Long-run performance of European IPOs, during a
period economic distress: 2007-2012
Amsterdam Business School
Name
Jaïr Igor Veder
Student number
10660593
Program
Economics & Business
Specialization
Economics and Finance
Number of ECTS
12
Supervisor
Ilko Naaborg
Target completion 29 / 06 / 2016
Abstract
Many different theories and results have been presented regarding the long-run
underperformance of initial public offerings (IPOs). However, most of the research has been conducted on US-corporations. Although long-run underperformance is confirmed within most research, recent research indicates that underperformance is insignificant (Ritter J. R., 2011) . There are also researchers who find that the long-run performance depends on the time period for the research (Carter, Dark, Floros, & Sapp, 2011). In this thesis 87 European IPOs are analyzed during the period 2007-2012. The reason behind this period is that during these years both the global financial crisis and the European sovereign debt crisis occurred. Therefore, analyzing whether economic distress has an influence on the IPO underperformance could give more insight on the price performances associated with IPOs.
Statement of originality
This document is written by Jaïr Igor Veder who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of Contents
1. Introduction ... 2
2. Literature review ... 3
2.1. Theories regarding initial public offerings ... 3
2.2. Possible explanations underperformance ... 4
2.3. Empirical findings in the literature ... 5
2.4. Conclusion on the literature ... 7
3. Methodology and Data ... 8
3.1. Methodology ... 8
3.2. Data and descriptive statistics ... 10
4. Empirical Results ... 15
4.1. Buy-and-hold returns and buy-and-hold abnormal returns ... 15
4.2. Three factor regression results ... 16
4.3. Conclusion on the results ... 18
5. Conclusion and discussion ... 18
1. Introduction
When a corporation is in need of financing it can either raise money by issuing debt or it can go public. Issuing debt can be costly or not desirable due to restrictive debt
covenants and therefore, going public can be a better alternative. When a corporation issues equity on the securities exchange for the first time it is called an initial public offering (IPO). After the initial public offering the corporation is being publicly traded and therefore, a lot of investors can take an interest in the corporations performance. Consequently investors benefit from knowing the performance patters associated with IPOs.
There are two anomalies associated with IPOs. The first anomaly is the
underpricing of the stock. The second anomaly is the underperformance of stocks in the long-run. Much research has been done on both anomalies, however, most of this research consists of US-stocks. In the research that consists of European-stocks, there are about ten much cited papers on long-run performances, but these papers research a sample in the period before the European Union. For the US market Ritter (1991) argues that investing in an IPO would have resulted in 83% earnings in the three year long-run compared to a matching group of companies.
Despite the fact that much research has been conducted on the long-run underperformance of IPOs, there has been no prior research on periods of economic distress. The global financial crisis and the European sovereign debt crisis could have a significant influence on the long-run performance of IPOs. Therefore, analyzing a sample of European IPOs in the two largest euro countries could give more insight on the long-run performance of IPOs in times of economic distress. The hypothesis tested in this thesis will be:
Where alpha is the abnormal trading day return.
The countries researched in this thesis are Germany and France. There are two reasons for these countries. The first reason is that they have a relatively large amount of IPOs compared to other European countries. Therefore, the sample will be larger and the statistical power will be higher. The second reason is that they are both part of the European Union, and therefore, the global financial crisis and European sovereign debt crisis will impact them in a similar matter. From these countries a total of 87 IPOs will be selected and matching corporations will be assigned to the IPOs.
The remainder of this thesis is organized as follows. The second chapter will discuss the existing theories and literature on IPOs. The third chapter will discuss the methodology and data. The fourth chapter will present and discuss the results from the research. The fifth chapter will discuss the results and conclude.
2. Literature review
This section will exist of four parts. First the basics of initial public offerings and the anomalies will be discussed, second the possible explanations of underperformance will be discussed, third the empirical findings in the literature will be discussed and the last part will be a conclusion of the literature.
2.1. Theories regarding initial public offerings
When a corporation issues equity on the public securities exchange for the first time it is called an initial public offer (IPO), also known as going public. Zingales (1995) finds that for most corporations an initial public offering is the largest equity offering it ever makes. On average one-third of the total equity raised every year is raised through initial public offerings. According to Lowry (2003) the demand for financing depends on the growth expectation of the economy. When the growth forecast are positive the demand for financing is relatively high and therefore, the amount of IPO’s is high in periods of positive growth expectations, also known as the hot markets phenomena. Ritter and Welch (2002) agree on the findings of Lowry and they find that the going-public decision depends on market conditions. When the market conditions are
favorable corporations decide to go public, but only when the corporation has reached a certain stage in the life cycle. This stage in the life cycle is determined by the
corporations size and the industries market-to-book ratio (Pagano, Panetta, & Zingales, 1998). When a corporations size and market-to-book ratio are higher, the bigger the likelihood of a corporation to go public.
When a corporation is in need of financing it has to make a choice between private and public financing. Maksimovic and Pichler (2001) find that although private financing offers the advantage of lower-cost financing it also requires more disclosure of corporate information than public financing. The loss on confidential information may be beneficial to industry competitors, however, the newly obtained access to private financing creates an opportunity for the issuer to grow and improve the corporations
tradeoff between financing at a lower cost and more disclosure of corporation information. Despite the fact that there are plenty of reasons for a corporation to go public, in most cases the reason is to create a public market in which shareholders and founders can convert their shares to cash at a later date (Ritter & Welch, 2002).
If a corporation decides to go public its first step is to select an investment bank to perform the underwriter functions (Ellis, Michaely, & O'Hara, 1999). The underwriter functions include every aspect associated with going public and a lockup provision is usually included (Bradley, Bradford, & Yi, 2001). A lockup provision is an arrangement between the underwriter and the insiders of a corporation undergoing the IPO. In this arrangement the insiders agree not to sell their shares for a specified period (usually 180 days) after issue. A lockup provision is included because the insiders usually own a significant economic interest in the corporation. Thus by restricting sales the lockup provision ensures that the insiders maintain a significant economic interest after IPO, thereby ensuring that the interests of old and new shareholders are aligned.
There are two IPO anomies which are mainly researched within the literature. The first anomaly is the ‘underpricing’ anomaly. Ritter and Welch (2002) find that the average first day returns of 18.8% cannot be a simple market misjudgment valuation or risk premia. Neither can it be blamed on bearing systematic or liquidity risk, because these risks do not disappear after the first day. Thus the underpricing is most likely caused by the underwriter, not normal interplay between supply and demand. One of the main explanations within literature is asymmetric information. Asymmetric information is when the corporation is more informed than investors, but they find no evidence of this being the single cause of underpricing. The second anomaly is the long-run ‘underperformance’ of IPOs. Under the assumption of an efficient market, when investors become aware of the IPO underperformance, they should adjust their
investment strategies and the underperformance should not be existing anymore from a point-forward basis (Ritter J. R., 2011).
2.2. Possible explanations underperformance
Miller (1977) finds that because of the lack of shorting options on some financial instruments, the demand for a particular security will come from the most optimistic investors. Therefore, the initial purchases by the most optimistic investors will increase the price and when over time the variance of investor opinions decrease, the share price will converge towards the mean valuation. This upward pressure on the price from the
most optimistic investors results in initially overpriced shares and a long-run underperformance.
Schultz (2003) argues that underperformance is due to pseudo market timing. He finds that when IPOs perform well more corporations will go public as long as abnormal returns are positive. Therefore, a small fraction of IPOs issues before periods of positive abnormal returns and a large fraction issues before periods of negative abnormal returns. This large fraction of IPOs underperforms and if performance is being measured equal-weighted, on average IPOs underperform.
Brau, Couch and Sutton (2012) argue that an important motivation for the going-public decision is to create public shares for the acquisition of other corporations. They find that the IPOs that become acquirers within the first year after going public have significant negative excess return in the following years. They compare style-adjusted 3 year buy-and-hold returns and find that acquiring IPOs have a significantly lower return of 21.5% compared to non-acquiring IPOs.
According to Chang, Chung and Lin (2010) the underwriter reputation has an effect on the long-run abnormal returns. They find that prestigious underwriters will want to attain their reputation. Therefore, these underwriters will monitor and analyze any financial information on the IPOs. As a result the IPOs will exhibit less aggressive earnings management and perform better in the long-run, however, the long-run abnormal returns are still negative and therefore, it does not explain why IPOs underperform.
2.3. Empirical findings in the literature
Much research has been conducted on the long-run performance of IPOs, however, most of this research has been in the United-States. The first research on long-run
performance was conducted by Stoll and Curley. Stoll and Curley (1970) did their
research on a sample of 205 US-stocks and 425 S&P industrials in the period 1957-1963. After comparing the geometric average semiannual rate of return they find that IPOs underperform a portfolio of standard and poor’s industrial stocks. Ibbotson (1975) examines 120 IPOs which are issued in the period 1960 through 1969. He applies the returns across time and series model and the results suggest underperformance, but the standard deviation is big due to small sample size so equal performance cannot be rejected. Aggarwal and Rivoli (1990) find that for 1600 IPOs in the period 1980-1987 the initial abnormal returns are positive and significant. Despite the first day return they find that when an investor buys the stock at the end of the first day and holds the stock
for 250 trading days the stock underperforms the NASDAQ index by 13.73%. Their arguments for the long-run underperformance are unknown intrinsic value and speculative investors.
Ritter (1991) is one of the leading researchers of IPOs. He conducts his research on a sample of 1526 US IPOs and matching corporations in the period 1975-1984. He applies an ordinary least squares regression and performs an analysis on wealth relatives. He finds that for each dollar invested in IPOs an investor would have gained 83% return compared to matching corporations in the three years. Ritter and Welch (2002) research 6238 US IPOs in the period from 1980-2001 and find that comparing IPO returns to the broad market index results in a significantly lower return for IPOs. However, when they conduct a regression and compare IPOs to companies with similar size and book-to-market characteristics, the IPOs do slightly worse, but not significantly.
Purnanandan and Swaminathan (2004) research more than 2000 US IPOs from 1980 to 1997, split them up over three portfolios and match them on peer matching criteria. They analyze the long-run performance with multiple models and they find that IPOs underperform matching corporations in the 5 years after initial public offering. Zheng (2007) disagrees with the findings of Purmanandan and Swaminathan. He finds problems in their selection criteria and after adjusting these criteria he analyzes the buy-and-hold returns and conducts a Fama and French three factor regression. In his research he finds that his sample of 2493 US IPOs from 1980 to 1997 does not significantly underperform matching corporations. When an IPO is issued during the period 1994-2002 a so called ‘hot period’ the Greek IPOs tend to overperform for a longer period, but still underperform in the three years after issue (Thomadakis, Nounis, & Gounopoulos, 2012). From a German sample of 189 IPOs in the period from 1970 to 1993 Ljungqvist (1997) finds that the buy-and hold return of IPOs is 12% lower than the market index in the first three years after issue. Ritter (2011) conducts his research on a sample of 7314 US IPOs in the period from 1980 to 2008. When determining buy-and-hold returns he finds that if an IPO is bought at the closing price of the first trading day the three year return is 20.8%, but the buy-and-hold return of corporations with approximately the same market capitalization and book-to-market ratio is 27.9%. Therefore, holding and IPO for three years would have resulted in a 7.1% lower return than comparable corporations. Agathee, Sannassee and Brooks (2014) analyze 44 IPOs on the Mauritian stock exchange in the years from 1989 until 2010. They first determine the cumulative abnormal returns and find that when an investor buys an IPO at the end of the first trading day the three year abnormal return is -16.53% compared to the market. When they determine the buy-and-hold abnormal returns they find that IPOs
underperform for all three years, however, the result becomes insignificant after two years. Carter, Dark, Floros and Sapp (2011) implement three different tests for long-run abnormal returns on 6686 IPOs in the period 1981 to 2005. For the buy-and-hold return they find that IPOs significantly underperform in the first three years. Investing in an IPO would have resulted in a 9.9% lower return compared to a matching corporation. The second method they use for testing long-run the performance of IPOs is by
conducting a regression. They perform a regression on the excess return of IPOs against the factors excess market return, small-minus-big, high-minus-low, momentum,
liquidity, skewness and investments. They find an insignificant monthly negative alpha of 0.046%. Therefore, in this test they cannot reject IPO equal performance. The third method they apply is by conducting a Fama and French three factor regression. They conduct the regression for the entire period and three 6-10 years sub periods. They find that for the entire period the IPOs alpha of 0.09% per month is both economically and statistically insignificant. However, when analyzing the results of the period from 1981-1987 they find a significant negative alpha of 0.31% per month, for the period 1988-1997 they find insignificant results and for the period from 1998-2005 they find a significant positive monthly alpha of 0.54%. They are reluctant to conclude that the long-run underperformance has disappeared, but they do suggest that more research should be conducted on recent IPO performances to determine whether a shift has occurred in the IPO long-run performance.
2.4. Conclusion on the literature
For corporations the choice of going public should be made with caution. The tradeoff between information disclosure and lower cost financing has to be considered. There are explanations for underperformance, but not a single argument is widely accepted in this subject’s literature. The majority of research suggests that IPOs have a tendency to have a high first day return and underperform in the long run. However, under the assumption of an efficient market, investors should not hold IPOs if they significantly underperform in the long-run. Therefore, if investors are still holding IPOs either the market is inefficient or IPOs do not underperform the market.
3. Methodology and Data
This section will first discuss the methodology which will be used to test the hypothesis, then in the second part it will discussed what kind of data is used, how this data is acquired, the presentation and the description of the data and the last part will be the conclusion of this section.
3.1. Methodology
The sample contains 87 IPOs and 87 operating profit margin matching corporations. Two methods will be applied to determine whether IPOs underperform matching corporations in the long-run. First the IPOs and matching corporations will be compared by determining the buy-and-hold returns and buy-and-hold abnormal returns. Second the Fama and French three-factor regression will be applied.
To determine whether IPOs underperform there are two common procedures (Ritter J. R., 2011). First there is an event study. In an event study the IPOs are matched based on firm specific characteristics and the buy-and-hold return of the IPOs is
compared to the returns of matching corporations. The second procedure is to run a multifactor time-series regression with excess return as a dependent variable. Zheng (2007) applies both procedures in his research and is one of the few IPO researchers who finds that the IPO long-run performance is not significantly lower than the
performance of industry matching corporations. He finds matches for the IPOs based on comparable sales and EBIDA profit margin. In this thesis the operating profit margin is applied to find matching corporations for the IPOs. First the operation profit margin is calculated for IPOs and industry peers, second the operating profit margin of industry peers is deducted by the operating profit margin of the IPO and last the industry peer with the smallest absolute difference with IPO operating profit margin is selected. This method is not consistent with the paper of Zheng (2007), but matching corporations on what percentage of revenue is left after variable costs are deducted is the method used. The operating profit margin is calculated the following way
To determine whether the three year performance of IPOs is significantly less than the performance of matching corporations both an event study and a time series regression
will be conducted. For the event study the buy-and-hold return and the buy-and-hold abnormal return will be determined. The following formulas are used to calculate both returns: ∏ ∏ ∏
Where is the daily return on an investment measured by the following equation:
Where is the price of a security at a given time.
The buy-and-hold return will be calculated for both the IPOs and the matching corporations. The buy-and-hold return is an equation which results in the return an investor would have gained or lost if the investor had held the portfolio for the entire duration. For the buy-and-hold return a t-test will be conducted to determine whether IPOs minus matching corporations are performing significantly below zero. Consistent with Zheng (2007), the tests will be conducted for both equal-weighted portfolios and value-weighted portfolios and for multiple six month time periods as well as the entire time period. The results of the buy-and-hold equation have to be compared to the market in order to determine how the portfolio performed compared to the economy as a whole. The buy-and-hold abnormal returns equation calculates the buy-and-hold returns for the portfolio and deducts this by buy-and-hold returns for the market.
The buy-and-hold returns are a good way to compare corporations holding returns, but it does not account for risk characteristics (Zheng, 2007). Therefore, this measurement of the returns might be a wrong method. Fama and French (1992) argue that market returns do not provide accurate risk measures and therefore, include a size and a book-to-market ratio to the model. This results in the following model:
( )
Where is defined as the return on a portfolio, is the overnight EURIBOR interest rate, the daily risk-adjusted abnormal return as a percentage, ( ) is the daily excess return of the market portfolio, is the daily return on small corporations
minus the daily return on large corporations, is the daily return on high book-to-market stocks minus the daily return on low book-to-book-to-market returns, is an error term and , and are factor loadings.
To determine whether IPOs underperform matching corporations a set of regressions will be conducted. The regressions will be conducted on IPOs, matching corporations and IPOs minus matching corporations. The regressions will both be for equal-weighted portfolios and value-weighted portfolios and will be conducted for two different time periods. The periods will be after the first day till the end of the sixth month and from the start of the seventh month till the end of year three. The reason behind the separation of the regression in two time periods is because of the lockup period. The lockup might cause the stock to perform well during the first six month and , therefore, this effect has to be separated from the long-run effect. Alpha is daily risk-adjusted abnormal return and if alpha is positive the portfolio performed better than expected according to the model. The regression output will be used to determine whether the alphas are significantly different from zero and whether alpha is below zero for the regression on IPOs minus matching corporations. Is the market return factor and is expected to be significant in every regression on IPOs or matching corporations. The market return is the most important factor in the model and if is not significant the model being used is not a good approximation of expected stock return. The factor loading and are expected to be significant in some of the regressions.
3.2. Data and descriptive statistics
For IPOs to be included in the sample they have to meet a set of criteria to prevent a bias. Most of these criteria are also included in the paper by Purnanandam and Swaminathan (2004), although some will be adjusted for a European market. The following criteria will be included for an IPO to be in the sample:
The information regarding prices, market capitalization and operating profit margin should be available on Datastream. If a corporation does not have the information available it is not possible to analyze the data correctly.
The IPO should issue ordinary shares. If a company issued different types of shares or multiple types it could lead to misinterpretation of the results. If for example a corporation issued preference shares, the price changes of the shares might not reflect the performance of the corporation the way ordinary shares do.
The IPO has to be a non-financial corporation. This criteria is implemented in the research of Purnanandam and Swaminathan (2004) and thereafter by other
researchers.
They also implemented that the IPO has to have a minimum offer price of €5.00. The minimum price is to eliminate many of the smaller IPOs. The smaller IPOs are more likely to underperform in the long run and can, therefore, lead to a bias in the results.
The IPOs have to be assigned to an industry by Datastream. This criteria is included because if IPOs are not assigned to an industry it is not possible to assign them to a matching corporation.
In order to research the three year performance the IPOs have to be traded for at least three years after going public and on at least 65% of the trading days. This criteria also excludes bankrupt and acquired IPOs. It is not possible to calculate three year returns when there is insufficient data on the years.
The sample first acquired for this research consisted of 58 German IPOs and 76 French IPOs in the period from 2007 till the end of 2012. However, because some of these IPOs did not fulfill the criteria or did not have a matching corporation the sample is reduced to 44 German IPOs, 43 French IPOs and matching corporations for every IPO. The information on German IPOs is collected from Xetra1. Xetra is the largest German
trading venue and keeps track of the new issues. The information for the French IPOs is collected from Euronext2. Euornext keeps track of all the corporations that go public in
France, Holland and Belgium. The stock prices, market capitalizations and other company data are collected from a program called Datastream. Datastream is a global database which contains information on equities, stock market indices, currencies, company fundamentals, key economic indicators and fixed income securities in 175 countries.
Table 1 presents the market capitalization information per industry. The sample contains corporations from 25 different industries. Therefore, the tests on this sample will be conducted on a diversified portfolio. However, the average market capitalization differs a lot from industry to industry and this will have effects on the value-weighted portfolio results.
Table 1: Market capitalizations per industry(€m) , 2007-2012, Germany & France
Industry
Sample
size
Mean
Minimum
Median
Maximum
Alternative energy
2
973.92
53.57
973.92
1894.27
Automobiles & parts
3
239.13
91.71
151.25
474.43
Chemicals
3
1004.86
54.52
162.06
2797.99
Construction & materials
4
199.59
42.55
201.63
352.55
Electricity
1
85.72
85.72
85.72
85.72
Electronic & Electronical equipment
5
872.64
20.25
45.89
4183.55
Food producers
4
65.04
33.95
61.72
102.76
Gas, water & utilities
1
11.67
11.67
11.67
11.67
General industrials
1
1655
1655
1655
1655
General retailers
2
95.9
6.7
95.9
185.1
Healthcare equipment & services
8
199.93
6.97
32.75
1204.19
Industrial engineering
6
462.83
18.52
304.21
1328.55
Industrial transportation
5
991.96
41.29
114.96
4328.07
Leisure goods
2
121.68
32.16
121.68
211.2
Media
8
310.66
15.55
32.24
2006.1
Mining
1
19.55
19.55
19.55
19.55
Mobile telecommunications
1
1511.1
1511.1
1511.1
1511.1
Nonlife insurances
1
4648.08
4648.08
4648.08
4648.08
Oil & equipment services
1
52.83
52.83
52.83
52.83
Personal goods
6
111.77
4.65
107.11
226.35
Pharmaceuticals & Biotechnology
5
77.26
27.67
91.81
118.69
Software & computer services
8
94.92
19.36
40.31
300
Support services
3
1001.41
21.45
407.55
2575.23
Technology hardware & equipment
3
165.98
49.45
162.04
286.45
Travel & leisure
3
97.1
27.5
65.29
198.5
Table 2 presents the summary statistics of the IPOs and matching corporations for both countries. The mean IPO market capitalization for German companies is 587 million whereas the French companies have a mean market capitalization of 259. The result of the difference in mean market capitalizations will be a favorable value-weighting for German corporations compared to French corporations. The market capitalizations of IPOs are lower than the market capitalizations of matching corporations. This is an expected result because the matching corporations are selected on operating profit margin and not size.
Table 2: Market capitalizations, 2007-2012
Mean(€m) median(€m)
standard
deviation(€m)
min(€m) max(€m)
German market
Initial public offering
586.92
156.65
159.48
19.36
4648.08
Matching
3469.37
481.82
1354.49
14.47 53170.42
French market
Initial public offering
258.78
45.89
753.38
4.65
4183.55
Matching
2862.11
65.4
7982.50
0.9 36347.06
Total
Initial public offering
424.73
85.72
927.66
4.65
4648.08
Matching
3169.23
149.16
8454.83
0.9 53170.42
Source: Datastream
Table 3 presents the amount of IPOs for the years within the sample for both countries. The year 2007 is the biggest IPO year within the sample and can be considered a “hot period”. In the year 2008 the global financial crisis emerged and because of the relative importance of 2007 this could affect the outcome of the research.
Table 3: IPOs per year
Year
2007
2008
2009 2010
2011
2012
total
German market
20
3
1
4
10
6
44
French market
18
0
2
2
12
9
43
For the German market proxy the DAX30 is used. The DAX30 measures the performance of the 30 largest German corporations. The corporations in the DAX30 give an estimate of how the German economy is performing as a whole. For the French market proxy the CAC40 is used. The CAC40 measures the performance of the 40 most important French traded corporations and is therefore, a good estimate of the French market.
The risk-free interest rate is acquired from the website of the Bundesbank3. The
Bundesbank supplies the EURIBOR interest rate which is the rate for which banks can borrow mutually.
To determine the German Small-Minus-Big (SMB) and the High-Minus-Low (HML) factors the DAX30 and MDAX are used. For each corporation the prices from the beginning of 2007 till the end of 2015 have to be available. There also needs to be information available for the market capitalization and book value of equity from 2006 up until 2014. Every year six portfolios are constructed based on the market
capitalization and book-to-market value. For the French factors the same procedure is applied, but on the CAC40 instead of German market indexes. Table 4 is a representation of how the portfolios are determined according to Fama and French (1992) and from the French4 website.
Table 4: Fama and French 6 portfolios
Median ME
70
thBE/ME percentile
30
thBE/ME percentile
Small Value Big Value
Small Neutral Big Neutral
Small Growth Big Growth
Where BE is the Book-value of equity and ME is the Market-value of equity
Source: French website
Small-Minus-Big (SMB) is calculated by taking the average of the three small portfolios minus the average of the three big portfolios and High-Minus-Low (HML) is calculated by taking the average of the two value portfolios minus the average of the two growth portfolios.
3 Risk-free interest rate from:
https://www.bundesbank.de/Navigation/EN/Statistics/Time_series_databases/Money_and_capi tal_markets/money_and_capital_markets_list_node.html?listId=www_s510_gmt_neu
4 French website:
4. Empirical Results
This section will first discuss the buy-and-hold returns and buy-and-hold abnormal returns, the second part will discuss the results of the Fama and French three factor regression and the final part will be the conclusion of the results.
4.1. Buy-and-hold returns and buy-and-hold abnormal returns
The buy-and-hold returns (BHR) and buy-and-hold abnormal returns (BHAR) are presented in table 5. For the equal weighted first six months the BHRs are negative for both IPOs and matching corporations. Therefore, holding either an IPO or a matching corporation would have resulted in a negative return, however, the BHARs are less negative because the market has a negative return. When comparing IPOs with matching corporations, the IPOs would have earned 1.96% less than matching
corporations, but the result is insignificant. The value-weighted returns are similar, but the matching corporations perform better than the market.
Table 5: Buy-and-hold and Buy-and-hold abnormal returns (Germany & France)
period 0- 6mth 6- 12mth 12-18mth 18- 24mth 24-30mth 30- 36mth 0- 36mth Equal-weighted % return BHR IPO -6.95 -6.75 -18.02 5.29 22.02 6.82 -9.27 BHR Matching -6.41 -1.59 -8.96 4.67 15.48 10.59 28.29 BHAR IPO -3.18 -2.81 -15.00 5.01 8.88 -0.12 -16.55 BHAR Matching -2.64 2.35 -5.95 4.39 2.34 3.65 21.01 IPO-Matching -0.54 -5.16 -9.06 0.62 6.54 -3.77 -37.56 p-value 0.46 0.06 0.00 0.54 0.87 0.19 0.07 period 0- 6mth 6-12mth 12-18mth 18-24mth 24-30mth 30-36mth 0- 36mth Value-weighted % return BHR IPO -4.74 0.22 -4.54 1.35 21.42 8.11 16.82 BHR Matching -2.51 -1.70 -7.75 -9.39 19.82 8.11 3.45 BHAR IPO -0.97 4.16 -1.52 1.07 8.28 1.17 9.54 BHAR Matching 1.26 2.24 -4.74 -9.67 6.68 1.16 -3.83 IPO-Matching -2.23 1.92 3.22 10.74 1.60 0.00 13.36 p-value 0.21 0.83 0.82 1.00 0.70 0.50 0.83
For the equal-weighted portfolio the periods 6-12 months and 12-18 months have similar results. The IPOs and matching corporations both have negative BHR, but
the return of matching corporations is less negative. Investing in IPOs would have resulted in a loss of 5.16% at a 90% significance level and 9.06% at a 99% significance level compared to matching corporations. The value-weighted IPO BHRs are 6.97% and 13.48% higher in the same period. The matching corporations returns decreased in the 6-12 month period and increased in the 12-18 month period. For both periods investing in IPOs would have resulted in a higher return than investing in matching corporations, but the results are insignificant.
The 18-24 months period shows for the equal weighted portfolio that matching corporations perform slightly better but not significantly, however, when the
performances are value weighted the IPOs perform 10.74% better and significant at 99% level. This means that IPOs perform better than matching corporations in this value-weighted time period.
For the period 24-30 months the BHRs are more than 10% positive for both equal-weighted and value-weighted portfolios. Despite the positive BHRs, the BHARs are lower and this means that the market had high returns in this time period. The difference between IPOs and matching corporations are insignificant for both weights.
For the period 30-36 months IPOs underperform in the equal-weighted portfolio with 3.77%, but not significantly and for the value-weighted portfolio the difference between IPOs and matching corporations is 0.
For the entire period when comparing IPOs to matching corporations the IPOs underperformed with 37.56% in the equal-weighted portfolio. This indicates that when investing in an IPO versus a matching corporation, and holding it for three years, the investment in an IPO would have resulted in a negative return of 37.56% This result is significant at a 90% level . When evaluating the value-weighted return the results are positive for IPOs. Investing in an IPO versus a matching corporation would have resulted in a positive three year return of 13.36%, but this is not significant.
4.2. Three factor regression results
Table 6 presents the Fama & French regression for the period from the end of the first trading day until the end of 6 months (Fama & French, 1992) . In both the equal-weighted and value-equal-weighted portfolio the market excess return is significant in most regressions. Despite the significant market excess return the size and book-to-market factors are insignificant for the most part. This indicates that for the first six months these factors do not explain the return of the corporations. For the equal-weighted portfolios both the alpha of IPOs and matching corporations are significantly below
zero. This suggests that investing in either IPOs or matching corporations would have led to a significant negative abnormal trading day return of 0.1% and 0.05%. However, when comparing IPOs to matching corporations, the alpha is negative, but not
significant. For the value-weighted return the alphas of both the IPOs and matching corporations are insignificantly negative and when IPOs are compared to matching corporations the alpha is slightly negative, but again insignificant.
Table 6: Three factor regression for first six months (Germany & France)
Equal-weighted portfolio IPO -0.0010** 0.3974*** -0.0234 -0.0209 Matching -0.0005** 0.5807*** 0.0752* 0.0554** IPO-matching -0.0005 -0.1776*** -0.1012* -0.0744*** value-weighted portfolio IPO -0.0003 0.7650*** 0.0680 -0.0823 Matching -0.0001 0.8289*** -0.0009 0.0313 IPO-matching -0.0002 -0.0491 0.0676 -0.1077
Where *,** and *** indicate significance at 10%, 5% and 1% statistical significance. In table 7 the Fama and French (1992) three factor regression results are presented for the period from the beginning of the seventh month to the end of three years. The factor market excess return is statistically significant in all the regression results, indicating that this factor is a good predictor for returns. The size and book-to-market factors are statistically significant for the equal-weighted return, but not for the value-weighted return. The alpha of IPOs in the equal-weighted portfolio is significant with a negative abnormal trading day return of 0.03%. The matching corporations have an insignificant positive alpha of 0.01% per trading day. When comparing IPOs to matching
corporations the result is a negative abnormal trading day return of 0.04%. This indicates that holding IPOs versus matching corporations would have resulted in a negative abnormal trading day return of 0.04% per trading day. For the value-weighted portfolio the results are similar, but higher. Holding an IPO would have resulted in a significant negative abnormal trading day return of 0.1% per trading day. Whereas holding a portfolio of matching corporations would have resulted in an insignificant negative abnormal trading day return of 0.03%. Therefore, holding IPOs compared to matching corporations resulted in a significant negative abnormal trading day return of 0.08%.
Table 7: Three factor regression for the beginning of month seven till the end of three years (Germany & France)
Equal-weighted portfolio IPO -0.0003** 0.3594*** 0.0394** 0.0298*** Matching 0.0001 0.4542*** 0.0325** 0.0183*** IPO-matching -0.0004** -0.0948*** 0.0069 0.0116 value-weighted portfolio IPO -0.0010*** 0.5272*** 0.0367 0.0452 Matching -0.0003 0.4300*** 0.0299 0.0123 IPO-matching -0.0008** 0.0973*** 0.0067 0.0330
Where *,** and *** indicate significance at 10%, 5% and 1% statistical significance.
4.3. Conclusion on the results
For the buy-and-hold returns the results are inconclusive. IPOs underperform matching corporations in the equal-weighted portfolio and IPOs perform better than matching corporations in the value-weighted portfolio, but the majority of results are
insignificant. For the three-factor regression the first six months indicate that IPOs do not significantly underperform matching corporations. However, when analyzing the returns of the 7 to 36 months period the IPOs significantly underperform matching corporations for both the equal-weighted and value-weighted portfolio.
5. Conclusion and discussion
Using a sample of 87 German and French IPOs, the long-run underperformance has been analyzed in a period in which Europe experienced the consequences of the global
financial crisis and the European sovereign debt crisis. Therefore, the hypothesis is tested to determine whether IPOs underperform matching corporations in time of economic distress.
Aggarwal and Rivoli (1990) find that their sample of 1600 IPOs underperforms the market index by 13.73% per year. The buy-and-hold abnormal returns in this sample for the sum of the periods 0-6 and 6-12 months are higher for both the equal-weighted portfolio and the value-equal-weighted portfolio. This indicates that the
research of Aggarwal and Rivoli. Their research was conducted in a period of good market performance, whereas for thesis the market performance in the first year is negative. Therefore, this result suggests that IPOs still underperform the first year, but less in time of financial distress.
Ritter (1991) researches 1526 IPOs and finds that holding a portfolio of IPOs would have resulted in 83% return compared to matching corporations in the first three years. In this thesis the buy-and-hold returns compared to matching corporation varies with the type of portfolio. The equal-weighted portfolio would have resulted in a negative return of 37.56% at a 90% significance level, whereas the value-weighted portfolio would have resulted in a return of 13.36%, but the result is insignificant. Therefore, the three year BHRs suggest IPO underperformance, but the results are inconclusive.
Purnanandam and Swaminathan (2004) conduct their research on more than 2000 US IPOs and compare them to matching corporations. Their results indicate that IPOs have a positive six month abnormal return, but significantly underperform in the long-run. Zheng (2007) disagrees with their findings and conducts his research on a sample of 2493 IPOs and compares them to matching corporations. For the three-factor regression he finds that IPOs significantly outperform the market in the first six months and do not significantly underperform the market in the next 4.5 years. The results for the first six months in this thesis vary between the equal-weighted portfolio and the value-weighted portfolio. The equal-weighted portfolio results indicate that IPOs have a significant negative abnormal trading day return, which is opposite to the findings of Zheng (2007) and Purnanandam and Swaminathan (2004). Although the equal-weighted returns are significant, the value-equal-weighted portfolio results suggest underperformance, but the results are insignificant. For the 6-36 month period this thesis finds significant underperformance for both portfolios. The long-run results are in line with the findings of Purnanandam and Swaminathan (2004), but not with the research of Zheng (2007). The difference in results can be explained by the period in which the IPOs where issued. He researched a sample from 1980 to 1998, whereas this research was conducted on a sample from 2007-2012. For the period 2007-2012 both the global financial crisis and European sovereign debt crisis could have influenced the performance. However, because this thesis analyzes a single period, and there is no previous research on IPO performance during times of economic distress, it is unclear if the period had a significant influence on performance. Despite the uncertainty regarding the influence on the IPO performance, the evidence suggests that in times of economic distress IPOs underperform in the long-run.
Carter, Dark, Floros and Sapp (2011) conduct their research on a sample of 6686 IPOs in the period from 1981 to 2005. They perform multiple Fama and French three-factor regressions on different time periods and find that the long-run performance varies over time. Their results indicate that the three year post issue abnormal performance in the period from 1998 to 2005 is significantly positive. Their overall findings suggest that the underperformance of IPOs has disappeared, however, they are reluctant to draw that conclusion. The results in this thesis contradict the findings of Carter, Dark, Floros and Sapp (2011). The regression results from the Fama and French three-factor regression indicate that IPOs underperform matching corporations in the long-run, and therefore, the results suggest that the underperformance has not disappeared over the years. However, it is important to note that the sample period in this thesis is in times of economic distress, whereas their research period was for the most part in times of economic growth. Therefore, IPO long-run underperformance could still have disappeared, but does seem to exist in times of economic distress.
The sample size in this research is relatively small compared to most IPO long-run performance research. This does not disempower the findings in this research, but a bigger sample could result in more statistical power. Moreover long-run event studies are subject to statistical difficulties that weaken the results (Brav, 2000). The difficulties with long-run event studies are non-normality and cross-sectional correlation of
abnormal returns. Despite these difficulties, in this thesis not only buy-and-hold returns, but also Fama & French regressions for both equal-weighted and value-weighted
portfolios have been applied to minimize the problems associated with long-run performance research.
Although IPO long-run performance is dependent on the sample period, the results in this thesis suggest that IPOs underperform matching corporations in times of economic distress. Therefore, from the investor’s point of view an investment in IPOs is a bad investment in times of economic distress, but because of inconsistencies the results should be interpreted with caution.
In the literature of IPOs multiple methods are applied to determine the long-run performance, but not a single method is widely accepted. Therefore, future research should put more emphasis on what methods to use.
For IPOs multiple researchers agree that there exists time variation in the long-run performances, however, there is little literature on this subject. Therefore the results in this thesis surrounding IPO long-run performance during times of economic distress are suggestive but not conclusive. Thus, time-variation in long-run IPO
performance could be an interesting topic for future research, especially performance variation between times of economic growth and times of economic distress.
