission from the underwriters in order to be able to sell their shares before the lockup expiration. Ofek and Richardson [2000] state (p.4): “from time to time, the underwriting group does grant permission and allows early sales of shares by locked up shareholders, however, the percentage of shares that are unlocked prior to expiration is generally small.”
Across countries and across firms we see a huge variety in lockup characteristics. Lockup agreements in the US and the UK are voluntary, while lockup contracts in continental Europe are often regulated. Although often regulated, Euronext has no coordinated set of rules regarding lockup requirements.
The LSE imposes no minimum lockups since January 2000. Prior to 2000 only mineral companies and scientific research-based companies were subject to lockup periods. Espenlaub et al. [2002] discuss that UK lockup contracts show more variety in terms of contractual detail than US lockup contracts. For instance, issuers in the US use absolute calendar days or periods while in the UK the expiry date is often around other company announcements or events. The average duration of firms with absolute lockup days in the UK is 561 days. Lockup agreements in the US are often more standardized at 180 days [Ofek and Richardson, 2000]. The Neuer Markt in Germany requires all initial shareholders to lockup for 6 months in which no new issues are allowed. Although Nowak and Gropp [2000] discuss that the Deutsche Borse can give certain exemptions regarding this 6 month lockup rule. Goergen et al. [2006] conclude that many firms in Germany have voluntary lockups in place, since they find that only 43 percent of the lockup agreements follow the minimum requirements of a 100 percent lockup. Compared to Germany where all the incumbent shareholders are locked up, France has different rules for different insiders. The Nouveau Marche requires directors to be locked up, where the issuing firm has the choice to impose lockups for 80 percent of their shares for 1 year or 100 percent for 6 months. In France but also in Germany many firms impose more stringent lockups than necessary. Goergen et al. [2006] find that 52 percent of the firms impose stricter lockups than required.
The Nieuwe Markt in the Netherlands requires all founding shareholders, all managing shareholders and shareholders, sitting on the supervisory board, to be locked up for a minimum of 360 days and for 80 percent of their holdings. Before 24 November 2000, all shareholders with a holding of 5 percent or more were locked up depending on the firm’s results. Shares were locked up for the full 100 percent until a positive operating income and net income were reported, while 50 percent of the shares were locked up until the firm had reported positive operating income and net income for three years within a 5 year time span. The lockup would have been fully released
2
Gao [2005] notes (p.1): “the lockup restriction generally prohibits insiders from short selling, hedging
after the firm had reported both positive operating income and net income for three years within 5 years. Euronext Amsterdam requires an issuer’s founders, directors and supervisory board members not to dispose their shares for period of 180 days and in certain exceptional cases a lockup period of 360 days applies. This 360 days lockup rule applies to certain issuers that are active in the new economy.
A lockup contract sends a credible signal to outsiders that insiders will not sell their shares in advance of bad news. Brau et al. [2005] argue that insiders of good firms signal by putting and keeping (locking up) their money where their mouths are. While Courteau [1995] argues that a lockup signal will only be used by higher-quality firms, because insiders expose themselves to more risk with longer lockups (by holding undiversified portfolios). Brav and Gompers [2003] argue that a lockup can serve as a commitment device, because committing the insiders to keeping their shares over a certain period of time after the IPO makes it more likely that any private information becomes public knowledge (reducing moral hazard incentives).
Numerous studies on IPO lockup agreements, especially in the U.S., for example Field and Hanka [2001] and Brav and Gompers [2003], but also Angenendt et al. [2006] in Europe have focused on the lockup expiration effects of price and trading volume. Significant share price changes are possible at the lockup expiration as market participants infer private information from insider activity, since the lockup expiration date represents the first occasion for insiders to sell their shares in the secondary market after the public offering. The expiration date is interesting from an economic viewpoint since the parameters of the lockup are well specified in the prospectus. Therefore, firms should not, on average, be subject to an abnormal price reaction at the time of the lockup expiration, if markets correctly and rationally estimate and anticipate the release of shares at the lockup expiration. This argument is based on the semi-strong form of the Efficient Market Hypothesis.
The main objective in this study is to provide a detailed analysis on shareholder lockup agreements in the Netherlands. This main objective boils down to two main areas. First, this study wants to explore the influence of firm and shareholder characteristics on lockup contracts in the Netherlands, including the influence of regulations, while focusing on the lockup lengths part of lockup contracts. Second, this study wants to examine whether there are significant abnormal returns at or around the lockup expiration date, including a marginal exploration of the firm and shareholder characteristics that potentially cause these significant abnormal returns.
detailed analysis on the stock price reactions and its determinants at or around the IPO lockup expiration.
2
Theoretical Background
2.1 The Influence of Firm and Shareholder Characteristics on Lockup Contracts
2.1.1 Information Asymmetry
The IPO market represents a classic example of information asymmetry. At the time of the flotation, corporate insiders (founders, management, officers, and directors) have inside information about firm characteristics, but above all firm value, while outsiders have little information about the prospects of the firm. In other words, incumbent shareholders, often part of management, possess superior information relative to outside investors. This information advantage is especially profound for IPO firms, since IPO firms are typically young and their values are still uncertain. Due to superior knowledge of insiders relative to outside shareholders about fundamentals and potential growth, insider trading is strictly regulated by the Securities Exchange Act of 1934 in the U.S. Also in the Netherlands it is strictly forbidden to trade upon insiders’ information.
Overcoming asymmetric information problems, by obtaining and verifying detailed information, may not be economically practical for outsiders, even if possible, because direct information transfers are untrustworthy since insider interests are best served by inflating the value of the firm, when they go public.
Information asymmetries with the going public process create a Myers and Majluf [1984] adverse selection problem. Myers and Majluf [1984] show in their model that equity financing always has negative consequences due to the existence of information asymmetry between insiders and outsiders. Not only with respect to the value of the firm's assets in place but also with respect to the net present value of new investment projects. Their model shows that when new investment opportunities become obtainable, that management will favor an equity issue if the firm is overvalued. Investors being aware of this self-fulfilling behavior will defend their own interests by revaluing the stock negatively. Thus, selling overvalued shares in an equity offering will benefit existing shareholders at the expense of new shareholders. Investors interpret the new issue as a signal of overvaluation, if new shareholders are aware of management’s incentives and lack complete information about the offering.
The Signaling Hypothesis
and Pyle [1977] develop a model where insiders can signal higher quality at the time of the IPO by retaining a large stake after the flotation. They argue that insiders in high-quality firms are willing to hold more of the firm’s stock after the IPO and therefore remain more undiversified as they can thereby commit to the quality of their company. Leland and Pyle [1977] show, since words are unreliable and only self-fulfilling, that the information transfer needed to solve the adverse selection problem can only be accomplished through actions. By either investing or holding shares in their own firms and committing to forego the possibility of selling secondary shares in the market (bearing asset risk by holding undiversified portfolios), insiders of high-quality firms signal information to new shareholders who in turn can use the signaling information to value the firm in question more accurately. However, Gale and Stiglitz [1989] argue that a signaling strategy is not convincing to investors when insiders can sell their remaining shares on the secondary market immediately after the issue.
Courteau [1995] was the first who added voluntary lockups to the Leland and Pyle model. In her model high quality firms signal their quality by agreeing to longer lockups. A lockup signal will only be used by higher-quality firms, because insiders expose themselves to more risk with longer lockups (by holding undiversified portfolios). Oppositely, insiders of low-quality firms are unwilling to accept the additional cost of longer lockups. Courteau [1995] concludes that the length of the lockup period serves as a signaling mechanism that complements ownership retention. As a result the lockup agreement serves as a certain legal device enabling insiders to pre-commit to keeping a high stake after the offering. Goergen et al. [2006] also argue that lockups may serve as a complement mechanism to the percentage of shares retained by insiders after the IPO, where more stringent lockups add credibility to the signal.
Alternatively, Goergen et al. [2006] reason that if a sale of a large holding is regarded as a bad signal, firms may want to neutralize a bad signal by imposing stricter lockups. Thus, the lockup agreement may act as a certain substitute mechanism to the percentage of shares retained by insiders after the offering.
Brav and Gompers [2003] expect that firms, signaling quality through longer lockups, will either raise their offering price to get “more money at the table” or are likely to raise money in a following equity offering. Besides arguing that you can signal through the length of the lockup period, they mention that you can also signal the quality of a company with two other mechanisms, namely through underpricing and through the fraction of shares locked up in the lockup.
The Commitment Hypothesis
As mentioned before a lockup is a commitment by insiders not to sell shares for a specific length of time in the future. Therefore the commitment hypothesis is based on the idea that lockups might serve as a commitment device that alleviates moral hazard problems. Moral hazard problems exist because firm quality is only observable ex post and insiders may act in their own interest and not in the interest of shareholders in the after market. Brav and Gompers [2003] describe the commitment hypothesis as follows (p.5): “Holding constant the quality of the firms, those that, ex ante, suffer from a greater potential for insiders to take advantage of shareholders would need longer lockups to induce investors to buy into the offering.” Thus companies whose moral hazard incentives in the aftermarket are likely to be large, for example, younger firms and firms with low operating cash flows, will have to accept a longer lockup, if they want to persuade outsiders to buy stock in the offering. Before the lockup expiration date, information regarding the firm’s prospects, but also managements’ actions, can be revealed, for example, through product releases and project completions. Hence, investors are more confident about acquiring shares in an offering, where the ability to expropriate outsiders is greatly reduced. Therefore a lockup can serve as a commitment device, because committing the insiders to keep their shares after the IPO will increase the likelihood that negative private information becomes public knowledge. Brau et al. [2005] mention the two primary differences between the signaling and the commitment theory (p.524): “with signaling, insiders know more about the quality of the firm at the time of the IPO. With commitment, insiders know more about their actions (effort) after the IPO.” Obviously the key distinction exists between the timing and the nature of the information asymmetry.
2.1.2 The Influence of Underwriters and Venture Capital
The role of underwriters in the IPO market is twofold, namely underwriters facilitate the distribution of shares from the issuer to the investor and, secondly, underwriters reduce the inefficiency inherent in the IPO market (by overcoming less than optimal decisions at the corporate level).
Beatty and Ritter [1986] argue that reputable underwriters decrease asymmetric information between insiders and outsiders. IPO underpricing literature also shows that higher reputable Investment banks seem to underprice less, for instance, Franzke [2003] finds that companies backed by a prestigious underwriter demonstrate less underpricing at the offering. Loughran and Ritter [2004] find evidence that reputable underwriters have gained market share over time, potentially suggesting that firms experience advantages when they issue shares with good quality underwriters. However, they also [2004] find that historically high quality underwriters only underwrote less risky issues, although they find recent evidence that might suggest that this has changed.
Brav and Gompers [2003] put forward the idea that prestigious underwriters are unwilling to risk their own reputations with firms that will more likely expropriate new shareholders. Thus underwriters have a certain certification role that will reduce the need for longer lockups.
Goergen et al. [2006] also argue that more reputable underwriters are associated with higher quality issues and therefore firms that go public with more prestigious underwriters need less stringent lockup contracts. They also suggest that underwriters write lockup agreements in order to safeguard their reputations, since underwriters may lose reputation if unfavorable price movements or adverse information effects occur after the issue. Thus, contrary to less stringent lockups argued above, a reputable (lead) underwriter will demand more rigorous lockup agreements to safeguard its reputation in general.
Venture capitalists are important providers of capital to new firms. Private equity capital is typically provided by specialized and institutionally backed outside investors to new and growing firms. Venture capital investments are usually high risk, but also potentially very profitable ventures. When pioneering and growing firms have ideas and opportunities but need capital and resources, while venture capitalists need ideas and opportunities to grow their capital, then venture capitalists and new and growing firms are almost perfect complements in high risk, high reward projects.
exchange capital for an equity stake, but also provide monitoring and screening and more importantly are also involved in the decision making process [Jain and Kini, 2000]. In other words, the presence of venture capital signals quality because they are involved in all the key decisions in the firm. Megginson and Weiss [1991] argue that venture capitalists have a very strong incentive to keep up to their reputation with investors and underwriters in order to keep access to the IPO market. They also argue that the presence of venture capital, through their certification role, reduces the level of information asymmetry between issuer and prospective investors and consequently lowers the cost of going public. Espenlaub et al. [2002] mention that (p.5): “Venture Capitalists play a similar part as underwriters and auditors, in that, as repeated players in the IPO market, they may be able to commit themselves credibly to the accuracy and completeness of disclosed information as false certification would lead to the loss of valuable reputation…The cost, stringency and limited availability of VC backing may act as a screening device as only high-quality firms that expect to gain most from VC involvement seek investment by VCs.”
Goergen et al. [2006] argue that the presence of venture capitalists, given their certification role of firm quality, reduces the call for more stringent lockups. Also Brav and Gompers [2003] argue that adverse selection problems are less severe (signaling good quality) in venture capital backed firms and for this reason the length of the lockup period should be smaller. Espenlaub et al. [2002] also state that shorter lockups might also be the result of conflicts of interest between the venture capitalist and the underwriter; because of the affiliation between the underwriter and the venture capitalist (the underwriter might use, for example, the shorter lockup as an incentive for future business).
2.2 Price Reactions at Lockup Expirations
The lockup expiration date represents the first occasion for insiders to sell their shares in the secondary market after the IPO. The parameters of the lockup, the length and the number of shares, are well specified in the prospectus and hence widely known in advance of the lockup expiration date. Some prospectuses even explicitly warn of the negative price impact of insider selling upon lockup expiration.
market), on average, and therefore the average abnormal return should be insignificantly different from zero. This argument is based on the semi-strong form of the Efficient Market Hypothesis (EMH) by Fama [1965], which states that all public information about a company is already reflected in its stock price.
To understand the price reactions at the lockup expiration better, this study divides the lockup price reactions literature into two parts. The first and most important part investigates the reasons for the observed price reactions at the lockup expiration, while the second part looks at literature that tries to explain the magnitude of the price reactions.
2.2.1 Reasons for Observed Price Reactions at the Lockup Expiration
Temporary Price Pressure
Field and Hanka [2001] suggest that on the lockup expiration day, a large flow of insider sell orders may momentarily lower the share price due to price pressure. The temporary price pressure hypothesis is based on the idea that a temporary price decline (due to insider sell orders) may be essential in equilibrium in order to catch the attention of liquidity providers. With regard to downward sloping demand curves, the defining difference between price pressure and downward sloping demand curves is that the effect, a negative abnormal return, should be only temporarily. If the result is lasting (no rebound), it is interpreted as a downward sloping demand curve. The temporary price pressure hypothesis might be consistent with some forms of market efficiency, while the downward sloping demand curve hypothesis violates the semi-strong form of efficiency.
Worse-than-Expected Insider Sales
Thus, the “worse-than-expected insider sales hypothesis” is based on the idea that the negative average abnormal return could be driven by information signals that future insider sales are likely to be large to lower the future price, or in other words, large enough to drive the price down the demand curve. This also means that if there are no insider sales at the lockup expiration day that the abnormal return should be equal to zero or should be even positive.
Liquidity Effect
Ofek and Richardson [2000] also argue that the price reaction at the end of the lockup period could be explained by a liquidity effect. The liquidity based hypothesis is based on the idea that investors require a liquidity premium for illiquid stocks. Liquidity deals with the cost of buying and selling a stock. Brokerage fees, market-impact costs, and the bid-ask spreads are considered costs of buying and selling a stock. Stocks that are cheaper to trade are considered more liquid than stocks that trade more costly. The trading cost for an illiquid stock reduces the total return to an investor. Hence, investors demand a higher expected return when investing in a stock with higher trading costs. This higher expected return implies a higher cost of capital to the firm. Thus an improvement of market liquidity will lower the required rate of return. In other words, firms with low volume prior to the lockup expiration should enjoy greater liquidity effects and as a result should experience a smaller price drop at or around the expiration date.
2.2.2 Magnitude of the Price Reaction at the Lockup Expiration
Downward Sloping Demand Curves
As mentioned above a downward sloping demand curve exists if there is no rebound in the price after the lockup expiration. The standard economic assumption is that the demand for a stock is perfectly elastic (an infinite coefficient of elasticity of demand) at the prevailing price. However, Field and Hanka [2001] suggest that stocks are also subject to downward sloping demand curves like nearly all markets for products. The “downward sloping demand curve hypothesis” predicts that the public’s demand curve slopes downward, in other words a permanent shift in supply will lower the share price. This effect is often named by practitioners “scarcity premium”. The “scarcity premium” should be higher for IPO firms with a smaller free-float.
Ofek and Richardson [2000] also take a look at downward sloping demand curves and argue on why companies that have just gone through IPOs are arguably the most attractive companies in terms of the downward sloping demand curve explanation (p.21): “First, newly-public stocks expand asset payoffs in ways which are not easily duplicated by existing assets. In other words, on average, stocks that have just gone through IPOs do not have clear asset substitutes. If IPOs represent an important part of the economy, yet cannot be duplicated by existing assets, investors will sacrifice the arbitrage price to access these assets.” Another argument in favor of the existence of downward sloping demand curves, according to Ofek and Richardson [2000], is that (p.21) “heterogeneity across investors can lead to downward sloping demand curves. In particular, if economic agents have differences of opinion about asset prices, then in equilibrium the asset price may not be a sufficient statistic for convergence of these opinions.” Finally they argue that also unanimity about firm value maximization might generate downward sloping demand curves for shares.
Commitment Hypothesis
Brav and Gompers [2003] mention that the stock price reaction should be lower for firms that have had good news or are less subject to moral hazard problems, because actions are better observable after the IPO. In line with the arguments used above, firms that experience less information asymmetry are firms that have highly reputable underwriters, are profitable, and are firms with lower price volatility. Also Angenendt et al. [2006] argue that signals of shareholder commitment (reducing uncertainty and revealing quality) may result in smaller price reactions at the lockup expiration day. They also argue that longer lockup lengths, reputable underwriters and venture capitalists backing (as proxies for shareholder commitment) reduce the price drop at the lockup expiration day. Angenendt et al. [2006] also mention that underpricing can be used to signal firm quality at the offering (as a substitute signal), because firms with large underpricing at the public offering are subject to less negative abnormal returns, since underpricing is a costly signal as it leaves “money at the table” for the selling shareholders.
Short Selling Constraint
2.3 Empirical Evidence on the Influence of Firm and Shareholder Characteristics on Lockup Contracts
Table 1 provides an overview of literature that explores the influence of firm and shareholder characteristics on lockup contracts.
Table 1: Overview of Literature on the Influence of Firm and Shareholder Characteristics on Lockup Contracts
The Underwriter and Venture Capital variables are either substitute or complement mechanisms for the dependent variable lockup length.
Author Period Country Sample
General Explanation for
Existence of Lockups Underwriter
Venture Capital Brav and Gompers [2003] 1988-1996 US 2749 Commitment Hypothesis Substitute Substitute Brau et al. [2005] 1988-1998 US 3049 Signalling Hypothesis Substitute Substitute Espenlaub et al. [2001] 1992-1998 UK 52 N/A No relation Complement Bradley et al. [2001] 1988-1997 US 2529 N/A N/A Substitute Goergen et al. [2006]a 1996-2000 GER/FRA 406 Commitment Hypothesis Complement No relation
2.3.1 Information Asymmetry
The Signaling Hypothesis
Brav and Gompers [2003] find no evidence that lockups solve a pre-issue adverse selection problem. According to Brav and Gompers [2003], already mentioned before, firms either signal because they want a higher offer price at the IPO or want to get a higher price in a following seasoned offering. However, they find a contradicting statistically significant negative relationship between lockup length and firms that revise their offer price upwards or between lockup length and follow-on seasoned equity offerings (SEOs). They therefore conclude that the length of the lockup is not used by insiders to signal firm quality.
Brau et al. [2005] find evidence of a positive relationship between lockup length and information asymmetries. In other words, firms with higher uncertainty have longer lockups. Brau et al. [2005] also conclude that shorter lockups are associated with a lower need for signaling. Thus, small and young firms have longer lockups than large and mature firms. Also when looking at industry specifics they find that more transparent and regulated utility companies, but also investment funds generally have shorter lockup periods. Lockup length also decreases when a venture capitalist certifies the firm’s quality.
Goergen et al. [2006] find evidence that lockups can be used as a pre-commitment device, since they find that the higher the fraction of shares sold at the IPO the longer the lockup period. This indicates that the length of the lockup can also serve as a substitute mechanism to signal the commitment of insiders. When dividing insiders between executive and non-executive directors, and using an interaction variable between the fraction of shares sold (free float) and a dummy equal to one if the selling shareholder is an executive, then Goergen et al. [2006] find evidence that suggests that executive directors in Germany and in France are subject to stricter lockups, while certain shareholders come away with less stringent lockups (the free float coefficient is no longer significant after adding interaction variable). On the other hand, Goergen et al. [2006] also argued, as mentioned before, that lockups might serve as a complement mechanism to the percentage of shares retained by insiders after the IPO, where stricter lockups add credibility to the signal. They find, however, weak evidence that supports this reasoning (depending on the test performed). Goergen et al. [2006] also find that underpricing acts as a substitute signal in France, although no significant results are found for firms in Germany. Thus, in contrast to the results in France, the market in Germany doesn’t distinguish a longer lockup as a signal of firm quality as it is not a substitute for underpricing.
The Commitment Hypothesis
book-to-market ratios and higher cash flow margins are negatively related to the lockup length, indicating that less risky, lower growth firms are associated with shorter lockups.
Goergen et al. [2006] find that firms newly listed in Germany and France with less uncertainty about their future value are subject to minimum requirements. In other words, they find a negative relationship between the lockup length and both the age of the firm at the time of the IPO and the market capitalization at the offer price for a sample of IPOs on the German Neuer Markt and the French Nouveau Marché. They also report that larger and older firms in France would rather like to be locked up for 6 months with 100 percent of the shares than being locked up for 1 year with only 80 percent of the shares. Apparently, French firms regard the latter lockup requirement as a stricter option.
2.3.2 The Influence of Underwriters and Venture Capital
Goergen et al. [2006] document a statistically significant positive relationship between reputable underwriters and the length of the lockup period in Germany. This result implies that German underwriters try to protect their reputational capital and force firms into longer lockups. For a sample of 186 UK IPOs issued between 1992 and 1998, Espenlaub et al. [2001] do not find a relationship between the length of the lockup and the quality of the underwriter. Brav et al. [2000] find that highly regarded investment banks impose stricter lockups if the likelihood of moral hazard problems in the aftermarket is higher. They also find that early release is positively related to underwriter reputation. They conclude that this is evidence of the commitment hypothesis since investment banks will only allow insiders of firms that have greatly reduced moral hazard risk to sell their shares before the lockup expiration. Brau et al. [2005] find a statistically significant negative relationship between lockup length and the prestigiousness of the underwriter.
Brav et al. [2000] find empirical evidence that adverse selection problems are less severe in firms with venture capital backing. They find that for a U.S. sample between 1988 and 1996 that venture capital backed firms are subject to 191 lockup days while firms that have no venture capital backing are subject to 264 days. They also report that the fraction of shares that are subject to a lockup is larger for venture capital backed firms than for others.
Bradley et al. [2001] find that 1,160 venture backed firms have a significantly lower average lockup period (192 days) than 1,511 non-venture backed firms with an average lockup period of 249 days.
Brau et al. [2005] find a statistically significant negative relation between the length of the lockup and firms that are venture capital backed. It appears that adverse selection problems are less severe in U.S. venture capital backed firms and for this reason the length of the lockup period can be smaller.
2.4 Empirical Evidence on Price Reactions at Lockup Expirations
The IPO lockup expiration has received much attention after the previous bull market. Several studies on IPO lockup expiration have focused on the lockup expiration effects on price and trading volume. Many studies on lockups, especially in the U.S., report negative significant abnormal returns around the expiry of lockups. Although, especially in Europe, several studies document an average abnormal return equal to zero. Table 2 shows an overview of literature on price reactions around the lockup expiration date.
Table 2: Overview of literature on lockup expiration date
***, ** and * stand for statistical significance at the 1%, 5% and 10% level, respectively.
Author Period Country Sample
Abnormal Return Expiration Day Event Window CAR CAR Increased Trading Volume Field and Hanka [2001] 1988-1997 US 1948 -0.90% [-1,1] -1.50% *** Yes Brav and Gompers [2003] 1988-1996 US 2749 -0.12% [-1,1] -0.79% Yes Bradley et al. [2001] 1988-1997 US 2529 -0.74% [-2,2] -1.61% *** Yes Brau et al. [2004] 1988-1998 US 3049 -0.38% [-4,0] -1.53% Yes Ofek and Richardson [2000] 1996-1998 US 1053 -1.15% [-4,0] -2.03% *** Yes Ofek and Richardson [2003] 1998-2000 US 305 -1.99% [-4,0] -4.11% *** Yes Espenlaub et al. [2001] 1992-1998 UK 52 -0.71% [0,1] -0.96% -Bertoni et al. [2002] 1999-2001 ITA 45 -1.40% [-5,-1] -1.42% * Yes Nowak and Gropp [2000] 1997-1999 GER 142 -0.19% [-1,30] -7.95% *** Yes Angenendt et al. [2005] 1996-2005 FRA 147 N/A [-5,5] -1.33% * Yes Goergen et al. [2006]a 1996-2000 GER/FRA 406 N/A N/A N/A N/A a) Goergen et al. [2006] find no evidence of abnormal returns.
They further state that the abnormal return is not the result of a few extreme observations. Sixty-three percent of the firms experience a negative abnormal return, while the median return is more negative than the mean. The anomalous abnormal return is robust to various sample specifications and benchmarks, while the abnormal return tends to increase in the later years of the sample when controlling for cross-sectional correlation of errors. Also the difference in the abnormal return for venture-capital-backed firms and non-venture-capital-backed firms tends to increase in the later years of the sample.
Ofek and Richardson [2000] find, while using a sample of 1.053 firms in the US, a significant (1%) abnormal return on the expiration date of -1.15 percent. Also a statistically significant, at less than 1 percent, five day abnormal return of -2.03 percent is found between 1996 and 1998. They also find a large jump in volume at the expiration date, with a mean that is 61 percent larger than the time period before the expiration, while a permanent shift in volume of approximately 38 percent is found. Interestingly, they also find that market participants made the same systematic errors each year, because they do not find evidence of a variation in price from year to year. They conclude that the lockup phenomenon is a consistent anomaly through time.
Brav and Gompers [2003] find a non-significant abnormal three day return of -0.79 percent, while a -0.12 percent abnormal day return is reported at the end of the lockup period. They also report a buy-and-hold abnormal return, while this abnormal return peaks at -2.14 percent (day 8) and is significant. With regard to abnormal volume, they find that volume until day -3 is insignificant, from day -3 until -1 marginally significant, but abnormal volume from day 1 until day 10 is positive and significant, while the volume from day 10 and beyond has enduringly changed to a higher level.
Bradley et al. [2001] also find that lockup expirations are accompanied with statistically significant abnormal price declines. The negative abnormal returns are mostly caused by venture capitalists, losing approximately 3 to 4 percent, while accounting for 45 percent of their sample. They also divide the sample in venture capital backed and non-venture capital backed firms and examine the influence of market capitalization, pre-expiration day stock price performance, underwriter reputation, stock price volatility, the fraction of shares subject to lockup, trading volume and other variables. No results are found for the non-venture capital backed sample, while the largest drops in value in the venture capital backed sample are experienced by firms with big pre-expiration day stock price performance, high trading volume and high volatility.
-1.6 percent and are statistically significant at the 1 percent level, while the non-venture capital backed returns range from -0.2 to -0.8 percent.
Angenendt et al. [2006] find marginally significant negative abnormal returns around the event date, however the cumulative abnormal returns for insiders is strongly statistically significant and negative.
Nowak and Gropp [2000] also find negative abnormal returns around the lockup expiry in Germany. They also find that the abnormal returns are more negative for firms with high stock price volatility, better pre-expiration day stock price performance, and for firms with a higher percentage of shares subject to a lockup.
Goergen et al. [2006] find no significant abnormal returns at the expiration date for Germany and France, regardless of the category of lockup contract and the type of insider locked up.
2.4.1 Reasons for Observed Price Reactions at the Lockup Expiration
Temporary Price Pressure
Field and Hanka [2001] reject the temporarily price pressure hypothesis because the abnormal return around the lockup expiration is not temporarily, but permanently negative with little or no recovering in the following days or weeks. They also state that the drop in value does not represent an arbitrage opportunity since the bid and ask prices do not fall enough to reward the strategy of selling short at the bid price before the unlock day and then covering at the ask price after the lockup expiration day. Brav and Gompers [2003] find evidence that rejects the temporarily price pressure idea, while Ofek and Richardson [2000] also find that the stock price does not recover after the days following the lockup expiration.
Worse-than-Expected Insider Sales
Field and Hanka [2001] find among 58 firms that report insider sales near the expiration day a mean three-day abnormal return of -4.5 percent, while they find among 276 firms that do not report insider sales a mean abnormal return of -2.5 percent. This result shows evidence that the negative return may be caused by worse-than-expected insider sales, nonetheless, the negative abnormal return also remains negatively when no insider sales are reported and therefore the abnormal return cannot be solely driven by worse-than-expected insider sales.
Liquidity Effect
Ofek and Richardson [2000] find a large jump in volume at the expiration date, with a mean that is 61 percent larger than the time period before the expiration, while a permanent shift in volume of approximately 38 percent is found. The significant increase in volume around the lockup expiration should cause the stock price to rise instead of fall at the expiration date. When dividing the firms between low and high volume, Ofek and Richardson [2000] find no smaller price drop for higher dollar volume stocks, which is not in line with the liquidity hypothesis, because firms with low volume prior to the expiration date lockup should experience more of a liquidity effect, and thus a smaller price drop, at the expiration of the lockup. Therefore they reject the liquidity hypothesis.
2.4.2 Magnitude of the Price Reaction at the Lockup Expiration
Downward Sloping Demand Curves
Downward sloping demand curves can be detected when looking at the relationship between abnormal return and the total three-day trading volume, measured as the fraction of the shares that could trade before the expiration date. Field and Hanka [2001] find significantly more negative abnormal returns when a larger fraction of the outstanding shares are locked up. However they also find significant abnormal returns when trading volume is less than one percent of pre-expiration float or below the pre-unlock mean trading volume. This implies that the abnormal return cannot be entirely explained by downward sloping demand curves. Brav and Gompers [2003] find evidence that is relatively consistent with downward sloping demand curves.
Ofek and Richardson [2000] also take a look at downward sloping demand curves and find evidence that generally support the hypothesis of downward sloping demand curves, although a non-significant result is obtained for the variable volume around market closures (used as a proxy for downward sloping demand curves). Their results show that an one standard deviation increase of the standard deviation of analysts earnings forecasts, as a measure of divergent opinions (reason for downward sloping demand curves), results in a 0.54 percent drop in the stock price.
Commitment Hypothesis
proxy for information asymmetry) increases the price drop at the expiration. On the other hand, they also find evidence that venture capital backing has a negative impact on the abnormal return. Also firms that have a smaller free float are associated with a larger price decline. Overall, they conclude that the results are consistent with the idea that lockups are in place to reduce or overcome information asymmetry.
Angenendt et al. [2006] find no evidence that the duration of the lockup serves as a commitment device. They also find no evidence that underpricing serves as a substitute signal for firm quality and therefore reduces the negative market reaction at the lockup expiration. No results are found on the impact of venture capital backing or voting right schemes on the negative market reactions at or around the expiration day. Also no relationship was found between the abnormal return and stricter lockup contracts, neither between venture capitalists’ reputations. More reputable underwriters will less likely risk a public floating of a low quality firm, indicating a positive relationship between abnormal return and the prestigiousness of the firm’s underwriter, however, firms with more reputable underwriters have more negative abnormal returns at the expiration date, indicating that underwriter reputation serves as a complementary signal of firm quality instead of a substitute signal.
According to Espenlaub et al. [2002], the commitment hypothesis by Brav and Gompers [2003] explains why high-tech firms commonly voluntarily choose pre-specified fixed dates instead of a lockup expiration on a corporate event (like an earnings release). They also explicitly state that the negative returns experienced at the expiration date regarding the expiry of directors’ lockups may be indicative of information asymmetries. Bradley et al. [2001] find results that are in contrast to the commitment hypothesis since they find marginally significant negative result for high quality underwriters. They also find results that are in line with the commitment hypothesis since the coefficient on high-tech firms is negative and significant (at the 1 percent level), indicating that high-tech firms decline significantly more in value at the expiration day. Also in line with the hypothesis is the result that larger firms bear slighter declines in value than smaller firms.
Short Selling Constraint
3
Hypothesis Development
The IPO market represents a classic example of information asymmetry due to the superior knowledge of insiders relative to outside shareholders about fundamentals and the potential growth of a firm.
Courteau [1995] finds evidence that the length of the lockup period serves as a signaling mechanism that complements ownership retention. In other words, stricter lockups give credibility to the signal of ownership retention. Credibility, since the signal is costly for insiders who are bearing asset risk by holding undiversified portfolios. As a result, the lockup agreement serves as a certain legal device enabling insiders to pre-commit to keeping a high stake after the offering. However, Goergen et al. [2006] argue that firms that issue a bigger part of their equity at the IPO have longer lockups. I come to the following hypothesis in line with Courteau [1995]:
Hypothesis 1:
H0: There is no relationship between ownership retention and the length of the lockup period
H1: Firms with higher ownership retention use stricter lockups to add credibility to the ownership
retention signal
Brav and Gompers [2003] argue that a lockup can serve as a commitment device, because committing the insiders to keeping their shares over a certain period of time after the IPO makes it more likely that any private information becomes public knowledge. In other words, companies whose moral hazard incentives in the aftermarket are likely to be large will have to accept a longer lockup, if they want to persuade outsiders to buy stock in the offering. Goergen et al. [2006] find that firms with more uncertainty about their value are subject to stricter lockups. This leads to the following hypothesis:
Hypothesis 2:
H0: There is no relationship between firms that are subject to more uncertainty about their value
and the length of the lockup periods
H1: Firms that are subject to more uncertainty about their value have stricter lockup agreements
argue that underpricing can be used as a substitute signal for lockup length and the fraction of shares locked up. Brav and Gompers [2003] find evidence that firms can signal firm quality through underpricing. I formulate the following hypothesis:
Hypothesis 3:
H0: There is no relationship between IPO underpricing and the length of the lockup period
H1: Firms that use more underpricing at the IPO face lenient lockups
Brav and Gompers [2003] find evidence that highly regarded underwriters are associated with higher quality issues and therefore firms that go public with more prestigious investment banks need less rigorous lockup contracts. However, Goergen et al. [2006] also suggest that underwriters write lockup agreements in order to buttress their reputations. Underwriters may lose reputation if unfavorable price movements or adverse information effects occur after the issue. Contrary to less stringent lockups hypothesized above, they argue that a reputable lead underwriter will demand stricter lockup agreements to safeguard its reputation in general. These arguments yield the following competing hypotheses:
Hypothesis 4a:
H0: There is no relationship between reputable underwriters and the length of the lockup period
H1: Lockups are less strict when firms go public with more reputable underwriters
Hypothesis 4b:
H0: There is no relationship between reputable underwriters and the length of the lockup period
H1: Reputable underwriters demand stricter lockups when firms go public to safeguard their
reputation
Hypothesis 5:
H0: There is no relationship between firms that are venture capital backed and the length of the
lockup period
H1: The certification role of venture capitalists decreases the need for stricter lockups
Many studies, especially in the U.S. but also in the UK, Germany and France find (significant) negative abnormal expiry returns at the end of the lockup expirations.
According to the Efficient Market Hypothesis firms should not be subject to an abnormal price reaction at the time of the lockup expiration. The market should correctly estimate and anticipate the release of shares at the lockup expiration, since the parameters of lockups are well specified in prospectuses. This leads me to the following hypothesis:
Hypothesis 6:
H0: At the lockup expiry the average abnormal return is equal to zero
H1: At the lockup expiry the average abnormal return is not equal to zero
While Field and Hanka [2001] and Ofek and Richardson [2000] focus mainly on downward sloping demand curves but also bid-ask spreads and short selling constraints respectively, this study focuses partly, like Brav and Gompers [2003] on the potential level of asymmetric information related to firm value. The commitment story predicts lower abnormal returns for firms that have had good news or are less subject to moral hazard.
Courteau [1995] argues that the length of the lockup period serves as a signaling mechanism that complements ownership retention. The signal is credible, since the signal is costly for insiders who are bearing asset risk by holding undiversified portfolios for a longer period of time. Therefore I come to the following hypothesis:
Hypothesis 7:
H0: There is no relationship between lockup length and the abnormal return at the lockup
expiration
H1: IPO Firms with longer lockups, complementing the ownership retention signal, experience
Firms can also signal their commitment by IPO underpricing. The signal of shareholder commitment is credible since underpricing leaves “money on the table” for issuers. If insiders don’t want to lockup for a longer period of time, then they need to signal firm quality by underpricing more at the flotation. This leads to the following hypothesis:
Hypothesis 8:
H0: There is no relationship between underpricing and the abnormal return at the lockup
expiration
H1: IPO Firms that underprice more experience less negative abnormal returns at the lockup
expiration
Angenendt et al. [2006] argue that highly regarded underwriters are associated with higher quality issues (lower uncertainty) and therefore firms that go public with more prestigious investment banks have less negative abnormal returns around the lockup expiry. Bradley et al. [2001] find evidence that firms that are underwritten by the top 3 underwriters have less negative abnormal returns. I formulate the following hypothesis:
Hypothesis 9:
H0: There is no relationship between underwriter reputation and the abnormal return at the lockup
expiration
H1: IPO Firms with more reputable underwriters experience less negative abnormal returns at the
lockup expiration
Jain and Kini [2000] argue that venture capital backed firms are less subject to uncertainty about their value or moral hazard problems, since venture capitalist not only exchange capital for an equity stake, but also provide monitoring and screening and are especially involved in the decision making process. Therefore I propose the following hypothesis:
Hypothesis 10:
H0: There is no relationship between firms that are venture capital backed and the abnormal
return at the lockup expiration
H1: IPO Firms with venture capital backing experience less negative abnormal returns at the
4
Data and Methodology
This study employs an initial sample of 77 Initial Public Offerings floated on the Amsterdam Stock Exchange from the 1st of June 19943 until the 31st of December 2005.A list of all the IPOs over the period 1994-2005 was obtained from the Listing and Primary Market Department of Euronext Amsterdam. The Thomson Research Database was used to collect all the prospectuses. I exclude IPOs by mutual funds, real estate investment trusts and financial institutions. The prospectuses that were not obtainable from the Thomson Research Database, were obtained by contacting the respective firm investor relations’ manager. After eliminating the non-obtainable prospectuses the final sample consists of 71 firms that have gone public on the Amsterdam Stock exchange.
Information regarding the IPO characteristics, for example the lockup length and the number of shares locked up, were obtained from the prospectuses. In case only a price range is offered in the prospectus, Het Financieele Dagblad was consulted to obtain the accurate offer price. Price data and volume data are obtained from Thomson Datastream. Accounting data and SIC Codes were extracted from Thomson One Banker, an online financial database application of the Thomson Financial Group that extracts its data from industry-leading sources such as First Call, Worldscope and Thomson SDC. Thomson One Banker is also used to obtain data on the IPO underwriters, while missing data on underwriters was obtained from offering prospectuses. Venture capital backed IPOs are identified through consulting the offering prospectuses. An IPO is regarded venture capital backed if the issuing firm had a venture capitalist shareholder or director at the time of the public offering. Venture capital backed firms were identified through consulting the Peat Marwick Netherlands Venture Capital Guide.
In this study a regression model relates the lockup length to several independent variables. The independent variables, besides the retained shares variable (by insiders after the offering) are: a dummy variable for venture capital backed firms, an underwriter reputation dummy, a technology dummy, an operating cash flow variable, a firm size variable and an underpricing variable. This study will also explore the price reaction at the time of the lockup expiration through event study methodology that was first described by Brown and Warner [1980] and in a more recent paper further demonstrated by MacKinlay [1997]. Next a regression model will relate the
abnormal return to several independent variables to provide a more detailed analysis on the firm characteristics that cause the potential abnormal returns around the IPO lockup expiration date.
4.1 Lockup Length using a Regression Analysis
This study uses Ordinary Least Squares (OLS) regression models of the following form:
y
t= á + â
1x
t+ â
2x
2t+ â
3x
3t+ â
4x
4t+ â
5x
5t+ â
6x
6t+ â
7x
7t+ å
t (1)Where yi is the length of the lockup, x1 the percentage of shares retained after the IPO, x2 a
venture capital backed dummy, x3 an underwriter reputation dummy,x4 the level of underpricing,
x5 firm size,x6 the operating cash flow margin and x7 a technology dummy.
Dependent Variable Lockup Length
The length of the lockup period is the logarithm of the lockup length in days. The minimum lockup length in the Netherlands is 180 days and for new economy firms sometimes 360 days.
Independent Variables
The Percentage of Shares Retained after the IPO
The percentage of shares retained by insiders after the IPO or the opposite of free-float gives us an interpretation of the total shares that are not sold in the offering and shows us the overall ownership retention after the going public process.
Venture Capital Backed Firm Dummy
The venture capital backed firm dummy equals one if a firm is backed by a venture capitalist and zero otherwise. The venture capital backed firms were identified through consulting the Peat Marwick Netherlands Venture Capital Guide.
Underwriter Reputation Dummy
investment bank was the top 2 lead-underwriter in public offerings in the Netherlands then the underwriter reputation dummy takes the value of a one but a zero otherwise.
Underpricing
Underpricing is often used as a measure for short term performance (in this study first day underpricing) for firms that have gone public on a stock exchange. Underpricing is calculated by deducting the offer price from the first day closing price divided by the offer price.
Uncertainty Variables: Size, Operating Cash Flow Margin and Technology
Size, operating cash flow margin and technology are firm transparency measures and can be used as proxies for uncertainty. The operating cash flow margin is calculated as the ratio of operating cash flow to sales. Size or market capitalization is calculated by taking the logarithm of the number of shares offered at the IPO times the offering price. The technology dummy takes a value of 1 if the firm is a technology firm (SIC codes 28, 35, 36, 38, 50 and 73, respectively Biotechnology and Drugs, Computer and Related, Electronics and Communication, Medical Equipment, Computers and Peripherals and finally Software) and a 0 otherwise.
Brau et al. [2005] state (p.525): “Large firms may be more transparent than small firms because large firms typically are followed by more analysts, are more likely to be in the news, have products or services used by more customers, and usually have a longer public history of performance.” Ritter [1998] also argues that firm size is a proxy for asymmetric information and affects firm value. Technology firms are often associated with a higher level of uncertainty due to their intangible asset structure.
A firm with either a lower cash-flow margin (more uncertainty) or a smaller market capitalization exhibits higher information asymmetries and has a stricter lockup. Also a firm active in the technology sector has more uncertainty around its value and is therefore more likely to expropriate outside investors and subsequently has more need for the commitment of a longer lockup.
Multicollinearity
Table 3: Correlation Matrix LOCKUP LENGTH VC % RETAINED SHARES UNDERW RITER
UNDER-PRICING OCF TECH SIZE
LOCKUP LENGTH 1.000 -0.109 -0.162 0.106 -0.134 0.247 -0.015 -0.022 VC 1.000 0.015 -0.021 0.025 -0.082 0.048 -0.262 % RETAINED SHARES 1.000 0.145 0.300 -0.066 0.132 0.134 UNDERWRITER 1.000 -0.235 -0.124 -0.061 0.336 UNDERPRICING 1.000 -0.114 0.000 -0.122 OCF 1.000 0.152 0.054 TECH 1.000 -0.190 SIZE 1.000
Since, I also want to comprehend how the various independent variables influence the dependent variable, multicollinearity in the OLS regression would be a problem. No signs of multicollinearity were found and there is no need to compute the variance inflation factors (VIF) to measure the severity of any multicollinearity among the independent variables.
Heteroscedasticity
The standard error estimates could be wrong, if the errors are heteroscedastic, but are assumed to be homoscedastic (var (åt) = ó
2
). In order to detect heteroscedasticity a White’s test (no cross terms) is executed. No significant results were found for any signs of heteroscedasticity. This study concludes that the variance of the errors is constant or homoscedastic.
4.2 Event Study
Event study methodology will be used, in order to test for abnormal returns at the lockup expiration date. The helpfulness of event studies arises from the fact that the extent of the abnormal performance at the time of the lockup expiration provides a measure of the (unanticipated) impact of the expiration on the residual claim to equity holders. Or in other words, the event study examines the impact of an event (newly disseminated information) on the stock returns of a firm.
This study uses the daily Return Index (RI) for the offering firms from Thomson Datastream. The RI corrects the PI (Price Index) for additional returns of a stock like dividends. For the market index this study uses the Return Index (RI) for the Amsterdam All Share Index, which is a weighted index of all shares traded at the Amsterdam Stock Exchange (by market capitalization).
The Time Frame
The event day (day 0) represents the lockup expiration date. The initial event window starts the day before the lockup expiration date and ends the day after the lockup expiration date. This event window is based on previous papers by Brav and Gompers [2003] and Field and Hanka [2001]. MacKinlay [1997] argues that by taking an event window you can examine the periods surrounding the event and correct, for instance, for information leakages. The estimation window, the control period prior the event, is used to estimate the parameters of the normal performance model and is subsequently used to calculate the abnormal stock returns over the event window. The estimation window is a 73 day period ending 11 days before the event day. This relatively short estimation window (MacKinlay [1997] uses a 120 day estimation window) is a trade-off, since a longer window is desirable for more precise estimates, but the length of the estimation period is restricted by comparatively short lockups. In addition, the five trading days after the offering are also excluded from the estimation window to account for the underwriter stabilization period.
Models
The general function to calculate the abnormal return is the following [MacKinley, 1997]:
it it
it
K
R
(2)Rit is the actual return on time t for firm i, Kit the normal return on time t for firm i and åit is the
abnormal return on time t for the same firm i. Rewriting equation (2) gives:
it it
it
R
K
Market Adjusted Return Model4
The market adjusted return model makes use of the fluctuations of the general market where the stock is listed. The market adjusted model relates the return of any given stock to the return of the market portfolio. It adjusts the daily return index of a firm for the daily return index of the total Dutch market return (thus beta equal to one).
it mt
it
R
R
(4)Rit is the actual return on time t for the firm i; åit is the abnormal return on time t for firm i. When
rewriting equation (5) you find the following formula for the abnormal returns (ARit):
mt it
it
R
R
AR
(5)Before we can test the model the Average Abnormal Returns (AAR) and Cumulative Average Abnormal Returns (CAAR) need to be calculated for the two models separately (also variance). The methodology described in MacKinlay [1997] is followed:
N i it i AR N AAR 1 1 (6)
N i t i N AAR 1 2 2 1 ) var(
(7)If a reaction took place caused by the event, then the AARi should be significantly different from
zero. Next CAAR (t1, t2) can be calculated:
2 1 ) , (1 2 t t t t AAR t t CAAR (8)
2 1 ) var( )) , ( var( 1 2 t t t t AAR t t CAAR (9) DistributionA robustness test is required to demonstrate that the observed values of the abnormal returns are statistically different from zero. In the case of normally distributed data a Student t-test can be applied and in the case of a non-normal distribution a non-parametrical test has to be applied.
