MSc Finance Master thesis
Lock-up Agreement and Aftermarket Performance of PE-backed IPOs in Europe
An empirical research of European Union IPOs between 2000 and 2013
Abstract
This thesis investigates the relationship between lock-up agreements and long-term aftermarket performance, and the difference in lock-up length within this relationship, for private-equity (PE)-backed initial public offerings (IPOs). More specifically, it studies the effect of lock-up length on post-IPO performance and abnormal returns around the lock-up expiration date, using the data of PE-backed IPOs listed in the European-Union (EU) markets between January 2000 and December 2013. The long-term performance is calculated for the 1-, 2-, and 3-year holding periods based on Fama and French’s (2016) five-factor model. The results find significant negative abnormal returns after one and two years following the IPO. Also, the results find statistical evidence of negative abnormal returns around the lock-up expiration date, and statistical evidence of a positive influence of the lock-up period on the 3-year, long-term aftermarket performance.
Key words: Initial public offerings; private equity; underpricing; long-term performance; lock-up agreement; lock-up length
JEL classification: G12, G14, G24, G34
Caroline Wendelmoet Christine Prince 1091RP, Amsterdam
Caroline.Prince@hotmail.com Swammerdamstraat 21-3
s2201380 July 2017
Contents
1. Introduction ... 3
2. Literature Review ... 6
2.2. Initial public offerings ... 6
2.3. Initial-public-offering underpricing ... 7
2.4. Initial-public-offering aftermarket underperformance ... 8
2.5. The IPO lock-up agreement ... 12
2.6. The IPO lock-up agreement and abnormal returns on the lock-up expiration date .... 13
2.7. The IPO lock-up agreement and aftermarket performance ... 14
3. Methodology ... 16
3.1. Fama and French’s five-factor model ... 16
3.2. Event study ... 17
3.3. Ordinary least-squares regression ... 19
3.3.1. Explanatory variable and control variables ... 20
3.4. Data collection ... 22
4. Data and descriptive statistics ... 24
4.1. Descriptive statistics ... 24
5. Results ... 30
5.1. Fama and French’s five-factor model ... 30
5.2. Event study ... 31
5.3. Regression ... 32
6. Conclusion and Limitations ... 34
7. References ... 37
1. Introduction
An IPO is the sale of a firm’s equity shares to investors on a public stock exchange for the first time. There are several reasons firms decide to go public: one reason to create an IPO is to raise capital, and a second reason is to create a way for the founders and early stakeholders to convert their share of the firm into cash at a future date (Ritter & Welch, 2002).
The aftermarket performance of IPOs has received much attention in the literature, and can be measured over short and long terms. Short-term performance is measured over a period of a few days to a month after the IPO, while long-term performance is measured over a period of more than a year after the IPO (Ritter, 1991).
Initial public offerings are underpriced in the short term (Logue, 1973; Ibbotson, 1975; Ritter, 1991), where underpricing is usually defined as the difference between the price at which the shares are offered to the market and the price at which the shares are traded at the end of the first day of trading. In the long term, IPOs tend to show underperformance (Ibbotson, 1975; Ritter, 1984; Ritter, 1991; Ibbotson & Ritter, 1995; Loughran & Ritter, 1995). For the UK, a similar pattern for post-aftermarket performance is reported (Levis, 1993; Espenlaub et al., 2000). Long-term underperformance means that the subsequent share prices of IPOs are lower than they were on their first day of trading, which results in lower returns in the long run for investors holding IPO stocks. Most overvalued IPOs run up the first day and revert to their fundamental value in the long run (Purnanandam & Swaminathan, 2004).
Private-equity-backed IPOs exhibit, on average, a lower degree of underpricing in the short term than their non-backed counterparts (Bergström et al., 2006; Barry et al., 1990; Megginson & Weiss, 1991). The evidence of under-pricing and underperformance in the aftermarket for PE-backed IPOs and other non-backed IPOs is consistent with the mispricing view of IPO underperformance, where investors who buy the IPOs initially tend to be optimistic about their future performance (Ritter, 1991; Loughran & Ritter, 1995; Miller, 1977). According to Rock (1986), firms intentionally underprice IPOs to induce uninformed investors to participate in the market, thereby increasing demand for the shares. Moreover, high-value firms underprice their shares because of the large demand for the stock by well-informed investors.
no evidence of market underperformance (Degeorge & Zeckhauser, 1993; Holthausen & Larcker, 1996; Mian & Rosenfeld, 1993, 1996; Cao & Lerner, 2009). One explanation for this performance is the PE sponsor’s involvement, which does not completely end after the PE-backed IPO (Levis, 2011).
Due to lock-up agreements, PE sponsors retain holdings in the firm for a period of time. This continuing involvement enables closer monitoring and a lower degree of both information asymmetries and potential conflicts with other stakeholders (Levis, 2011), leading to better operating performance compared to their non-backed counterparts. Furthermore, PE-backed IPOs tend to be less underpriced and have superior aftermarket performance over the long term. Studies on lock-up agreement find that when lock-ups expire, average trading volumes increase and a significant negative abnormal return occurs (Ofek & Richardson, 2000; Field & Hanka, 2001). No abnormal price reaction should occur at the lock-up expiration date, though, because the length of the lock-up period is known prior to the public offering. Even though the demand curves for stocks slope downward, investors should be able to correctly forecast the average number of shares that insiders will sell at the lock-up expiration date (Brav & Gompers, 2003). Consequently, the average abnormal return around expiration should be zero (Scholes, 1972).
The main objective of this thesis is to assess the influence of a lock-up agreement’s length on the aftermarket performance of PE-backed IPOs in the EU. Field and Hanka (2001) have studied the effect of lock-up agreements on the entire American IPO market. However, this thesis tests only the effects of lock-up periods surrounding European PE-backed IPOs. Thus, this thesis emphasises interest in the PE-backed IPO aftermarket performance of countries other than the US.
The main research question of this thesis is as follows:
What is the effect of the lock-up period on abnormal returns in the long-term aftermarket performance and around the lock-up expiration date of PE-backed IPOs in Europe?
To guide the research to an answer to this question, the following three hypotheses were formulated:
Hypothesis 1: Private-equity-backed firms have negative long-term stock-price performance in the post-IPO period.
Hypothesis 2: For PE-backed IPOs, there is a negative abnormal return around the expiry date of the lock-up period.
Hypothesis 3: Shorter lock-up periods are associated with better long-term aftermarket performance for PE-backed IPOs.
For the first hypothesis, Fama and French’s (2016) European five-factor model is used to test aftermarket performance at the 1-, 2-, and 3-year marks for holding periods. Second, an event study methodology was used to analyse the impacts of the lock-up agreement on the abnormal returns around the expiration date. Thirdly, this thesis uses the ordinary least-square (OLS) regression analysis to test the effect of lock-up length on long-term aftermarket performance.
This study’s use of an EU sample contributes to the existing European literature on the long-term performance of PE-backed IPOs (Levis, 1993; Schultz, 2003; Jaskiewicz et al., 2005; Espenlaub et al., 2000), as confirmed by Levis (2011), who finds that “the aftermarket performance of PE-backed IPOs outside the US is sparse and inconclusive” (p257).
exhibit superior long-term aftermarket performance compared to their non-backed counterparts.
Additionally, this thesis is—to the best of my knowledge—the first that specifically focuses on the lock-up agreements surrounding PE-backed IPOs in Europe between 2000 and 2013.
This study of 139 PE-backed IPOs in the EU between 2000 and 2013 provides statistical evidence of underperformance of the PE-backed IPOs in the first three years following the IPO. The results obtained from the event study find negative abnormal returns around the lock-up expiration date. Also, this thesis provides a positive statistically significant relationship between the lock-up length and the long-term aftermarket performance after a 3-year holding period following the IPO. However, this effect is not statistically significant in the first two years following the IPO.
The remainder of this thesis is structured as follows: Chapter 2 provides a literature review analysing the impact of the lock-up agreement around the lock-up expiration date. In addition, the literature review analyses the influence of the length of the lock-up agreement on the aftermarket performance for IPOs. Chapter 3 describes the methodology, and the data and descriptive statistics are shown in Chapter 4. Chapter 5 details the results, and Chapter 6 concludes and discusses this study’s limitations and suggestions for further research.
2. Literature Review
2.2. Initial public offerings
Several benefits arise from the decision to go public. One reason to do so is to raise new capital instead of taking on excess debt. Secondly, an opportunity arises for current shareholders to cash out or exit completely. Thirdly, the firm’s stock price is a good indicator of performance. This is an advantage both for investors, as the stock prices can signal whether the firm is a good investment, and for the firm’s managers, since they can be rewarded with respect to stock performance. Fourth, the firm can reduce its borrowing costs and diversify its sources of finance because its information is public and widely available (Brealey et al., 2012).
The capital structure of leveraged buyouts (LBOs) is a vital part of the PE business, and is superior to any existing capital structure (Jensen, 1989). One of the characteristics of PE finance is that investors hold their investments for only a limited period of time—usually less than 10 years (Gompers & Lerner, 1999). After the investment horizon, PE firms use several exit channels, for example, the sale of the firm within the portfolio to an outside strategic buyer, a secondary buyout, or an IPO (Baker et al., 2015). For PE firms, the IPO route is an important divestment strategy (Muscarella & Vetsuypens, 1990; Kaplan, 1991).
2.3. Initial-public-offering underpricing
There is widespread evidence for the underpricing phenomenon of IPOs in the short term (Logue, 1973; Ibbotson, 1975; Ritter, 1991). The initial underpricing of IPOs leads to over-performance in the long-term aftermarket (Purnanandam & Swaminathan, 2004). Table 1 provides a summary of the literature on the underpricing of IPOs.
Table 1: Summary of the literature on IPO underpricing
Author Country Sample Period Number of IPOs Methodology Main Results Ibbotson (1975) US 1960–1969 2,504 Risk-adjusted
returns Average positive initial performance of 11%. Ritter and Welch
(2000)
US 1980–2001 6,249 Equally weighted BHAR
Evidence of an average first-day return of 18%.
Ritter (1991) US 1975–1984 1,526 Event-time CAR,
BHAR Average initial return of 16%. Moreover, the extents of this underpricing are highly cyclical, and, in the long term, IPOs are overpriced.
Logue (1973) US 1965–1969 250 Equally
weighted BHAR Evidence of underpricing. Levis (1990) UK 1985–1988 123 Capital-asset pricing model (CAPM) abnormal returns. Evidence of an average
underpricing of 8%. However, the average underpricing decreases to 5% when the application becomes smaller, conditional on being allocated stock. Amihud et al., (2003) Israel 1989–1993 284 Equally weighted initial returns
Uninformed investors earned, on average, a negative allocation-weighted initial return of 1%.
investors bid on underpriced IPOs, and uninformed investors bid on relatively overpriced IPOs. Hence, IPO companies use underpricing to compensate for the losses of uninformed investors, since underpricing prevents uninformed investors from withdrawing from the markets. In addition, Benveniste and Spindt (1989) find that underpricing compensates better-informed investors for truthfully revealing their information before the issue price is settled, and therefore reducing the expected amount of money left on the table.
Another theory related to asymmetric information is the signalling theory (Allen & Faulhaber, 1989). A firm can signal its quality through underpricing. High-quality firms are able to underprice their IPOs, since they are confident that they can earn back the underpricing losses through their future performance. Low-quality firms need to exert more effort to earn back their losses; therefore, signalling is costlier for low quality firms. Private-equity-backed IPOs are classified as high-quality firms, given the degree of support in terms of capital, industry experience, and active and professional management (Bergström, 2006).
Private-equity-backed IPOs exhibit, on average, lower degrees of underpricing in the short term than their non-backed counterparts (Barry et al., 1990; Megginson & Weiss, 1991; Bergström et al., 2006; Levis, 2011). Private equity firms are subject to more information disclosure and scrutiny, which contributes to greater informational homogeneity amongst investors, and reduces ex ante uncertainty about the value of the issuing firm. This, in turn, alleviates the adverse selection problem and need for underpricing (Bergström et al., 2006).
2.4. Initial-public-offering aftermarket underperformance
Table 2: Summary of the literature on IPO long-term aftermarket performance
Author Country Sample Period Number of IPOs Methodology Main Results Ritter (1991) US 1975–1984 1,526 Event-time CAR,
3-year BHAR
Average holding of 34% of IPOs underperform in the three years after going public.
Loughran and
Ritter (1995) US 1970–1990 4,753 Event-time BHAR Negative 5-year abnormal returns of 7% per year for IPOs and seasoned equity-issuing firms. Ritter and
Welch (2002)
US 1980–2001 6,249 Event-time BHAR
Evidence of 23% 3-year negative abnormal returns.
Gompers and Lerner (2003)
US 1935–1972 3,661 Event-time BHAR and CAR
Significant IPO
underperformance of 19% when the BHAR is used. However, this underperformance disappears when the CAR is used (2%). Aggarwal and
Rivoli (1990) US 1970–1990 4,753 Event-time BHAR Evidence of 13% 2-year negative abnormal returns for buying shares at the closing price and offer price. Schultz (2003) Continental Europe and Sweden 1988–1998 973 Event-time BHAR and matched portfolio approach
No underperformance in the long term. Over shorter measurement horizons, over-performance is found. Jaskiewicz et al. (2005) Spain and Germany 1990–2000 554 Event time BHAR
Three years after the IPO, negative abnormal returns of 32% for Germany and 36% for Spain.
Levis (1993) UK 1980–1988 712 Event-time BHAR and matched portfolio approach
Evidence of 3-year negative abnormal returns of 22%.
Espenlaub et al. (2000)
UK 1985–1992 588 Event-time BHAR
Negative abnormal returns after the first three years (15%). However, over the first five years after the IPO, the abnormal returns diminish (7%).
between pessimistic and opportunistic investors, and therefore the share price will fall (Miller, 1997). However, in an efficient market, if prices get out of line, then arbitrage forces them back. The arbitrageur buys the underpriced securities (pushing up their prices) and sells the overpriced securities (pushing down their prices). The arbitrageur earns a profit by buying low and selling high, and waiting for prices to converge (Purnanandam & Swaminathan, 2004).
While there is widespread evidence of IPOs’ long-term aftermarket underperformance, studies find that PE-backed IPOs outperform non-backed IPOs. Table 3 presents a summary of the findings on the long-term aftermarket performance for PE-backed IPOs.
Table 3: Summary of the literature on PE-backed IPOs’ long-term aftermarket performance
Author Country Sample Period Number of IPOs Methodology Main Results
Private-equity-backed firms exhibit significant superior long-term stock-price performance. Evidence of 1-year abnormal return of 19% versus 9% for non-backed counterparts. Katz (2009) US 1980–2005 147 Event-time BHAR and
matched portfolio approach Holthausen and Larcker (1996) US 1983–1988 90 Accounting ratios to measure performance, operating income, and operating cash flows
No evidence of negative excess returns of RLBOs after the IPO, compared to the underperformance of non-sponsored IPOs.
Degeorge and
Zeckhauser (1993) US 1983–1987 62 Event-time CAR, 3-year BHAR
They find that over the two years following the IPO, RLBOs' stocks outperform comparison firms. Mian and Rosenfeld
(1999)
US 1983–1988 85 Event time average abnormal returns and CAR
Study on RLBOs. The 3-year performance significantly outpaces comparable firms with the same four-digit industry code.
Brav and Gompers (1997)
US 1972–1992 3,407 Fama-French’s (1993) three-factor asset-pricing model
Venture Capital-backed IPOs outperformed non-VC-backed offerings in the five years following the offer.
Levis (2011) UK 1992–2005 554 Event-time BHAR
Private-equity-backed IPOs show better 3-year aftermarket performance when compared to both venture capital and their non-backed counterparts as well as the market as a whole.
Bergström, Nilsson,
and Wahlberg (2006) UK and France 1994–2004 125 Event-time abnormal returns, CAR
Across different time periods, PE-backed IPOs outperform non-PE-backed IPOs.
Cao and Lerner
(2009) US 1980–2002 526 Wealth-relative performance measure
Evidence exists for the positive impact of PE on IPO aftermarket performance. Firms with majority ownership by PE sponsors experience better long-term financial stock-price performance when they go public (Katz, 2009). Studies on the long-term performance of RLBOs also find better long-term aftermarket performance among RLBOs before floatation, compared to their non-backed counterparts, and even find no evidence of aftermarket underperformance (Degeorge & Zeckhauser, 1993; Mian & Rosenfeld, 1993; Holthausen & Larcker, 1996; Neus & Walz, 2005; Cao & Lerner, 2009). Overall, PE-backed IPOs demonstrate better aftermarket performance when compared to both those supported by venture capital (VC), their non-backed counterpart IPOs, and the market as a whole. Private-equity sponsors want to build a good reputation amongst investors; therefore, they do not want to behave opportunistically in an IPO. Hence, this non-opportunistic behaviour would imply lower first-day returns and thus, better long-term aftermarket performance for sponsor-backed IPOs, compared to their unsponsored counterparts (Cao & Lerner, 2009). Barry et al. (1990) have found that PE firms are active in financing and monitoring their portfolio companies; hence their monitoring skills are associated with lower underpricing at the time of the IPO, and greater operational improvements and value maximisation of the stock price are expected. Although it is expected that these value-enhancing features are in place during the PE-ownership period, it is often reasonable to expect that management and financial practices established during the time under PE ownership will be maintained for at least some time after the exit (Levis, 2011). Besides, according to Levis (2011), p.254: “PE-backed IPOs provide a fertile ground for assessing the impact of PE in general as they offer a wider perspective of the contribution made by the PE sponsor through the full length of their involvement with the firm”.
Considering the above, the first hypothesis is as follows:
2.5. The IPO lock-up agreement
Lock-up agreements are contracts between the underwriter and the pre-initial public offering (pre-IPO) inside block-holders to guarantee outside investors that the stock will not experience substantial insider selling immediately following the IPO.
When a firm goes public, outside investors have little information about the firm, while inside shareholders tend to have more information about the firm’s prospects. One of the reasons for the existence of a lock-up agreement is to consequently offer outside investors protection from being exploited by well-informed insiders (Brav & Gompers, 2003). Another motivation for introducing lock-up agreements is to prevent flooding the market with an oversupply of shares at the time of IPO. In addition, by restricting shares, lock-up agreements assure public investors that inside investors continue to hold a significant interest in the firm (Bartlett, 1999). While firms can decide whether to implement lock-ups or not, 80% to 90% of US PE-backed IPOs have a lock-up length equal to 180 days, and almost no lock-ups are less than 90 days (Brav et al., 2000; Field & Hanka, 2001; Mohan & Chen, 2002; Bradley et al., 2004). The average shares that owners offer in IPOs is not the same in every country. Table 4 presents a summary of both the average share’s lock-up and length of the lock-up period in the US and Europe.
Table 4: Regulations regarding the minimum lock-up percentage in the US and EU
France Germany Italy The Netherlands Belgium UK US The firms could either lock up 100% of the shares for a period of 6 months or lock up 80% of the shares for a period of 1 year.
All the initial shareholders and the firm lock up 100% of the shares for a period of 6 months. Shareholders, directors, managers, and founders lock up 80% of the shares for a period of 1 year. All the shareholders lock up 80% of the shares for a minimum period of 360 days. All managing shareholders lock up 80% of the shares for a period of 1 year. No minimum lock-up period is required. The Security Exchange Commission (SEC) does not impose any lock-up agreements.
Lock-up agreements in the US are primarily voluntary; however, contracts are mostly standardised (Bradley et al., 2004). On the other hand, the continental European markets implement a required minimum period for lock-up agreements.
of the post-expiry insider sellers; however, they account for 30% of the insider sales in value. It can be concluded that VCs sell out more quickly, whereas other initial shareholders may face more stringent lock-ups in order to signal the quality of the firm to the market (Cao et al., 2004).
2.6. The IPO lock-up agreement and abnormal returns on the lock-up expiration date
Recent studies document abnormal stock returns around lock-up expiration (Brav et al., 2000; Ofek & Richardson, 2000; Brav & Gompers, 2003; Field & Hanka, 2001; Bradley et al., 2004). Table 5 provides a summary of the literature on abnormal returns at the expiration date.
Table 5: Summary of the literature on lock-up agreements
Author Country Sample Period Number of IPOs Methodology Main Results Field and Hanka
(2001)
US 1988–1992 1,948 Event-time CAR A 40% increase in trading volume and a significant 3-day abnormal return of -1% around the lock-up expiration. Ofek and
Richardson (2000)
US 1996–1998 1,662 Event-time CAR A 1%–3% drop in the share price, and a 40% increase in volume, when the lock-up expires.
Brav et al. (2000) US 1975–1992 4,526 Event-time BHAR Negative abnormal returns of 1% on the lock-up expiration date.
Brav and Gompers
(2003) US 1988–1996 2,794 Event-time BHAR Abnormal returns around lock-up expiration date. Bradley et al. (2004) US 1988–1997 2,529 Event-time CAR The IPO lock-up expirations
are, on average, associated with significant abnormal price declines, in particular for VC-backed IPOs.
Yung and Zender (2010)
US 1988–2006 4,025 Event-time CAR Cumulative abnormal returns around lock-up expiration. Negative abnormal returns are most prominent at PE-backed IPOs.
Goergen et al.
(2006) German and French
1996–2000 173 Multinomial logit
regressions No significantly abnormal returns at the expiry for the main Continental European market.
Espenlaub et al.
(2003) UK 1992–1998 186 Event-time BHAR and CAR No significant abnormal returns around the lock-up expiration date.
Hanka, 2001; Yung & Zender, 2010). However, studies on the UK and the continental European markets have not found a significant market reaction at the time of expiration (Espenlaub et al., 2003; Goergen et al., 2006).
If the market is efficient, the length of the lock-up agreement is common knowledge to investors at the time of the IPO; therefore, no abnormal returns should occur at expiration (Ofek & Richardson, 2000; Brau et al., 2004; Hoque & Lasfer, 2009). Specifically, before the end of the lock-up period, the price impact should be built into the IPO-traded price. One explanation for the abnormal returns around lock-up expiration date is that the shares are assumed to have a downward-sloping demand curve. When inside shareholders sell their shares, public investors are asked to hold a larger number of shares. The insider sales around the lock-up expiration date shift the equilibrium price to a point where larger quantities of shares are sold, so the share price falls (Field & Hanka, 2001). After the depreciation of the share price, prices return to their fundamental value (Scholes, 1972). Concluding the above, the second hypothesis is as follows:
Hypothesis 2: For PE-backed IPOs, there is a negative abnormal return around the expiry date of the lock-up period.
2.7. The IPO lock-up agreement and aftermarket performance
Several theories have been developed to explain the negative abnormal return around lock-up expiration. The first theory—signalling theory—uses lock-up agreements as a signalling device under conditions of information asymmetry. In the second theory—commitment theory—lock-up agreements are used as a commitment device to reduce moral-hazard challenges.
at the IPO (Welch, 1989). Private-equity sponsors are important providers of external finance to new firms, and provide the necessary capital. Furthermore, the involvement of PE sponsors prior to the IPO signals quality, since they usually monitor the firm and are part of the decision-making process (Barry, 1994; Jain & Kini, 2000). Thus, a longer lock-up period is expected for PE-backed IPOs because they want to signal the quality of the firm, and hence obtain a higher offer price.
However, Field and Hanka (2001) have found that, for a sample of 1,948 US IPOs with lock-ups, the managers, the selling-shareholders in the IPO, and the shareholders owning more than 5% of the equity are almost always locked up, given the high homogeneity of lock-up agreements in the US. They have also found that US firms do not use the length of the lock-up as a signal of their value. Since the continental European markets implement a required minimum period for lock-up agreements, it could be expected that the length of the lock-up agreement does not signal quality in Europe (Goergen et al., 2004). Furthermore, Brau et al. (2004) doubt that lock-up contracts are effective in mitigating the asymmetries of information between insiders and outsiders, given their short length. The average lock-up duration of 180 days will, at most, cover two quarterly-earnings announcements.
Next, according to the commitment theory, the lock-up agreement serves as a commitment device to alleviate moral-hazard challenges (Brav & Gompers, 2003). Managers could act out of self-interest rather than maximising shareholder wealth. Firms with larger moral-hazard incentives in the aftermarket consequently accept longer lock-up periods to convince the public to buy the stock at the IPO. Potential investors are more willing to buy the stocks at the IPO if the ability to take advantage of them is reduced because of longer lock-up periods (Gompers, 2003).
from institutional investors (Megginson & Weiss, 1991). These methods of gaining credibility imply that reputable sponsors should be associated with shorter lock-up agreements (Brav & Gompers, 2003), yielding the third hypothesis:
Hypothesis 3: Shorter lock-up periods are associated with better long-term aftermarket performance for PE-backed IPOs.
3. Methodology
This section discusses the methodologies that were used to test the long-term aftermarket performance of PE-backed IPOs and the performance around expiration of the lock-up as well as the effect of lock-up length on the long-term aftermarket performance. Firstly, to estimate the aftermarket performance of PE-backed IPOs, Fama and French’s five-factor model was used. Secondly, following MacKinlay’s (1997) market model, an event study was used to analyse the impact of lock-up length on the abnormal returns around the lock-up expiration date. Thirdly, this study used an OLS regression analysis to test the effect of lock-up length on long-term abnormal returns of PE-backed IPOs.
3.1. Fama and French’s five-factor model
To estimate the long-term performance of the PE-backed IPOs, abnormal returns were calculated for 1, 2, and 3 years following the IPO. Day zero was defined as the initial return, and the aftermarket period includes the three years following that day, where years are calculated based on trading days. Following the previous literature, a month is defined as 21 trading days, with 252 days in a year, 504 days in 2 years, and 756 days in 3 years (Loughran & Ritter, 1995; Brau et al., 2004). The Fama-French model regresses the yearly returns on stocks on five market factors that the authors identified (Fama and French, 2016). The model’s regression equation is as follows:
𝑅!"− 𝑅!"= 𝛼!+ 𝛽! 𝑅!− 𝑅! + 𝑠!𝑆𝑀𝐵!+ ℎ!𝐻𝑀𝐿!+ 𝑟!𝑅𝑀𝑊!+ 𝑐!𝐶𝑀𝐴!+ 𝜀!" (1) Where Rit is the stock return on date t. Rf, is the risk-free rate1, and Rm is the return on the
value-weight market portfolio. Rm – Rf is the market risk premium, and 𝛽!, 𝑠!, ℎ!, 𝑟!, and 𝑐! are sensitivities or factor loadings. If sensitivities to the portfolio returns in (1) capture all
1 The risk-free rate is based on the one-month Treasury bill rate from Ibbotson Associates.
variations in expected returns, then the intercept 𝛼! is zero for all securities and portfolios i (Fama & French, 2016).
The small-minus-big (SMBt) and high-minus-low (HMLt) factors are the date t returns
of factor-mimicking portfolios designed to capture size and book-to-market characteristics, respectively. The average difference between the returns on robust operating profitability and weak operating probability is represented by the robust-minus-weak (RMWt) factor. Finally,
the conservative-minus-aggressive (CMAt) factor is the difference between the returns on the
diversified portfolios of the low and high investment firms’ stocks, called conservative and aggressive, respectively (Fama & French, 2016). All daily return data were adjusted for dividends and capital appreciations by taking the return-index (RI) values from DataStream.
The approach that was followed is a two-step process. First, a time-series regression was used to estimate both the alpha and factor loadings for the factors for each of the 139 IPOs. This resulted in a set of 139 abnormal returns as the alpha for each fund for the 1-, 2-, and 3-year holding period was determined. A positive value for the alpha indicates that the IPO has outperformed the benchmark, while a negative value indicates that the IPO has underperformed, compared to the benchmark.
To assess whether the abnormal returns are statistically significant, a Student t-test was used. Under the null hypothesis, all the alphas are equal to zero, and under the alternative hypothesis, the alphas are not equal to zero. The hypothesis that the long-term alphas are equal to zero across i IPOs at date t, where t = 1,2,3 (1, 2, and 3 years following the IPO), was tested.
𝐻!: 𝛼!,! = 0 (2)
𝐻!: 𝛼!,! ≠ 0 (3)
3.2. Event study
after the event. The market model parameters are estimated from the pre-event estimation period for days –90 to –10. In the same way, the parameters estimated from the post-event estimation are used for days +10 to +130 after the event period. The pre-event estimation period is set at 90 days because of the minimum lock-up period of 90 days, since no stock returns are public information before the IPO date. The event date is referred to as the lock-up expiration date for the PE-backed IPOs. The timeline around the IPO lock-lock-up expiration date is graphically depicted in Figure 1.
Figure 1: Event Timeline
Pre-estimation window Event period Post-estimation window
T = (–90) T = (–10) T = (0) T = 10 T = (130)
To determine the performance of each firm, the abnormal returns were calculated with the following formula:
𝐴𝑅!,! = 𝑅!,!– 𝐸(𝑅!,!) (4) Where Ri,t is the daily stock return, and E(Ri,t) is the expected return in the absence of the
event.
To calculate the returns (Ri,t), the return index was used, taking the logarithmic returns per
day, as in (5).
𝑅!,! = 𝑙𝑛 !"!,!
!"!,!!! (5)
Then, the market model was used to estimate the expected returns in the absence of the event. The market model is defined as follows:
𝐸 𝑅!,! = 𝛼! + 𝛽!𝑅!"+ 𝜀!,! (6) The day t value-weighted market return is denoted by Rmt. The MSCI Europe Index is used as
When the outcomes of equations (5) and (6) are inserted into the formula in equation (4), the abnormal returns are calculated.
The mean returns in MacKinlay’s (1997) methodology are calculated as the sum of all returns for one event divided by the number days of the event. The CARi,t is the sum of all
abnormal returns across the 10 days around the lock-up expiration date, where day zero is the lock-up expiration date.2
𝐶𝐴𝑅!,! = ! 𝐴𝑅!,!
!!! (7)
The CARs were calculated for the event windows [–1, +1], [–2, +2], [–3, +3], [–4, +4], [–5, +5], [–6, +6], [–7, +7], [–8, +8], [–9, +9], and [–10, +10]. To assess whether these CARs are statistically significant, a Student t-test was applied. Additionally, to test whether the abnormal returns are influenced by an effect prior to or after the event, a second Student t-test was applied. The CARs were calculated for the event windows [–4, 0], [–3, 0], [–2, 0], [–1, 0], [0, +1], [0, +2], [0, +3], [0, +4]. The hypothesis that these CARs are equal to zero across i IPOs at date t was tested.
3.3. Ordinary least-squares regression
To test the third hypothesis, a regression model was constructed to test the effect of the PE-backed IPO’s lock-up period on long-term abnormal returns. Finding the specific effect of the lock-up agreement on aftermarket performance, this model controls for the firm’s age at offering; IPO proceeds; market conditions (IPOs three months prior to IPO); initial returns (underpricing); leverage; sales; and earnings before interest, taxes, depreciation, and amortisation (EBITDA). The 2008 global financial crisis and manufacturing industry were included as dummy variables. Table 6 presents the definition, symbols, and signs of all variables used in this thesis.
An OLS regression analysis was performed on the observed long-term returns against the explanatory variables presented in Table 6. Consistent with Chen et al. (2012), (8) was estimated. 𝐴𝑅! = 𝑎 + 𝛽!𝐿𝑜𝑐𝑘𝑢𝑝 + 𝛽!𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒 + 𝛽!𝐸𝐵𝐼𝑇𝐷𝐴𝑚𝑎𝑟𝑔𝑖𝑛 + 𝛽!𝑃𝑟𝑜𝑐𝑒𝑒𝑑𝑠 + 𝛽!𝐴𝑠𝑠𝑒𝑡𝑡𝑢𝑟𝑛𝑜𝑣𝑒𝑟 + 𝛽!𝐴𝑔𝑒 + 𝛽!𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔 + 𝛽!𝑀𝑎𝑟𝑘𝑒𝑡𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛 + 𝛽! 𝑇𝑖𝑚𝑖𝑛𝑔 + 𝛽!"𝐷𝑢𝑚𝑚𝑦𝑀𝐴𝑁 + 𝛽!!𝐷𝑢𝑚𝑚𝑦𝐶𝑅𝐼𝑆𝐼𝑆 (8)
Table 6: Variable definitions, symbols, measurement, and signs
Type Variable Symbol Measure Signs
Dependent Abnormal return AR Fama and French’s (2016) five-factor model
generated abnormal returns. Dependent variable Cumulative abnormal
return CAR Cumulative return adjusted by the MSCI Europe Index return. Dependent variable Explanatory Lock-up length Lockup Logarithm of the length of lock-up in calendar
days. (–), (+)
Control Leverage (%) Leverage Total debt divided by total assets at the end of
the fiscal year of the IPO. (–) EBITDA-margin EBITDAmargin Earnings before interest, taxes, depreciation, and
amortisation divided by total sales at the end of
the fiscal year of the IPO. (+) IPO proceeds Proceeds The total size of the proceeds at the time of the
IPO is measured by taking the logarithm number of the shares offered, multiplied by the offer price (€).
(+) Asset turnover (%) Assetturnover Total turnover divided by total sales at the end of
the fiscal year of the IPO. (+) Firm age at offering Age The age of the firm is calculated by taking the
logarithm of the days between the firm’s
founding year and the year of the IPO. (+) Underpricing (%) Underpricing Calculated as the first-day return. Closing price
minus the offer price divided by the closing price of the first day.
(–) Market condition Marketcondition Measured by taking the logarithm of the total
number of IPOs in the EU in the three months
prior to each IPO. (–)
Market timing Timing The logarithm number of days between the IPO
and January 1 of the applicable year of the IPO. (–) Dummy Dummy manufacturing DummyMAN Industry-manufacturing dummy.
Dummy crisis DummyCRISIS A dummy variable included for the 2008 global financial crisis.
3.3.1. Explanatory variable and control variables
shares for longer periods of time in order to signal the quality of the firm to the market. On the other hand, PE-backed and highly reputable underwriters are associated with shorter lock-up agreements (Brav & Gompers, 2003).
A firm’s total debts and liabilities to total assets determine whether financial institutions consider the firm to be healthy, since they either lend money or indicate that the firm’s equity is relatively low and that the firm is considered to be in financial distress. According to Levis (2011), leverage positively influences stock returns (Levis, 2011). However, Cao and Lerner (2009) have found that stock returns are not affected by leverage. Moreover, a negative relationship has also been found (Loughran & Ritter, 1995). Therefore, this study expects a negative relationship between leverage and aftermarket performance.
A firm’s ability to generate profit can be measured by the EBITDA-margin, which reflects the ability of the firm to use its assets to generate revenue. A positive and significant effect of the EBITDA-margin has been found on the aftermarket performance for IPOs (Levis, 2011). Therefore, a positive effect was expected.
Higher proceeds usually indicate a more established firm, lower observed risk, and ultimately a lower underpricing discount. In the long term, larger firms are expected to yield better performance (Ritter, 1991; Loughran & Ritter, 1995; Levis, 2011). Therefore, this study expects a positive relationship between the offer-proceeds amount at the time of the IPO and the long-term performance in the regression.
The asset-turnover ratio measures the efficiency of a firm—how a firm uses its assets to generate operating revenues. A high asset turnover is an indication of the efficiency of the firm, and is therefore expected to have a positive effect on the long-term aftermarket performance (Levis, 2011).
The age of the firm at offering has often been used as a proxy for risk (Ritter, 1991). Older firms are likely to have less information asymmetry due to their longer operational history. Schultz (2003) finds that the probability of failure decreases with increasing age. Demers and Joos (2007) find that younger (less established) firms are likely to fail. Moreover, according to Ritter (1984), younger firms are generally associated with more risk, leading to higher underpricing. Therefore, this study controlled for the age of the firm at the time of the IPO, and expected a positive impact of age on post-IPO survival.
worst in the long-term aftermarket. Therefore, it is expected that underpricing negatively affects long-term aftermarket performance.
The long-term underperformance of public offerings is concentrated amongst years with a high volume of IPOs (Ritter, 1991). If a firm is IPO-ed during a period in which a relatively large number of companies went public—a so-called ‘hot issue period’—it is expected that the IPO price would be relatively high, leading to aftermarket underperformance (Ritter & Welch, 2002). Alternatively, if the firm is IPO-ed during a relatively calm period, a lower offer price is expected, and hence the firm exhibits better aftermarket performance. For this reason, this study expected a negative relationship between the number of IPOs in the preceding three months and the aftermarket performance of the IPO.
The performance measures of firms that have been IPO-ed at the beginning of the year are more accurate than the performance measures of firms that are IPO-ed at the end of the year (Gompers & Lerner, 1999). A greater probability of changes in the performance measures exists because the measures are not yet available to the public. Cao (2011) has found evidence of deteriorating operating performance in the aftermarket performance of IPOs. Therefore, it is expected that IPO timing is negatively related to long-term aftermarket performance because of the deterioration in aftermarket performance.
The industry is based on the NACE industry classification codes, which is the official industry-standard classification system used in the EU. Most of the firms that went public between January 2000 and December 2013 are concentrated within the manufacturing industry. Therefore, a dummy variable for the manufacturing industry was taken into account. To capture the effect of the financial crisis, a dummy variable was created for the firms that went public in 2008 and 2009. Low stocks prices were expected for this period because, during the financial crisis, investors were not willing to pay high prices for stocks. With low stock prices, there was a potential for better aftermarket performance. It was expected that firms that went public in 2008 or 2009 exhibited a better aftermarket performance.
3.4. Data collection
A PE-backed IPO is defined as a firm in which PE sponsors have a controlling interest at the time of the buyout (Cao, 2011). January 2000 was chosen as the start date of the sample due to the limited availability of data before this date. The time frame ends in 2013 because of the need for 3-year post-IPO return data so that recent data can be included
Initially, a sample of 539 completed and confirmed PE-backed IPOs in the enlarged EU was retrieved from Zephyr. We eliminated all financial firms that contained the standard classification code (SIC) 6xxx, including investment trusts and venture-capital trusts (VCTs); utility firms with the SIC code 49xx; foreign issuers; and re-admissions (Levis, 2011). Government frequently stipulates specific regulations for these industries, so the pricing and the returns of these firms could be biased (Peltzman, 1976). Information regarding the length of the lock-up was extracted from Thomson One, which makes available the prospectuses for 90 firms. However, if the lock-up length was not available in Thomson One, prospectuses were hand-collected and used to find the lock-up length. For 56 IPOs, it was not possible to find the lock-up length because the prospectuses of the IPOs were not found. Thomson One primarily employs the codes ‘management lock-up’, ‘firm lock-up’, and ‘selling-shareholder lock-up’. Since this study was interested in the lock-up period of PE sponsors, it used the lock-up period of the selling-shareholder lock-up. The underpricing of IPOs demonstrated an abnormal increase to 65% during the internet bubble (1997–2000). After the internet bubble, underpricing decreased to close to the average level (Loughran & Ritter, 1995). Therefore, PE-backed firms that were IPO-ed during the dot-com bubble in June 2000 were excluded from the sample (Levis, 2011). Ultimately, the stock returns on 11 firms were not available in DataStream, and were thus also eliminated from the sample. Table 7 presents a summary of the steps taken in the search strategy.
Table 7: Numerical overview of the search strategy
Search Strategy Zephyr Removed Added Total
Region: EU enlarged (28) 599,679
Time period: 01/01/2000–31/12/2013 281,943 317,736
IPO (Flag) 312,040 5,696
PE/VC-backed 5,157 539
Completed and confirmed IPOs 364 175
Dot.com bubble 10 165
Exclude SIC code 6xxx 7 155
Exclude SIC code 49xx 6 149
Thomson One lock-up length 149
Prospectus hand-collected 57 206
Prospectus not found 56 150
DataStream ERROR code 11 139
The market conditions determine whether the firms were IPO-ed during a relatively calm period or during the so-called hot-period. The market conditions were based on the total number of public offerings in the three months prior to each public offering. To find this information, data on all IPOs in the EU (PE- and non-PE-backed) were retrieved from Zephyr.
DataStream was used to collect the daily returns and the first-day closing prices. The PE-backed IPO firm’s leverage ratio, asset turnover, NACE industry code, and date of incorporation were all obtained from Thomson One. The data used for Fama and French’s (2016) five-factor model were taken from Kenneth French's homepage.3 It includes the five Fama and French European factors—market risk premium, SMB, HML, RMW, and CMA— as well as the risk-free rate.
4. Data and descriptive statistics 4.1. Descriptive statistics
Table 8 reports the geographical composition of the 139 PE-backed IPOs included in this thesis.
Table 8: Geographical composition of the total sample of PE-backed IPOs in Europe, 2000–2013
Country Freq. Relative (%) Amount Raised (EURm) Relative (%)
UK 49 36 13,134.68 30 Germany 21 14 9,962.82 23 France 15 10 4,764.99 11 Italy 13 9 3,899.79 9 Sweden 7 4 1,750.82 4 Spain 6 4 1,665.10 3 Austria 5 3. 817.69 1 Switzerland 4 2 846.29 1 Denmark 4 2 2,668.63 6 Netherlands 4 2 1,093.39 2 Belgium 3 2 2,025.38 4 Poland 3 2 136.93 0 Norway 2 1 242 0 Finland 1 1 54.4 0 Ireland 1 1 42.5 0 Luxembourg 1 1 156.75 0 Total 139 100 43,262.15 100
The IPOs are concentrated in 16 different countries, primarily in Western Europe, the UK, and Germany. The majority of the sample is concentrated in the UK and Germany, with 49 and 21 IPOs, respectively. These two countries together account for more than 51% of the total sample. The aggregate amount of capital raised by all IPOs is EUR 43,262.15 million.
3The factors SMB, HML, RMW, CMA, Risk-free rate and market risk premium are retrieved from the Kenneth
Table 9 reports both the annual distribution of the sample IPOs across completion years and the amount raised per year. One observation is the effect of the increase in the number of IPOs prior to the 2007–2008 financial crisis. Approximately 67% of the total number of IPOs took place in the years 2004, 2005, 2006, and 2007, with 22, 20, 33, and 17 IPOs completed, respectively. In 2008, only one IPO was completed. By contrast, in 2013, 19 IPOs were completed. The results indicate that companies prefer an IPO during favourable economic conditions.
Table 9: Annual distribution of PE-backed IPOs in Europe, 2000–2013
Year Freq. Relative (%) Amount raised (EURm) Relative (%)
2000 3 2 515.57 1 2001 3 2 175.90 1 2002 9 6 2,420.18 5 2003 2 1 1,662.74 3 2004 22 15 3,675.10 8 2005 20 14 5,432.58 12 2006 33 25 10,587.81 24 2007 17 11 5,314.40 12 2008 1 0.7 187.35 0 2009 1 0.7 98.50 0 2010 7 4 3,395.59 7 2011 2 1 497.98 1 2012 0 0 0 0 2013 19 13 9,298.45 21 Total 139 100 43,262.15 100
Table 10: Lock-up length of PE-backed IPOs in Europe, 2000–2013
Length Freq. Relative (%)
0 0 0 90 15 10 120 1 1 160 2 1 180 90 63 350 1 1 360 5 3 356 22 17 547 1 1 790 1 1
The lock-up period in the sample varies between 90 and 790 days. Although the most common lock-up period is 180 days, 90 firms reported this length as their restrictive period. Overall, the results from Panel A indicate that lock-ups in the EU are quite homogeneous in terms of their length. This is in line with the findings for US IPOs (Field & Hanka, 2001).
Table 11 provides an overview of the industry distribution of the IPOs in the sample, and presents the absolute and relative number of IPOs per industry, according to the NACE industry classification codes. In addition, the absolute and relative amounts raised per industry are also presented. The NACE classification codes are used to find the industry classification for each IPO.
Table 11: The industry distribution of PE-backed IPOs in Europe, 2000–2013
Industry Freq. Relative (%) Amount Raised (EURm) Relative (%)
Manufacturing 62 45 20,131.34 46
Wholesale and retail trade; repair of motor
vehicles and motorcycles 15 11 4,706.65 10
Information and communication 16 11 4,251.31 9
Professional, scientific, and technical
activities 16 11 4,141.00 9
Administrative and support-service activities 7 4 2,662.43 6
Mining and quarrying 2 1 667.41 1
Real-estate activities 6 4 2,788.69 6
Human-health and social-work activities 3 2 694.99 1
Arts, entertainment, and recreation 3 2 173.31 0
Transportation and storage 3 2 1,350.37 3
Accommodation and food-service activities 2 1 293.64 1
Construction 1 1 127.72 0
Public administration and defence;
compulsory social security 1 1 1,239.70 2
Education 0 0 0.00 0
Other service activities 1 1 13.06 0
Total 139 100 43,242 100
samples (Levis, 2011; Cao, 2011). Within the manufacturing industry, 46% of the total amount is capital-raised.
Table 12 reports the distribution of the PE-backed IPOs by sponsor, the average and median lock-up periods, and the total amount of capital raised per sponsor. The sponsor 3I Group was most frequently involved in public offerings, with 15 IPOs, followed by CVC Capital Partners, with seven. However, CVC Capital Partners raised the most capital at EUR 6.86 billion, followed by Goldman Sachs Capital Partners at EUR 3.11 billion. Considering that fact, the table indicates that PE sponsors are consistent players in the IPO market, suggesting that PE players care about their reputations.
Table 12: Distribution of PE-backed sponsors for IPOs in Europe, 2000–2013
Sponsor Frequency Lock-Up Period Offer Proceeds
No. of
Obs. Relative (%)
Mean Median Amount Raised
(EURm) Relative (%)
3I Group 15 22 203 180 2,070 7
CVC Capital Partners 7 10 273 180 6,859 24
APAX Partners Worldwide 5 7 180 180 380 1
Barclays Private Equity 3 4 180 180 239 1
Blackstone Group 3 4 180 180 1,354 4
Carlyle Group 3 4 242 180 1,734 6
EQT 3 4 217 180 1,605 5
Goldman Sachs Capital Partners 3 4 180 180 3,117 11
KKR 3 4 150 180 1,122 4
Advent International 2 2 180 180 823 2
Alpha Group 2 2 225 180 248 1
BC Partners 2 2 180 180 1,257 4
Credit Suisse First Bosten
Ceramic Partners 2 2 125 180 795 2
Doughty Hansen & Co 2 2 180 180 170 1
Enterprise Investors 2 2 180 180 28 0
Hicks Muse Tate & Furst 2 2 180 180 2,257 8
Lindsay Goldberg 2 2 363 180 669 2 Permira Advisors 2 2 180 180 935 3 Wellington Partners 2 2 270 180 75 0 Wendel 2 2 180 180 2,163 7 Total 67 100 27,899 100
terms of EBITDA-margin, IPO-proceeds leverage ratio, and asset turnover. For example, the leverage ratio varies widely amongst the IPOs in the sample set. The minimum leverage ratio almost equals zero (2%), and the maximum is above 100 (173%). The median of the leverage ratio is equal to 69%, which confirms that PE-backed IPOs make use of high levels of debt. The negative minimum EBITDA-margin is equal to 97%, with a maximum of 71%. However, the median is equal to 14%, indicating that several outliers influence the sample. The offer proceeds vary between the range 1.85 million and 2 billion. The minimum asset turnover equals 0.07%, and the maximum equals 546%. This variation is explained by the firms’ operations across different industries. For example, firms within the service-activities industry provide services and purchase a minimum amount of assets. Conversely, firms operating within the real-estate activities industry own and purchase assets, and sell them to customers.
Table 13: Summary statistics of the PE-backed operational firm, IPO and PE-firm characteristics
Variable No. Of Obs. Mean Median Std. Deviation Min Max Skewness Kurtosis Panel A. PE-backed operational firm characteristics
Leverage ratio (%) 118 69.58 69.26 28.26 2.84 173.45 –0.09 0.61 EBITDA-margin 105 13.88 14.35 26.02 –97.2 71.95 –2.17 7.91 IPO proceeds (EURm) 139 306.82 145.22 385.95 1.85 2,000.00 1.98 3.9 Asset turnover (%) 119 100.51 81.13 87.77 0.07 546.28 2.19 6.96
Panel B. IPO Characteristics and PE-firm characteristics
IPOs in the previous three months 139 134.58 126 59.24 43 271 0.22 –0.92 Lock-up in days 139 216.78 180 102.74 90 790.83 1.91 6.18 Underpricing (%) 134 5.87 3.28 14.65 –92.11 51.72 –1.59 15.29 Timing in years 139 0.56 0.49 0.25 0.08 0.99 0.2 –1.13 Timing in days 139 203.24 180 91.68 30 361 –0.2 –1.13 Age of firm, in years 139 35.89 10.02 49.49 1.16 160.04 2.2 2.63
lock-up period of 790 days. This finding is consistent with the literature on US PE-backed IPOs (Brav et al., 2000; Field & Hanka, 2001; Mohan & Chen, 2002; Bradley et al., 2004). The median value of underpricing is lower than the findings of European IPOs in general (Bergström et al., 2006; Levis, 2011). The median IPO timing is 180 days, indicating no strong preference amongst sponsors for taking the firm public at the beginning or at the end of the year. The median age of the firms involved is 10 years. The average age for a firm is 35 years due to the presence of firms established prior to 1900. On the contrary, younger firms established only one year prior to the IPO are also represented in the sample, which explains the high standard deviation.
Table 14 presents the descriptive statistics for the dependent variables used in this thesis. Panel A reports on the 1-, 2-, and 3-year abnormal returns using Fama and French’s five-factor model. Panel B reports the summary statistics of the abnormal returns around the lock-up expiration date, based on MacKinlay (1997). The number of observations, the mean, standard deviation, the median, the minimum and maximum, the skewness, and kurtosis are presented.
Table 14: Summary statistics on the IPO performance characteristics of PE-backed IPOs in Europe, 2000–2013
Panel A: IPO performance characteristics for 1-, 2-, and 3-year holding periods
Variable (%) No. Of Obs. Mean Median Deviation Std. Min Max Skewness Kurtosis One-year performance (t = 252) 139 –1.13 –1.133 0.007 –2.235 0.103 0.311 –1.289 Two-year performance (t = 504) 139 –1.046 –1.214 0.007 –1.993 0.042 0.352 –1.391 Three-year performance (t = 756) 139 –0.003 0.002 0 –0.207 0.091 –1.058 2.514 Panel B: IPO performance characteristics around the lock-up expiration date
The results in Panel A find negative abnormal returns after one-, two-, and three years following the IPO. The findings align with the existing literature indicating that IPOs, in the long run, tend to underperform (Levis, 1993; Ibbotson, 1975; Ritter, 1984; Ibbotson & Ritter, 1995; Ritter, 1991; Loughran & Ritter, 1995; Espenlaub et al., 2000). The results in Panel B correspond with the existing literature on negative abnormal returns around the lock-up expiration date for IPOs (Field & Hanka, 2001; Brav et al., 2000; Ofek & Richardson, 2000; Brav & Gompers, 2003; Bradley et al., 2004).
Finally, to avoid multicollinearity in the OLS regression, it was necessary to test whether the predictor variables mentioned above were correlated with each other. In the case of multicollinearity, the results from the regression can be distorted. To detect multicollinearity, a correlation matrix, which is presented in Table 19 in the appendix, was set up. The correlation coefficients of the main variables did not surpass 0.80; therefore, multicollinearity does not seem to be a problem.
5. Results
5.1. Fama and French’s five-factor model
Table 15 presents the results of the Student t-test for the 1-, 2-, and 3-year holding periods for 139 PE-backed IPOs in the EU between 2000 and 2013. The t-statistics and the significance level are reported for these holding periods with the IPO defined as day zero. Furthermore, the t-statistics and the significance level are reported for the one-two and two-three year-on-year long-term performance.
Table 15: Sample t-test long-term aftermarket performance of PE-backed IPOs
Alpha t Sig. (2-tailed)
α 1 –17.952 0.000*** α 2 –17.716 0.000*** α 3 –0.740 0.461 α 12 –14.609 0.000*** α 23 –10.387 0.000***
The results demonstrate that after the 1- and 2-year holding periods, the abnormal returns are statistically significant, and underperformance occurs in the first three years following the IPO. The negative abnormal returns are consistent with the previous literature on the aftermarket underperformance of IPOs in the EU over the long term (Jaskiewicz et al., 2005; Levis, 1993).
aftermarket performance compared to both VC-backed IPOs and their non-backed counterparts. Levis (2011) has found a negative 3-year abnormal return of 20% for non-PE-backed IPOs. The results of this thesis find a statistically negative abnormal return of 10%. Based on Levis (2011), it could be argued that PE-backed IPOs outperform non-backed IPOs in the three years following the IPO. However, the negative abnormal returns diminish over the first three years. These diminishing returns are in line with the findings by Espenlaub et al. (2000), who report negative abnormal returns after the first three years. The negative long-term performance is explained by Purnanandam and Swaminathan (2004), who described that overvalued IPOs run up the first day and revert to fundamental value over time. This reversion explains the negative abnormal returns after the first day of trading.
5.2. Event study
Table 16 displays the results of the CARs around the lock-up expiration date for 139 PE-backed IPOs in Europe between 2000 and 2013. The first column indicates the event window for which the CARs are calculated. The event windows are calculated from [–1, +1] to [–10, +10]. The event windows [–2, +2], [–3, +3], and [–4, +4] show statistically significant negative CARs. The event windows from [–5, +5] up to [–10, +10] exhibit a lower degree of CARs, and are not statistically significant.
Table 16: Cumulative abnormal returns around the lock-up expiration date of PE-backed IPOs in Europe, 2000– 2013
Event window t Sig. (2-Tailed)
CAR1 –0.380 0.705 CAR2 –1.786* 0.076 CAR3 –2.966** 0.004 CAR4 –2.609*** 0.010 CAR5 –0.769 0.443 CAR6 –0.217 0.828 CAR7 0.153 0.879 CAR8 –0.052 0.959 CAR9 0.100 0.920 CAR10 0.322 0.748
The *, **, and *** denote statistical significance at 10%, 5%, and 1%, respectively.
Furthermore, there is a marginal but statistically insignificant over-performance at the lock-up expiration date [–1, +1], which could be explained by the offer price, which is set higher than the fundamental value of the stock price.
The negative abnormal returns decrease when approaching the lock-up expiration date. The economic explanation for this decrease is related to the level of asymmetric information. For example, IPOs may be subject to less informational asymmetries following a greater amount of publicity surrounding the IPO (Bergström et al., 2006). Levis (2011) has found that the decrease in information asymmetries leads to less underpricing and to better aftermarket performance for PE-backed IPOs in the long run, as compared to their non-backed counterparts.
Table 17: Cumulative abnormal returns around the lock-up expiration date of PE-backed IPOs in Europe, 2000–2013 t Sig. (2-Tailed) CAR [–1,0] –0.737 0.463 CAR [–2,0] –1.400 0.164 CAR [–3,0] –1.987** 0.049 CAR [–4,0] –2.273** 0.025 CAR [0,1] 0.309 0.758 CAR [0,2] –0.665 0.507 CAR [0,3] –1.058 0.292 CAR [0,4] –0.779 0.437
The *, **, *** denote statistical significance at 10%, 5%, and 1%, respectively
5.3. Regression
Table 18 presents the results from the regressions. The regression models have different dependent variables. These are Fama and French’s abnormal returns, after a holding period of 1 year, 2 years, and 3 years, on the lock-up length and the control variables. One of the important observations from Table 18 is a positive and significant coefficient (at the 1% level) for lock-up length after a 3-year holding period.
financial distress. However, while the leverage ratio shows a negative coefficient after both the 2- and 3-year holding periods, it is not statistically significant.
Table 18: Regressions for the aftermarket performance of PE-backed IPOs in Europe, 2000–2013
This table presents the OLS regression results, where the dependent variables are the 1-year, 2-year, and 3-year abnormal returns, calculated based on Fama and French’s (2016) five-factor model. All regressions are performed with an intercept. The p-values are reported in parentheses, and the beta coefficients represent standardised coefficients. The explanations of the variables used in this regression are explained in Table 6.
Variables One-Year Holding Period Two-Year Holding Period Three-Year Holding Period Coefficient
(p-value) t-statistic Coefficient (p-value) t-statistic Coefficient (p-value) t-statistic Intercept –0.003 –0.264 –0.001 –0.053 –0.001 –0.972 Lock-up length 0.059 0.505 –0.005 –0.046 0.307*** 2.801 (0.006) Underpricing (%) –0.080 –0.669 –0.153 –1.301 –0.139 –1.240 Leverage (%) –0.203* –1.685 –0.146 –1.227 –0.102 –0.897 (0.096) Asset turnover (%) 0.095 0.744 0.147 1.161 0.042 0.351 EBITDA-margin 0.283** 2.016 0.325** 2.345 –0.031 –0.236 (0.047) (0.022) Market condition –0.086 –0.717 –0.104 –0.884 0.172 1.526 IPO proceeds –0.136 –1.042 –0.149 –1.162 –0.087 –0.709 Market timing –0.037 –0.314 0.026 0.230 –0.093 –0.849 Age 0.108 2.016 0.097** 0.799 0.231 –2.005 (0.049) Dummy manufacturing –0.003 –0.031 0.006 0.057 –0.019 –0.177 Dummy crisis 0.124 1.066 0.217* 1.891 –0.026 –0.236 (0.062) Adjusted R-squared –0.007 0.143 0.107 ANOVA (F) 0.941 1.150 1.948** (0.046) The *, **, and *** denote statistical significance at 10%, 5%, and 1%, respectively.
negative, after a 3-year holding period (but statistically insignificant). One explanation for this is described by Cao (2011), who finds a pattern of post-IPO deterioration. The return on EBITDA divided by sales is at its highest level just before the IPO, but decreases sharply in the period after the IPO, and exhibits underperformance in the long run. Moreover, the negative EBITDA-margin coefficient after three years is related to so-called ‘quick-flips’. The PE sponsors took these firms private and ‘flipped’ them back to the public market quickly.
The firm-age coefficient is positive (0.097) and statistically significant at the 5% level, after a 2-year holding period. This finding indicates that older firms do increase aftermarket performance. Older firms are likely to have less information asymmetry due to longer operating histories (Ritter, 1991). The results of this thesis’s research also support the findings of Schultz (1993) as well as Demers and Joos (2007) that the probability of failure decreases with increasing age, and that younger (less established) firms are more likely to fail.
The results find a positive and significant coefficient (at the 10% level) for the effect of the crisis dummy after a 2-year holding period. The dummy was assigned to IPO-ed firms in 2008 and 2009, taking into account whether the crisis affected the post-IPO aftermarket performance. A potential reason is the relatively low offer price during the crisis, since investors were not willing to overpay on the share prices. However, after the 3-year holding period, the results found a negative, but statistically insignificant, effect of the crisis on the aftermarket performance of IPOs, indicating that the crisis depreciated the stock prices.
6. Conclusion and Limitations
The objective of this thesis was to investigate the relationship between the lock-up period and long-term aftermarket performance of PE-backed IPOs, and to investigate the influence of the lock-up agreement on the returns around the lock-up expiration date for these IPOs. This research was conducted for PE-backed IPOs in the EU for the period from January 2000 to December 2013. In a sample of 139 IPOs, divided over 15 different industry classifications in 16 European countries, long-term abnormal returns were compared.