Short and long-term performance of European IPO’sAmsterdam Business School Name Bart van Wijck Student number 10573550 Program Economics & Business Specialization Finance & Organization Number of ECTS 12 Supervisor Drs. P.V. Trietsch Target completion 29/06/2016 Abstract A sample of 1,782 European initial public offerings (IPO’s) that go public in the 2000-2009 period is used to test whether these IPO have a significant effecton the stock price. To test wheter this is true, this study considers overperformance in the short run and/or underperformance in the long run. This study uses the buy-and-hold abnormal returns method to compare stock returns to the Euronext 100 market index. The IPO’s show positive and significant average initial returns of 8.82 percent. This study also finds significant buy-and-hold abnormal returns of -7.48 percent three years after IPO. The significant average five-year abnormal return is -6.46 percent. Cross-sectional analyses show a significant size effect that is negative in the short run and positive in the long run. There are no significant effects for country and industry in the long run. Keywords: Initial Public Offering, IPO, Europe, underperformance
Table of Contents 1. Introduction ... 3 1.1. Cause ... 3 1.2. Existing literature ... 3 1.3. Contribution ... 3 1.4. Research Question ... 3 1.5. Cross-sectional analyses ... 4 1.6. Data & Methodology ... 4 1.7. Outline ... 4 2. Literature review ... 5 2.1. An IPO’s first day ... 5 2.2. Why do companies overperform on the first day? ... 5 2.2.1. Analyst self-selection ... 5 2.2.2. Excess Demand ... 6 2.2.3. Principal-Agent problem ... 6 2.2.4. Investor compensation ... 6 2.2.5. Cascades ... 7 2.2.6. Signaling ... 7 2.3. What happens in the long-run? ... 7 2.3.1 Market efficiency ... 7 2.3.2. Investor overoptimism fades away ... 8 2.3.3. Wrong investor expectations ... 9 2.3.4. Higher market returns ... 9 2.4. Previous studies ... 9 2.5. Which variables have an effect on IPO performance? ... 10 2.5.1. Issue size ... 10 2.5.2. Momentum ... 11 2.5.3. Country ... 11 2.5.4. Industry ... 11 2.6. Caveats in performance research ... 12 3. Data and Methodology ... 13 3.1. Data ... 13 3.2. Methodology ... 14 3.3. Cross-Sectional Analyses ... 14 3.3.1. Size and Country analysis ... 14 3.3.2. Industry analysis ... 15 4. Analysis ... 16 4.1. Empirical Results ... 16 4.1.1. Short-term abnormal returns ... 16 4.1.2. Long-term abnormal returns ... 16 4.2. Cross-Sectional Analyses ... 18 4.2.1. Size & country effects ... 18 4.2.2. Industry effect ... 20 5. Conclusion and remarks ... 22 References ... 23 Appendix A ... 25 Appendix B ... 26 Appendix C ... 27
1. Introduction 1.1. Cause Over the years there have been a lot of initial public offerings (IPO’s) around the world. Some of these performed well while others failed. Many IPO’s show positive first day returns, followed by a period of underperformance when compared to a market index. There seems to be a positive short-term effect and a negative long-term effect. 1.2. Existing literature There are many articles about the performance of IPO’s. The short-term effects have been discussed by Ibbotson (1975); Aggarwal & Rivoli (1990); Ritter (1991) and Levis (1993). They all find positive and significant first-day returns. These returns are followed by a period of underperformance as shown by Aggarwal and Rivoli (1990), Ritter (1991), Loughran (1992), Levis (1993) and Loughran & Ritter (1995) in their research studies. 1.3. Contribution Since most of these studies only reviewed performance for three years after their IPO it is unclear how IPO’s perform after this period. According to his main empirical findings, Ritter argues that the observed underperformance does not continue much beyond 3 years (p. 24, 1991). This study is set up to find out whether there is overperformance in the short run and underperformance in the long run. If there is evidence of underperformance, this study finds out if it extends beyond three years after IPO. It focuses on the short (day-one) and long-term performance (up to five years) of European IPO’s in the 2000-2009 period, aiming to answer the main research question. 1.4. Research Question “Do European companies overperform in the short-run and underperform in the long-run after their IPO?”
1.5. Cross-sectional analyses To try and find an explanation for this underperformance, several cross-section analyses are conducted based on the issue size (in terms of gross proceeds), the country in which the IPO takes place and the industry the issuing firm is active in. 1.6. Data & Methodology A sample of 1,782 European IPO’s from the 2000-2009 period will be used to answer the research question. The performance of the IPO’s will be calculated using Buy-and-Hold Abnormal Returns (BHARs). 1.7. Outline The structure of this thesis is as follows. Chapter two will cover existing literature about IPO underperformance with focus on how underperformance arises, results from past researches and which variable may have an effect on IPO performance. Chapter three will cover the data and methodology with explanations about the sample, the method for calculating returns and the models used for the cross-sectional analyses. Chapter four covers the discussion of the empirical findings, including the cross-sectional analyses. Finally, chapter five gives the conclusion to this research, including the answer to the research question and remarks for future research.
2. Literature review This chapter will discuss the literature on IPO performance and the findings of other studies. 2.1. An IPO’s first day Before discussing the long-term performance of IPO’s it is important to first look at the short-run performance. There is a substantial amount of literature on positive first day returns for IPO’s. Table 1 shows the results of previous research studies covering first-day returns. Both Ibbotson (1975, pp. 246-248) and Ritter (1991, p. 11) compared the IPO performance to a sample of matching firms. Aggarwal & Rivoli (1990, p. 47) and Levis (1993, p. 33) compared IPO performance to a market index. Table 1: Findings in previous studies on first-day returns
Study Period Country Sample Size Day one Returns
% Ibbotson (1975) 1960-1969 USA 120 11.4 Aggarwal & Rivoli (1990) 1977-1987 USA 1,598 10.7 Ritter (1991) 1975-1984 USA 1,526 14.3 Levis (1993) 1980-1988 UK 712 14.3 2.2. Why do companies overperform on the first day? There are several theories as to why IPO’s show positive first-day performance. 2.2.1. Analyst self-selection McNichols & O’Brien (1997) argue that there is a relation between an analysts’ recommendation to buy shares and the information the analyst has about the future earning prospects this share has (1997, p. 167). Analysts only follow the firms that have good prospects. The study shows that this self-selection by analysts leads to overoptimism. This leads to more recommendations for the stock and thus more demand, which will result in a higher return (1997, p. 197).
2.2.2. Excess Demand Both Beatty & Ritter (1985) and Shiller (1990) argue that underwriters price the shares below their actual value to create excess demand. Beatty & Ritter argue that the offerings with price increases are the ones that are oversubscribed (1985, p. 215). They also state that investors will only buy if they are informed about the underpricing (1985, p. 215). Shiller (1990) compares underwriters to impresarios. They create an illusion of excess demand to make the event seem very popular. This leads to a higher trading volume and therefore a higher price (1990, p. 62). Rock (1986) argues that underwriters price shares below their value to make them easier to sell (1986, p. 205). 2.2.3. Principal-Agent problem Baron (1982) argues that investment bankers have superior information about demand in capital markets (1982, p. 956). He shows that the IPO is a Principal-Agent problem with the issuing firm as the principal and the underwriter as the agent (1982, p. 957). The compensation the underwriter will receive in the optimal contract is a function of IPO proceeds and the price of the shares. Because the underwriter has more knowledge about the market, it has an incentive to underprice the shares to attract more investors and spend less money on promotion (1982, p. 975). In a response to Baron’s (1982) research, Muscarella & Vetsuypens (1989) conduct a research with a sample of investment banks going public. Since these banks act as their own underwriter there should be no asymmetric information between them and the issuer (1989, p. 126). They find a mean initial return of 7.12 percent on the first day, implying there is still underpricing (1989, pp. 130-132). Therefore, Baron’s model cannot explain the underpricing phenomenon (1989, p. 135). 2.2.4. Investor compensation Benveniste & Spindt (1989) show the process of underwriters pricing and allocating IPO’s (1989, p. 343). Underwrites base price on interest they sense among investors. But these investors have no incentive to reveal their expectations. By keeping this information to themselves, they try to benefit and
make a profit (1989, p. 344). To make investors reveal their information, the underwriter underprices the stock as compensation. The more information the investors reveal, the more compensation they get, the more underpriced the stock is (1989, p. 344). 2.2.5. Cascades Welch (1992) shows that the IPO market is subject to cascades. Because of the limited distribution channels underwriters have, it takes time to approach all investors. The later informed investors have observed the IPO performance up to that point (1992, p. 695). These investors base their purchasing decision on the decisions of the earlier investors and completely ignore their own information (1992, p. 696). Because of this, underwriters set a lower price for the share to give the early investors an incentive to buy. According to the cascade in Welch’ model, later investors will now buy as well, irrespective of their own information (1992, p. 722-723). 2.2.6. Signaling Allen & Faulhaber (1988) find there is signaling in the IPO market. They find that the issuing firm has more knowledge about its future prospects than anyone else (1988, p. 319). To inform investors about these prospects they underprice their shares. They hope the price goes up before the firm does a seasoned equity offering (SEO) (1988, p. 304). Then, at the SEO, the firm can ask a higher price for the new shares because of the higher price of shares already on the market. A higher price for the new shares will result in more capital being raised (1988, p. 318). 2.3. What happens in the long-run? The reason the first day of IPO’s is discussed first is because the short-term effects lead to underperformance in the long run. Many theories point to a form of the efficient-market hypothesis as to why IPO returns drop over time. 2.3.1 Market efficiency If markets are efficient, it means that prices completely reflect all available information (Fama, 1997, p. 284). But, according to Fama, this is not the right
description for price formation (1997, p. 284). He argues that investors mainly focus on past performance. This leads to an overreaction and thus a higher price. Fama also states that this overreaction leads to poor long-term performance (1997, p. 286). However, in the model Fama creates he shows that, consistent with the market efficiency hypothesis, overreactions happen just as frequent as underreactions. Therefore claiming that these anomalies are subject to chance (1997, p. 304). Fama continues by arguing that only firms that show great potential for the future go public. This results in optimistic valuations and overreactions by investors (2004, p. 286). On the basis of the efficient market hypothesis, the long-run underperformance arises from price revaluations as more information reaches the investors over time (1997, p. 289). 2.3.2. Investor overoptimism fades away Chahine (2004) shows that strong past performance leads to overoptimism among investors at IPO and thus initial positive returns (2004, p. 87). The study shows analysts’ forecasts tend to systematically exceed actual earnings figure (2004, p. 100). As soon as more information about future company earnings becomes public, the price will correct. Now, the performance in the long run not only depends on past earnings, but also on the revision of future earnings (2004, p. 98). Miller (1977) shows that the buyers of IPO shares are the ones that are the most optimistic about them. The true value of these shares is uncertain (1977, p. 1154). The difference in expectations between optimistic and pessimistic investors increases with the uncertainty. Over time, more information about the company reaches the investors (1977, p. 1155). The difference will decrease, resulting in a price drop. According to Miller, this is the reason IPO’s underperform in the long run (1977, p. 1155). A result of the impresario theory by Shiller (1990) is that after the initial period with high returns, investors keep selling these stocks. This results in a price drop. Shiller argues that these selling activities lead to the long-run underperformance (1990, p. 63).
2.3.3. Wrong investor expectations Loughran & Ritter (1995) claim that investors are betting on long shots (1995, p. 47). They argue that if the true possibility of a firms’ success is three percent, but investors think it is four percent, the overvaluation is 33 percent. It then would take a long period of time over a large sample before investors revise their expectations. This results in underperformance. (1995, p. 47) 2.3.4. Higher market returns The previous theories all point to a decrease in stock returns as a reason for underperformance. Another possibility is that the long-term return on the market index is simply higher than the stock returns. According to Aggarwal & Rivoli the systematic risk is higher for IPO’s than for the market index (1990, p.46). This is shown by Balvers, McDonald & Miller (1988) in their study. The industries where firms went public in all show betas larger than one (1988, p. 617). Ibbotson (1975) finds a beta of 2.2 in the first month after IPO, and a beta of 2 in the second month (1975, p. 258). He also states that betas gradually decline over time (1975, p. 261). With larger than one betas, IPO’s can only underperform the market if its returns are more negative than the market returns. However, Aggarwal & Rivoli (1990) find an average one-year IPO return of -13.73 percent while the NASDAQ index had an average annual growth rate of 14 percent (1990, p. 50). 2.4. Previous studies Studies have shown that performance tends to drop over time. Table 2 shows the results of previous studies on long-term performance. Aggarwal & Rivoli (1990), for example, find negative returns after a 250-day (one year) holding period (1990, p. 46). This is the return for investors that buy stock on day one, after the initial performance has occurred. The researchers also find performance to be negative when stock is bought at offering, so including initial performance (1990, p. 50). Ritter (1991) finds underperformance after three years when comparing IPO’s to a set of matching firms. Loughran (1992) finds that the NASDAQ market index significantly outperforms IPO’s after a six-year holding period (1992, pp.
250 & 259). Loughran & Ritter (1995) also compared IPO performance to the returns from a set of non-issuing matching firms that are similarly sized (1995, p. 33). But these negative returns aren’t exclusive to US samples. In a study covering emerging markets, Aggarwal, Leal & Hernandez (1993) find underperformance in Brazil & Chili. They argue that these results are similar to US and UK returns despite the small amount of observations (1993, p. 52). Keloharju’s (1993) finds underperformance when comparing IPO returns to a value-weighted index composited from the Helsinki Stock Exchange (HSE). The performance is even worse when the returns are compared to an equally weighted market index (1993, p. 273). Levis (1993) finds that returns steadily decline during three years (1993, p. 35). He also states that the underperformance continues after the studied 36 months (1993, p. 35). However, he does not report evidence for this statement in his paper. Table 2: Findings in previous studies on long-term returns
Study Period Country Sample Size Long-term Returns
% Aggarwal & Rivoli (1990) 1977-1987 USA 1,598 -13.7 Ritter (1991) 1975-1984 USA 1,526 -29.1 Loughran (1992) 1975-1984 USA 3,656 -58.9 Loughran & Ritter (1995) 1970-1990 USA 4,753 -6.7 Aggarwal, Leal & Hernandez (1993) 1980-1990 Brazil 62 -47.0 Aggarwal, Leal & Hernandez (1993) 1982-1990 Chili 19 -23.7 Keloharju’s (1993) 1984-1989 Finland 80 -22.4 Levis (1993) 1980-1988 UK 712 -22.9 2.5. Which variables have an effect on IPO performance? In previous studies about IPO performance there has been a lot of discussion on which variables have an effect on the performance and which don’t. 2.5.1. Issue size Aggarwal & Rivoli (1990) split their sample up in four groups by the size of their gross proceeds. They find positive and significant initial returns for all groups
(1990, p. 50). The one-year returns are negative and significant for all groups (1990, p. 50). Ritter (1991) shows that smaller offerings show higher initial returns but also show a worse aftermarket performance (1991, pp. 13-14). The study also shows that all his size categories show underperformance in the long run. (1991, pp. 13-14). 2.5.2. Momentum Ritter (1991) finds that firms try to take advantage of periods with economic growth. He calls this phenomenon ‘windows of opportunity’ (1991, p. 4). This means that in times of economic growth, the number of IPO’s is larger than in times where there is no growth. This adds to a statement by Ibbotson & Jaffe (1975) who argue that underperformance only occurs during particular periods (1975, p. 1038). 2.5.3. Country Ibbotson & Ritter (1995) claim that underpricing happens in every country that has its own stock market. They add that the amount of underperformance would vary from country to country (1995, pp. 994-995). Paul & Mallik (2003) study the relation between stock returns, economic activity and inflation (2003, p. 23). They state that the gross domestic product (GDP) is an accurate measure of economic activity and therefore use it in their study (2003, p. 23). The study shows that stock prices and returns are related negatively to the interest rate and positively to GDP growth. The found no significant effect for inflation (2003, p. 29). 2.5.4. Industry Ritter (1984) studies hot issue markets in which there is a large volume of IPO’s compared to other periods (1984, p. 238). He finds that, in the hot issue market of 1980, the increase in volume only applies to natural resource firms. Underwriters exploit start-up companies in this industry during the oil and gas boom in 1980. When the boom was over, so was the large volume of IPO’s (1984, p. 239).
Ritter (1991) shows that long-run performance varies widely over different industries. The financial institutions show the best performance, Natural resource industries (oil & gas) underperformed (1991, p. 18). Chou, Ho & Ko (2011) find that industries have an effect on stock performance that cannot be explained by the CAPM, Fama-French three-factor and Carhart’s four-factor model (2011, p. 356). They also find that firms in the same industry show a higher correlation since they share similar fundamentals. Because larger firms respond quicker to information, they lead the smaller firms in the same industry (2011, p. 356). The researches conclude that stock returns are related to industries as well as size, book-to-market ratios and momentum (2011, pp. 368-369). 2.6. Caveats in performance research Ibbotson & Ritter (1995) notice a caveat when it comes to long-term performance research. In order to study performance, long holding periods are needed (1995, p. 1007). In a large sample this will cause overlap, resulting in limited independent observations. Previous results may show patterns because of factors than cannot be repeated (1995, p. 1007).
3. Data and Methodology This chapter covers the data used, the methodology and the cross-sectional analyses. 3.1. Data The sample for this research consists of 1,782 initial public offerings from the 2000-2009 period. IPO’s from 2010 and later will not be included in this research since it will not be possible to review performance for the full 5 years. The data comes from the Zephyr database and consists of company name; company country; percentage of total stock offered; the market the stock is listed on; Company SIC Codes; Company ticker symbols; Date of IPO completion; Offer price of stock; Stock price after completion and Stock price one month after completion. For an IPO to be selected for this sample it has to match the following criteria: • European company; • Listed on only one European stock exchange (no double listings); • Daily stock prices are available for 5 years after IPO • Minimum amount of €500.000 raised; There are IPO’s where the firm offers a small amount of equity by offering a small amount of total stock on the market (i.e. 1%). The criterion of minimum €500,000 raised filters these IPO’s out of the sample. The distribution of the sample can be found in Appendix A, table 10. To calculate company returns for the 5 years after their IPO, data from the Datastream database will be collected. This data will consist of daily share prices allowing the returns to be computed. Also, data from the Euronext 100 index will be used to calculate market returns, which will also be extracted from the Datastream database. This will consist of daily index values in the period 2000-2014. Finally, to measure economic growth in the countries in which an IPO takes place, GDP data will be collected from the Datastream database. This will consist of quarterly GDP values in the period 2000-2014 for all countries in the sample.
To be able to analyze a company’s performance, the Buy-and-Hold Abnormal Returns (BHARs) returns method will be used, as suggested by Lyon, Barber & Tsai because they argue it is a representation of the experience investors receive (1999, p. 138). For a company i at time t, it is calculated as follows:
BHARit = Share returnit – Market returnit
After the initial return is calculated, the aftermarket performance will be analyzed. To do so, BHARs will be calculated for two intervals: three years (t = 36) and five years (t = 60) after the first day of trading. This way the initial return period, the first day, will not be included in the long-term returns. To measure the country effect, GDP growth will be used. This is measured over the same period as the BHARs. Since these firms are new to the market, they have no past prices. Therefore it is not possible to establish the risk factor. For simplicity it is assumed that all IPO’s have a beta of one. 3.3. Cross-Sectional Analyses The cross-sectional analyses are based on size, country and industry of the IPO. 3.3.1. Size and Country analysis The effect of IPO size and country will be measured for both the short (day-one) and the long-term (three and five years), using the following regressions. For the short-term (day-one) effects: 𝐵𝐻𝐴𝑅!"# != 𝛼 + 𝛽!𝑙𝑛𝑠𝑖𝑧𝑒 + 𝜀
Where BHARDay 1 are the buy-and-hold abnormal returns for the first day of the
IPO and lnsize is the natural logarithm of IPO size in terms of gross proceeds. The size variable is being transformed into a logged variable to make the coefficient represent the percent change in the explanatory variable in relation to a percentile change of the dependent variable. The regressions for the long-term effects are: 𝐵𝐻𝐴𝑅!"#$ ! = 𝛼 + 𝛽!𝑙𝑛𝑠𝑖𝑧𝑒 + 𝛽!𝐺𝐷𝑃𝐺_𝑌3 + 𝜀 𝐵𝐻𝐴𝑅!"#$ ! = 𝛼 + 𝛽!𝑙𝑛𝑠𝑖𝑧𝑒 + 𝛽!𝐺𝐷𝑃𝐺_𝑌5 + 𝜀
Where BHARYear 3/5 are the buy-and-hold abnormal returns for three and five year holding periods. GDPG_Y3/5 is the GDP growth over the same period as the buy-and-hold abnormal returns. As mentioned earlier, Paul & Mallik (2003) state that the GDP is an accurate measure of economic activity (2003, p. 23). Therefore this research will use GDP growth as a variable to find country effects for IPO’s. Since it is not possible to calculate GDP growth for one day, it is assumed to be zero for the short-term and will not be included in the first regression. Based on findings by Aggarwal & Rivoli (1990) and Chou, Ho & Ko (2011) mentioned in the previous chapter, it is suspected that IPO size will have an effect on IPO performance. Ibbotson & Ritter (1995) claim that underperformance happens in every country with its own stock market (1995, pp. 994-995). Paul & Mallik (2003) find that stock returns are related to GDP growth (2003, p. 29). Therefore it is suspected that the country will also have an effect on IPO performance. 3.3.2. Industry analysis Based on the SIC codes, the sample is divided into the ten main industry groups. The distribution of the groups can be found in Appendix A, table 11. For every industry, the average buy-and-hold abnormal returns will be calculated over the initial return period (day-one) and the three- and five-year holding periods. After that, the difference in means will be calculated using a t-test for the three largest industries in terms of number of IPO’s (‘Manufacturing’, ‘Finance, Insurance and Real Estate’ and ‘Services’) to find out whether there are differences in abnormal returns among industries. The formula for the t-test can be found in Appendix C. Based on findings by Ritter (1991) and Chou, Ho & Ko (2011) it is suspected that industry will have an effect on IPO performance.
4. Analysis This chapter analyzes the calculated returns for the short and the long-run performance. This chapter also includes the cross-section analyses. 4.1. Empirical Results The calculated average BHARs for all 1,782 IPO’s in the sample that covers the 2000-2009 period are shown in table 3. The returns in the table are calculated based on holding periods and are not cumulative. The return for year one is the return for buying the stock on day one and holding it for twelve months. The table shows average abnormal returns for holding periods up to five years (60 months). This is done to show how the change in abnormal returns over time. Table 3: Buy-and-Hold Abnormal Returns by holding period
Mean Std. Err. t-stat
P>|t| Day 1 0.0882 0.3131 11.89*** 0.00 Year 1 -0.0395 0.5680 -2.94*** 0.00 Year 2 -0.0730 0.7451 -4.14*** 0.00 Year 3 -0.0748 0.7761 -4.07*** 0.00 Year 4 -0.0657 0.8787 -3.17*** 0.00 Year 5 -0.0646 0.9571 -2.85*** 0.00 * Significant at 10%, ** 5%, *** 1% or less. 4.1.1. Short-term abnormal returns The initial return is 8.82 percent with a t-statistic of 11.89 making these first day returns statistically significant. The raw day-one returns are 8.84 percent, making the 0.02 percent market return dismissible. These initial returns are similar to results found by Ibbotson (1975), Aggarwal & Rivoli (1990), Ritter (1991) and Levis (1993). 4.1.2. Long-term abnormal returns Table 3 also shows negative and significant returns for all long-term holding periods. The one-year average abnormal return is -3.95 percent which means that the average return dropped 12.77 percent in the first twelve months. In the second year abnormal returns dropped another 3.35 percent to -7.30 percent. The three-year buy-and-hold average abnormal return is -7.48 percent. After
three years the returns increase slightly to -6.57 percent after four years and -6.46 percent after five years. The movement of these returns is also shown in figure 1. Figure 1 shows the average BHARs on the vertical axis and the holding period form day one on the horizontal axis. The line in this figure clearly shows the decline in returns as the holding period get longer and a slight increase after three years. Figure 1: Buy-and-Hold Abnormal Returns by holding period These returns are similar to the findings in studies by Aggarwal & Rivoli (1990), Ritter (1991), Loughran (1992), Loughran & Ritter (1995), Aggarwal, Leal & Hernandez (1993), Keloharju’s (1993) and Levis (1993). However, these studies, except for Loughran & Ritter’s (1995), all show negative returns of thirteen percent or larger. A test to find if there is a difference between the three- and five-year buy-and-hold abnormal returns has given a t-statistic of -0.35 and a p-value of 0.7264. This means that the improvement of average returns between these holding periods in not significant. The formula for this test can be found in Appendix C. 8,82 -3,95 -7,30 -7,48 -6,57 -6,46 -10,00 -8,00 -6,00 -4,00 -2,00 0,00 2,00 4,00 6,00 8,00 10,00 Day 1 Year 1
(t=12) Year 2 (t=24) Year 3 (t=36) Year 4 (t=48) Year 5 (t=60)
4.2. Cross-Sectional Analyses These cross-sectional analyses are aimed to find more insight in the origin of IPO underperformance. 4.2.1. Size & country effects The regression to find short-term (day-one) effects for IPO performance only includes an explanatory variable for size. As mentioned earlier, this is because it is assumed that GDP growth for day one is zero. The outcome of the regression for the day-one effect can be found in table 4. Table 4 shows a positive and significant constant (alpha). The coefficient for lnsize is negative and significant. This means that there is a size effect for IPO performance in the short-run. The positive constant implies that the initial abnormal return for an IPO is positive, but will decrease as lnsize increases. A one percent increase in size will lead to a 0.0161/100 = 0.000161 percent decrease in abnormal returns. The abnormal return becomes negative when IPO issue size becomes larger than 6.75 billion euro’s. Table 4: Regression outcome for day-one returns
BHAR Day 1 Coef. Std. Err. t-stat P>|t|
lnsize -0.0161 0.0041 -3.91*** 0.00 Constant (α) 0.3644 0.0748 4.87*** 0.00 * Significant at 10%, ** 5%, *** 1% or less. The regressions to find effects for long-term performance include an explanatory variable for GDP growth. Table 5 shows the regression outcome for the three-year buy-and-hold abnormal returns. Both the constant and the coefficient for lnsize are significant. The coefficient for GDP growth is positive, but not significant. This implies that the country in which a firm goes public has no effect on long-term performance. Table 5: Regression outcome for three-year returns
BHAR Year 3 Coef. Std. Err. t-stat P>|t|
GDPG_Y3 0.0016 0.0022 0.71*** 0.48
lnsize 0.0622 0.0092 6.77*** 0.00
Constant (α) -1.1428 0.1595 -7.17*** 0.00
Remarkable is that the constant is now negative and the coefficient for lnsize is positive. This means that the long-term abnormal returns for an IPO are negative, but improve by 0.000622 percent when IPO size increases by one percent. Table 6 shows the regression outcome for the five-year BHARs. Similar to the model for three-year returns, the constant and the coefficient for lnsize are significant. The coefficient for GDP growth is, again, not significant. The constant in table 6 is more negative than in table 5 implying that long run abnormal returns decline as the holding period gets longer. The coefficient for lnsize increased compared to table 5. This implies that a one percent increase in size has a larger impact on abnormal returns over a five-year holding period than it does over a three-year holding period. Table 6: Regression outcome for five-year returns
BHAR Year 5 Coef. Std. Err. t-stat P>|t|
GDPG_Y5 0.0006 0.0004 1.60*** 0.11 lnsize 0.0872 0.0116 7.54*** 0.00 Constant (α) -1.5630 0.2014 -7.76*** 0.00 * Significant at 10%, ** 5%, *** 1% or less. The regression outcomes show that abnormal returns are positive and significant in the short-run and negative and significant in the long run. This corresponds to the average abnormal returns found earlier this chapter and findings in previous studies. The outcomes also show that the size of an IPO has a significant effect on both short and long term performance. This confirms the expectation of the size effect stated in the previous chapter. The size effect is negative in the short run and positive in the long run. This is similar to findings by Ritter (1991). Finally, the regressions show that there are no significant country effects. This is contradictory to the expectation stated in chapter three and findings by Paul & Mallik (2003) who did find a significant GDP effect on performance.
4.2.2. Industry effect As mentioned in the previous chapter, for every industry the average buy-and-hold abnormal returns are calculated for three intervals. These average abnormal returns can be found in Appendix B, table 12, 13 and 14. Table 12 shows that all industries show positive average day-one abnormal returns. These returns are all significant except for the Agriculture industry. Table 13 and 14 show the long-term average abnormal returns. The only significant returns are in the ‘Finance, Insurance and Real Estate’ and ‘Services’ industries for both periods. To find out if there are differences among industry returns, the means for the three largest industries are compared with a t-test. The formula can be found in Appendix C. The results are shown in the tables below. Table 7 shows the comparison for day-one returns. There is a significant difference in average abnormal returns between the ‘Finance, Insurance and Real Estate’ and the ‘Manufacturing’ industries and between the ‘Finance, Insurance and Real Estate’ and ‘Services’ industries. There is no significant difference between ‘Manufacturing’ and ‘Services’. This could mean there is an industry effect in short-term (day-one) IPO performance. Table 7: Day-one return industry comparison
Mean 1 Mean 2 t-stat P>|t|
Fin_Ins_RE - Manufacturing 0.0459 0.0710 -1.79** 0.07 Fin_Ins_RE - Services 0.0459 0.1217 -2.32** 0.02 Manufacturing - Services 0.0710 0.1217 -1.64** 0.11 * Significant at 10%, ** 5%, *** 1% or less. Table 8 and 9 show the comparison between the average long-term abnormal returns. There are no significant differences between any of the industries. This implies that there is no industry effect in long-term performance. This is contradictory to the expectations stated in the previous chapter. The findings in this study are also inconsistent with findings by Chou, Ho & Ko (2011) who did find a significant effect for industry on performance in their study.
Table 8: Three year return industry comparison
Mean 1 Mean 2 t-stat P>|t|
Fin_Ins_RE - Manufacturing -0.1184 -0.0379 -1.63 0.11 Fin_Ins_RE - Services -0.1184 -0.1178 -0.01 0.99 Manufacturing - Services -0.0379 -0.1178 0.20 0.84 * Significant at 10%, ** 5%, *** 1% or less. Table 9: Five year return industry comparison
Mean 1 Mean 2 t-stat P>|t|
Fin_Ins_RE - Manufacturing -0.1151 -0.0582 -1.00 0.32 Fin_Ins_RE - Services -0.1151 -0.1250 0.07 0.94 Manufacturing - Services -0.0582 -0.1250 0.39 0.70
5. Conclusion and remarks This study was conducted to find out whether there is IPO overperformance in the short run and underperformance in the long run. If the underperformance exists, this study empirically tests if it extends beyond three years after IPO. Using buy-and-hold abnormal returns for 1,782 European IPO’s in the 2000-2009 period this study is able to show a significant average short-run (day-one) return of 8.82 percent. This is similar to findings by Ibbotson (1975), Aggarwal & Rivoli (1990), Ritter (1991) and Levis (1993). The three- and five-year average abnormal returns are negative and significant at -7.48 percent and -6.46 percent. These long-term returns do not significantly differ from each other. The observed negative long-run returns are similar to findings in studies by Aggarwal & Rivoli (1990), Ritter (1991), Loughran (1992), Loughran & Ritter (1995), Aggarwal, Leal & Hernandez (1993), Keloharju’s (1993) and Levis (1993). Cross-sectional analyses show a significant effect of IPO issue size on both short- and long-term performance. This is similar to findings by Aggarwal & Rivoli and Ritter (1991). The effect for size is negative in the short-run and positive in the long run. Contradictory to findings by Paul & Mallik (2003), this study finds no effect for country, in terms of GDP growth, in long-term performance. Finally, the cross-sectional analysis for industry effects in the three largest groups shows significant differences between groups in the short-run but not in the long run. This is inconsistent with findings by Chou, Ho & Ko (2011) who did find a significant effect for industry on performance in their study. The sample for this study covers IPO’s in the 2000-2009 period. During this period there were two crises in the economy: the burst of the dot-com bubble in 2000, and the financial crisis that started in 2008. It is unclear if and/or how these crises affected IPO performance. This may be interesting for future research. Besides that, this study did not cover all industries to find an effect for performance. There could also be more variables that affect IPO performance not covered in this study. Therefore further research on this topic is needed.
References Aggarwal, R., Rivoli, P. (1990). Fads in the Initial Public Offering. Financial Management 19, 4 (1990), pp. 45-57 Aggarwal, R., Leal, R. & Hernandez, L. (1993). The Aftermarket Performance of Initial Public Offerings in Latin America. Financial Management 22 (1993), pp. 42-53 Allen, F., & Faulhaber, G.R. (1988). Signaling by Underpricing in the IPO Market. Journal of Financial Economics 23 (1989), pp. 303-323 Balvers, R.J., McDonald, B. & Miller, R.E. (1988). Underpricing of New Issues and the Choice of Auditor as a Signal of Investment Banker Reputation. The Accounting Review, Vol. 63, No. 4 (1988), pp. 605-622 Baron, D.P. (1982). A Model of the Demand for Investment Banking Advising and Distribution Services for New Issues. The Journal of Finance 37 (1982), pp. 955-976 Beatty, R.P. & Ritter, J.R. (1985). Investment banking, reputation, and the underpricing of initial public offerings. Journal of Financial Economics 15 (1986), pp. 213-232 Benveniste, L.M. & Spindt, P.A. (1989). How investment bankers determine the offer price and allocation of new issues. Journal of Financial Economics 24 (1989), pp. 343-361 Chahine, S. (2004). Long-run abnormal return after IPOs and optimistic analysts’ forecasts. International Review of Financial Analysis 13 (2004) pp. 83-103 Chou, P.H., Ho, P.H. & Ko, K.C. (2011). Do industries matter in explaining stock returns and asset-pricing anomalies? Journal of Banking & Finance 36 (2012), pp. 355-370 Fama, E.F. (1997). Market efficiency, long-term returns, and behavioural finance. Journal of Financial Economics 49 (1998), pp. 283-306 Ibbotson, R.G. (1975). Price Performance of Common Stock New Issues. Journal of Financial Economics 2 (1975), pp. 235-272 Ibbotson, R.G. & Jaffe, J.F. (1975). “Hot Issue” Markets. The Journal of Finance 30 (1975), pp. 1027-1042 Ibbotson, R.G. & Ritter, J. (1995). Initial Public Offerings. Handbooks in Operations Research and Management Science, Vol. 9 (1995), pp. 993-1016
Keloharju, M. (1993). The winner’s curse, legal liability, and the long-run price performance of initial offerings in Finland. Journal of Financial Economics 34 (1993), pp. 251-277 Levis, M. (1993). The Long-Run Performance of Initial Public Offerings: The UK Experience 1980-1988. Financial Management 22 (1993), pp. 28-41 Loughran, T. (1992). NYSE vs NASDAQ returns: Market microstructure or the poor performance of initial public offerings? Journal of Financial Economics 33 (1993), pp. 241-260 Loughran, T. & Ritter, J.R. (1995). The New Issues Puzzle. The Journal of Finance 50 (1995), pp. 23-51 Lyon, J.D., Barber, B.M. & Tsai, C. (1999). Improved Methods for Tests of Long-Run Abnormal Stock Returns. The Journal of Finance 54 (1999), pp. 165-201 McNichols, M. & O’Brien, P.C. (1997). Self-Selection and Analyst Coverage. Journal of Accounting Research, Vol. 35, Studies on Experts and the Application of Expertise in Accounting, Auditing, and Tax (1997), pp. 167-199 Miller, E.M. (1977). Risk, Uncertainty, and Divergence of Opinion. The Journal of Finance 32 (1977), pp. 1151-1168 Muscarella, C.J. & Vetsuypens, M.R. (1989). A simple test of Baron’s model of IPO underpricing. Journal of Financial Economics 24 (1989), pp. 125-135 Paul, S. & Mallik, G. (2003). Macroeconomic factors and bank and finance stock prices: the Australian experience. Economic Analysis and Policy 33 (2003), pp. 23-30 Ritter, J.R. (1984). The “Hot Issue” Market of 1980. The Journal of Business 57 (1984, pp. 215-240 Ritter, J.R. (1991). The Long-Run Performance of Initial Public Offerings. The Journal of Finance 46 (1991), pp. 3-27 Rock, K. (1985). Why New Issues are Underpriced. Journal of Financial Economics 15 (1986), pp. 187-212 Shiller, R.J. (1990). Speculative Prices and Popular Models. Journal of Economic Perspectives, Vol. 4, No. 2 (1990), pp. 55-65 Welch, I. (1992). Sequential Sales, Learning, and Cascades. The Journal of Finance 47 (1992), pp. 695-732
Table 10: Distribution of Initial Public Offerings by year, 2000-2009
All IPO’s Sample Included
IPO’s Proceeds Gross
IPO’s Proceeds Gross
IPO’s Proceeds Gross
€ Millions € Millions % % 2000 517 71,982 302 40,642 58.4 56.5 2001 192 37,077 114 23,573 59.4 63.6 2002 121 17,596 70 9,934 57.9 56.5 2003 87 8,114 47 5,290 54.0 65.2 2004 222 34,026 146 21,937 65.8 64.5 2005 377 47,743 297 44,174 78.8 92.5 2006 511 79,693 405 61,698 79.3 77.4 2007 432 52,618 298 42,760 69.0 81.3 2008 137 9,359 67 7,739 48.9 82.7 2009 64 5,142 36 5,089 56.3 99.0 Total 2,660 363,350 1,782 262,836 67.0 72.3 Table 11: Distribution of Initial Public Offerings by industry
SIC Code Industry No. of IPO’s Proceeds Gross € Millions 01-09 Agriculture, Forestry and Fishing 15 642 10-14 Mining 89 17,351 15-17 Construction 34 3,090 20-39 Manufacturing 424 52,070 40-49 Transportation, Communications, Electric, Gas and Sanitary service 172 67,331 50-51 Wholesale Trade 50 3,054 52-59 Retail Trade 54 8,399 60-67 Finance, Insurance and Real Estate 418 81,465 70-89 Services 524 29,340 91-99 Public Administration 2 96 Total 1,782 262,836
Table 12: Day-one BHARs for industry
Obs Mean Std. Dev. t-stat P>|t|
Agriculture 15 0,1533 0,3949 1,50*** 0,1558 Construction 34 0,0691 0,1527 2,64*** 0,0126 Fin_Ins_RE 418 0,0459 0,1735 5,41*** 0,0001 Manufacturing 424 0,0710 0,2292 6,37*** 0,0001 Mining 89 0,1140 0,2900 3,71*** 0,0004 PublicAd 2 0,0638 0,0052 17,22*** 0,0369 Retail 54 0,1286 0,3502 2,70*** 0,0093 Services 524 0,1217 0,3700 7,53*** 0,0001 Tr_Com 172 0,0960 0,5099 2,47*** 0,0145 Wholesale 50 0,1154 0,3271 2,49*** 0,0162 * Significant at 10%, ** 5%, *** 1% or less. Table 13: Three year BHARs for industry
Obs Mean Std. Dev. t-stat P>|t|
Agriculture 15 0,0580 0,8022 0,28*** 0,7836 Construction 34 -0,0769 0,8599 -0,52*** 0,6065 Fin_Ins_RE 418 -0,1184 0,5569 -4,35*** 0,0001 Manufacturing 424 -0,0379 0,8457 -0,92*** 0,3581 Mining 89 0,0208 1,0066 0,20*** 0,8419 PublicAd 2 -0,8120 0,5538 -2,07*** 0,2865 Retail 54 0,0042 0,7847 0,04*** 0,9682 Services 524 -0,1178 0,7923 -3,40*** 0,0007 Tr_Com 172 -0,0123 0,7827 -0,21*** 0,8339 Wholesale 50 -0,0533 0,9832 -0,38*** 0,7056 * Significant at 10%, ** 5%, *** 1% or less. Table 14: Five year BHARs for industry
Obs Mean Std. Dev. t-stat P>|t|
Agriculture 15 -0,1782 0,4909 -1,41*** 0,1804 Construction 34 -0,0081 1,4300 -0,03*** 0,9762 Fin_Ins_RE 418 -0,1151 0,7665 -3,07*** 0,0023 Manufacturing 424 -0,0582 0,8789 -1,36*** 0,1746 Mining 89 0,0334 1,1951 0,26*** 0,7955 PublicAd 2 -0,7612 0,1595 -6,75*** 0,0936 Retail 54 0,3042 1,3936 1,60*** 0,1155 Services 524 -0,1250 0,8928 -3,21*** 0,0014 Tr_Com 172 0,0493 1,1614 0,56*** 0,5762 Wholesale 50 -0,0036 1,4180 -0,02*** 0,9841 * Significant at 10%, ** 5%, *** 1% or less.
Appendix C t-test for comparing two means with unequal variances 𝑡 = (𝑥!− 𝑥!) 𝑠!! 𝑛!+𝑠! ! 𝑛! ~𝑡 𝑑𝑓 Where: 𝑑𝑓 = 𝑠!! 𝑛!+𝑠! ! 𝑛! ! 𝑠!! 𝑛! ! 𝑛!− 1 + 𝑠!! 𝑛! ! 𝑛! − 1 x = mean s = standard error n = amount of observations