IPO underpricing and performance around the
financial crisis
Amsterdam Business School Name Maarten Meiners Student number 10179585 Program Economics & Business Specialization Finance & Organization Number of ECTS 12 Supervisor Dr. Rafael Perez Ribas Target completion 16/06/2016 Abstract This thesis tests whether the financial downturns affect underpricing of initial public offerings (IPO) and their returns after 12 months. To do so, I investigate the periods before, during and after the financial crisis initiated in 2008. According to previous research a lower level of underpricing is expected during economic downturns. To test the hypothesis, I estimate three models that calculate the influences of sales and debt on the level of underpricing and the holding period return (HPR). This paper concludes that the level of underpricing is indeed lower during the financial crisis. In normal periods, the underpricing is a function of sales. But during the crisis, this relationship disappears. The HPR is higher after the crisis period. The HPR is lower in case of high underpricing. In normal periods, HPR is a function of company’s sales and debt. But during the crisis period, this relationship disappears as well. Keywords: IPO, Underpricing, Underperformance, Return, Financial CrisisTable of Contents Statement of Originality ... 1 1. Introduction ... 2 2. Literature review ... 3 3. Data ... 5 4. Empirical method ... 7 5. Results ... 8 6. Conclusion ... 18 Reference list ... 19 Statement of Originality This document is written by Student Maarten Meiners 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.
1. Introduction When firms go public, it is common that the first day closing price is higher then the initial offer price (Ritter and Welch, 2002). This phenomenon is called the underpricing of IPOs. IPO underpricing has changed over time. Loughran and Ritter (2004) suggest that this has to do with the time of the economic state of the market. The underpricing of IPOs was the highest in the years 1999-2000 during the Internet bubble and went down afterwards during the dotcom crisis in 2001-2003. I want to investigate what happened to the underpricing of IPOs during the financial crisis around 2008. On the other hand, in the long run IPOs are mostly underperforming compared to market return (Brav and Gompers, 1997). Here I want to investigate the underperforming with the 1-year holding period return (HPR) to see if you should invest in IPOs during financial downturns. The combination of this information resulted in the following research questions: What are the costs of IPOs in financial downturns? What are the implications of investing in IPOs during financial downturns? How are the underpricing and returns of IPOs related? This thesis investigates what happens before, during and after financial downturns. As a result of the research of Loughran and Ritter (2004) about the economic downturn in 2001-2003, we predict to see the same influence of the economic downturn around 2008 on the IPO underpricing. So, the main hypothesis is a lower level of underpricing during the crisis period. The hypothesis is tested with an ordinary least squares (OLS) regression. The first model in this paper estimates the influence of company’s sales and debt on underpricing during the different time periods around the 2008 financial crisis. We find out that the underpricing is indeed lower during the crisis period. The normal influences of company’s sales and debt before and after the crisis period are not influencing the level of underpricing during the crisis period. Besides the underpricing phenomenon, we study underperformance on the 1-year HPR. As a result of the research of Brav and Gompers (1997), my second
order hypothesis is that the underperformance of the IPO stock is negatively influenced by the level of IPO underpricing. We expect the 1-year HPR to be higher after the crisis period, as a restoration of confidence after the crisis period, will drive up the returns. We find a negative influence of the level of underpricing on the HPR. This means that a higher level of underpricing will result in a lower HPR. The normal influences of company’s sales and debt before and after the crisis period are not influencing the HPR during the crisis period. In the following chapter the existing literature will be discussed. The subject will be introduced with a broader view about firms going public. Then we will dig deeper into the literature and see what is important for this research. This research collected IPO data from 2003 till 2014 in Zephyr and Datastream, described in the data section. Following section will be the empirical method: the level of underpricing and HPR will be described with an OLS regression on the different periods before, during and after the crisis. Then the results will be explained before the discussion and conclusion. 2. Literature review Firms go public because of many reasons. Ritter and Welch (2002) suggest that the main reason is a desire to raise equity capital for the firm. Another reason is that a firm creates a public market in which the founders and other investors can convert some of their wealth into cash. A minor non-financial reason contains increased publicity. Ransley (1984) suggests that the prospects for growth by acquisition and the available funds for expansion are the main reasons for going public. Röell (1996) concludes that the main reasons are an informative stock price, more liquidity and more finance opportunities. These are all positive and strengthening factors for the firm to go public. Contrarily, the complexity of the public markets and the costs of going public are big disadvantages of IPOs. The costs of going public can be split up into direct and indirect costs. The direct costs are mostly banking fees of the underwriters helping the firm going public. The indirect costs are related to underpricing. These costs are a big offset for firms to go public and can be as high as 21.22%
according to Ritter (1987). The questions arise whether an IPO is the best way for a company to raise capital and why going public is better in some situations or times than in others. In the field of IPO underpricing studies, Ritter and Loughran (2002) made a major tribute to a clear description of underpricing. They explain that the underpricing is calculated by the first day closing price minus the offer price. This paper uses their explanation to calculate underpricing. There are multiple explanations for underpricing. Lucas and McDonald (1990) developed a model about imperfect information. They think that better-informed investors sometimes know more about the firm going public then the underwriter or others. The underwriter is the investment bank, which helps the firm to go public. Therefore there will be a difference in valuation, which causes the underpricing. Ritter and Welch (2002) support these findings, but add the fact that the underpricing is related to favorable market conditions. They find out that the market conditions are the most important factor for timing of IPOs. The market timing is more important then the life cycle stage of the firm itself. As an example, Ritter and Welch (2002) found high underpricing during the internet bubble in 2000. Another paper of Loughran and Ritter (2004) investigates the change of underpricing over time. They conclude that underpricing is high during a ‘hot issue’ market period. More firms will use a bubble period to go public. This is another reason why this paper expects a higher underpricing before the financial crisis as this was a bubble period. I want to investigate what happens to the cost of going public during financial downturns. The cost of going public, the underpricing, is changing because of the market cycles, in this case the financial crisis. The first hypothesis of this paper is based on these market cycle findings. Lower underpricing is expected during the financial crisis. As we see the underpricing changing during the financial downturns, we wonder what happens when we invest in IPOs during the crisis. The 12 months returns are based on the holding period return of the IPO stock. Brav and Gompers (1997) suggest that some IPO stock underperforms compared to the market.
Ritter (1991) finds out that the 3-year buy and hold return after the IPO is underperforming compared to similar matched firms, even across different industries. The biggest underperforming comes from an IPO market where investors are overoptimistic about the earnings potential of the firm that went public. After the hot issue market phenomenon, the IPO stock underperforms. So the second hypothesis of this paper says that there is an underperformance expected after the IPO. After the crisis this underperformance is expected to be lower. The second regression consists the 1-year holding period return of IPO stock to research this phenomenon. 3. Data In this thesis, data come from different sources. Database Zephyr of Bureau van Dijk has been used to select the IPOs. Zephyr has a lot of information for mergers & acquisitions and IPOs; therefore this was the best database to start selecting the IPOs. Datastream was used for additional information like company data and equities. The ISIN number was taken from Zephyr to find all company’s in Datastream and find additional company information and stock data for the different regressions.
Data cover the period between January 1st 2003 and December 31st 2014. The starting period has been determined as a normal economic period after the dotcom crisis in 2000/2001. The analysis continues up to December 31st 2014 because we need 1-year HPR data till December 31st 2015. Therefore this paper has IPO information before, during and after the 2008 financial crisis. The debt crisis in the American housing market was one of the main reasons for the financial crisis. This paper selects January 1st 2008 till December 31st 2011 as financial crisis period for the different datasets. The selected IPOs are from the US stock market, the NASDAQ and New York Stock Exchange (NYSE) national markets. The complete period contains 1492 IPOs. The required data were not available for every company; this results in a smaller number of IPOs in the sample. This gives 491 observations before the
crisis, 242 during the crisis and 496 after the crisis, in total 1235 observations. To introduce the different variables needed for this research, Table 1 is presented. Table 1: List of Variables Variable Description Underpr Log of first day closing price – log of offer price Sales Log of sales; represents gross sales and other operating revenue less discounts, returns and allowances, before the IPO date Debt (Long Term Debt + Short Term Debt & Current Portion of Long Term Debt) / Common Equity HPR 1-year Holding Period Return = log of closing price 1 year after IPO- log of first day closing price The first dependent variable is the underpricing. Underpricing is measured by the first day closing price minus the offer price of the IPO. 𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔 = ln 𝑓𝑖𝑟𝑠𝑡 𝑑𝑎𝑦 𝑐𝑙𝑜𝑠𝑖𝑛𝑔 𝑝𝑟𝑖𝑐𝑒 − ln(𝑜𝑓𝑓𝑒𝑟 𝑝𝑟𝑖𝑐𝑒) In most studies on IPOs the first day closing price is higher then the offer price. This phenomenon is therefore called underpricing of IPO. The second dependent variable is the 1-year Holding Period Return (HPR). 𝐻𝑃𝑅 = ln 𝑐𝑙𝑜𝑠𝑖𝑛𝑔 𝑝𝑟𝑖𝑐𝑒 1 𝑦𝑒𝑎𝑟 𝑎𝑓𝑡𝑒𝑟 𝐼𝑃𝑂 − ln(𝑓𝑖𝑟𝑠𝑡 𝑑𝑎𝑦 𝑐𝑙𝑜𝑠𝑖𝑛𝑔 𝑝𝑟𝑖𝑐𝑒) In the second regression the buy and hold return of the IPO stock is measured and compared to the underpricing. In the third regression the HPR is compared to the sales and debt of the IPO firm. For the first regression there are two independent variables, sales and debt. The sales represents gross sales and other operating revenue less discounts, returns and allowances on the IPO date: ln(sales). Debt is calculated by (long term debt + short term debt & current portion of long term debt) / common equity. This value is adjusted for outliers by winsor with 1%. The descriptive statistics of these variables are presented in Table 2.
Table 2: Descriptive Statistics
2003-2007 2008-2011 2012-2014 Total period Mean Std. Dev. Mean Std. Dev. Mean Std. Dev. Mean Std. Dev. Underpr 0.212 0.675 0.125 0.417 0.138 0.361 0.165 0.519 Sales 11.53 1.942 11.77 2.112 11.82 2.207 11.68 2.073 Debt 0.657 2.850 0.798 2.168 0.777 3.186 0.727 2.870 HPR -0.065 0.565 -0.059 0.440 0.031 0.532 -0.026 0.531 # obs 491 242 496 1235 There is a significant difference between the underpricing in 2003-2007 and 2008-2011. The mean of underpricing in 2003-2007 was 0.212 with a standard error of 0.675. However the underpricing coefficient in 2008-2011 was 0.125 with a standard error of 0.417. This is in line with the first hypothesis. The mean of the variable sales goes up over the different periods. The mean of the variable debt goes up during the crisis period. Besides that, the mean of the HPR is only positive in the period after the crisis. During the research multicollinearity is investigated as well. The highest correlation was between sales and underpricing consisting 0.1541. Also the highest variance inflation factor (vif) was sales with a value of 1.14. There is no significant correlation between the independent variables. 4. Empirical method To test my first hypothesis, I estimate the following model. This model calculates the influence of company’s sales and debt on the level of underpricing. 𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔! = 𝛼!+ 𝛽!𝑆𝑎𝑙𝑒𝑠! + 𝛽!𝐷𝑒𝑏𝑡!+ 𝜀!
where 𝛼!represents year-specific effects and 𝜀! is the error term. Following my hypothesis the research expects lower alpha and betas during the crisis because of less influence of the dependent variables on the level of underpricing. The second model that tries to answer the research question is the influence of underpricing and the year of the IPO on the 1-year holding period return. 𝐻𝑃𝑅! = 𝛼!+ 𝛽!𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔! + 𝜀!
where 𝛼!represents year-specific effects and 𝜀! is the error term. Following the second hypothesis we expect a higher alpha and beta after the financial crisis because we expect more influence of underpricing on the HPR. A third model is constructed to test the hypothesis. This model calculates the influence of company’s sales and debt on the 1-year holding period return. 𝐻𝑃𝑅! = 𝛼!+ 𝛽!𝑆𝑎𝑙𝑒𝑠!+ 𝛽!𝐷𝑒𝑏𝑡!+ 𝜀!
where 𝛼!represents year-specific effects and 𝜀! is the error term. Following the hypothesis we expect lower alpha and betas during the crisis because of less influence of the dependent variables on the HPR. 5. Results Results of the regression will be presented and analyzed. Besides that, some additional figures will be presented. Figure 1 is a distribution plot and shows the level of underpricing over the different years. It shows a lower median for the level of underpricing during the crisis period, years 2008-2011. The data shows low means especially for 2008 and 2009 compared to other years. Also the variances are low for 2008 and 2009, this means smaller differences between the single values within the year.
Figure 1: The Underpricing Changing over Time This distribution plot shows the level of underpricing. The boxes represent the 25th percentile, median and 75th percentile. The lines represent the upper and lower adjacent values, outside values are not displayed. The first regression is estimated for three different time periods: before, during and after the financial crisis; and one of the complete time period. First we see a relationship between sales and underpricing but no relationship with debt. The influence of sales has a negative and significant effect in the period before the financial crisis. There is no significant effect during and after the crisis. This means that there is enough evidence to say that there is a negative effect of sales on the level of underpricing before the crisis. So a higher amount of sales means a lower level of underpricing. The influence of sales and debt on the level of underpricing are presented in Table 3 as coefficients. The student’s t-test and the p-value give information about the significance of these coefficients. Financial crisis -. 4 -. 2 0 .2 .4 .6 U n d e rp ri ci n g 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Table 3: Regression Estimates Influence on Underpricing
2003-2007 2008-2011 2012-2014 Total period Name
variable Coef. (P>|t|) t Coef. (P>|t|) t Coef. (P>|t|) t Coef. (P>|t|) t Sales -.0639 -2.63*** (0.009) -.0037 -0.27 (0.791) -.0009 (0.862) -0.17 -.0264 -2.87(0.004) *** Debt .0142 1.19 (0.236) -.0079 -1.20 (0.232) -.0054 -1.15 (0.251) -.0005 -0.09 (0.925) cons .7716 2.68*** (0.008) .1853 1.24 (0.217) .1397 (0.036) 2.10** .3461 2.90*** (0.004) 2004 .1716 1.91* (0.057) .2166 2.52** (0.012) 2005 .2885 2.34** (0.020) .2869 2.38** (0.017) 2006 .1543 1.89* (0.060) .1428 1.95* (0.051) 2007 .1054 1.34 (0.182) .1151 (0.089) 1.70* 2008 .1861 1.48 (0.140) 2009 -.1290 -1.52 (0.130) -.0045 (0.947) -0.07 2010 .0557 0.61 (0.542) .1437 (0.041) 2.04** 2011 -.0068 -0.09 (0.931) .0953 (0.157) 1.42 2012 .1019 1.68* (0.094) 2013 .0326 0.99 (0.320) .1169 (0.042) 2.04** 2014 .0152 0.46 (0.645) .0803 (0.179) 1.34 # obs 430 213 420 1063 Note: *, **, and *** to indicate 10%, 5%, and 1% significance levels. Robust standard errors are used for this regression.
Figure 2 shows the effect of sales on underpricing from Table 3. We see a declining line of 2007, which represents the significant negative effect. The 2008 line is almost horizontal; this means there is no effect from sales on underpricing during the crisis. Figure 2: The Effect of Sales on Underpricing Before and During Crisis Figure 3 shows the effect of debt on underpricing from Table 3. There is no significant effect of debt on the level of underpricing in any of the periods; this means there is not enough evidence to support the results. There is a small positive effect before the crisis in contrast to a small negative effect during the crisis. It means with a higher level of debt, a firm going public has more underpricing before the crisis but lower underpricing during the crisis. 2007 2008 -. 4 -. 2 0 .2 .4 .6 U n d e rp ri ci n g 5 10 15 20 log Sales
Figure 3: The Effect of Debt on Underpricing Before and During Crisis The second regression is estimated for the same time periods. Underpricing has a negative effect on the 1-year holding period return. This is in line with second hypothesis. The effect is significant during and after the crisis period. This negative effect gets bigger over time. This means that a higher level of underpricing results in a lower HPR. The influence of underpricing on the HPR is presented as coefficients with the corresponding t-test and p-value in Table 4. 2007 2008 0 .5 1 U n d e rp ri ci n g -2000 -1000 0 1000 2000 Debt
Table 4: Regression Estimates Influence from Underpricing on HPR
2003-2007 2008-2011 2012-2014 Total period Name
variable Coef. (P>|t|) t Coef. (P>|t|) t Coef. (P>|t|) t Coef. (P>|t|) t Underpr -.0585 -1.71* (0.088) -.2418 -2.74*** (0.007) -.3654 -3.36*** (0.001) -.1598 -4.70(0.000) *** cons .0681 0.93 (0.354) -.2221 -2.29(0.023) ** .2504 5.19*** (0.000) .0767 1.04 (0.298) 2004 -.0064 -0.07 (0.941) .0006 0.01 (0.995) 2005 .0174 0.19 (0.846) .0512 0.56 (0.572) 2006 .0085 0.08 (0.933) .0346 (0.714) 0.37 2007 -.4727 -5.02*** (0.000) -.4549 -5.07(0.000) *** 2008 -.3534 -2.93(0.004) *** 2009 .2145 1.88* (0.061) -.0445 (0.638) -0.47 2010 .2731 2.41** (0.017) -.0656 (0.445) -0.76 2011 .1435 1.30 (0.197) -.1478 (0.092) -1.68* 2012 .0955 1.13 (0.257) 2013 -.1744 -3.00*** (0.003) -.0378 (0.640) -0.47 2014 -.2672 -4.01*** (0.000) -.1007 (0.245) -1.16 # obs 491 242 501 1234 Note: *, **, and *** to indicate 10%, 5%, and 1% significance levels. Robust standard errors are used for this regression.
Figure 4 shows the effect of underpricing on the HPR from Table 4. We see a declining line that gets steeper over time. This means the effect of underpricing on the HPR gets bigger. Figure 4: The Effect of Underpricing on HPR Before, During and After Crisis The third regression is estimated for the same time periods. Both sales and debt have an effect on the HPR. Sales has a positive effect on the HPR. This effect is bigger and significant before and after the financial crisis compared to the period during the crisis. This means that more sales will result in a higher HPR in the periods before and after the crisis. There is no significant effect from sales on the HPR during the crisis. Debt has a positive effect on the HPR. The effect is significant in the period before the crisis. During and after the crisis, debt has no significant effect on the HPR. This means that before the crisis, higher level of debt means a higher HPR. The influence of sales and debt on the HPR are presented as coefficients with the corresponding t-test and p-value in Table 5. 2013 2008 2007 -1 .5 -1 -. 5 0 .5 1 H PR -2 0 2 4 6 Underpricing
Table 5: Regression Estimates Influence from on HPR
2003-2007 2008-2011 2012-2014 Total period Name
variable Coef. (P>|t|) t Coef. (P>|t|) t Coef. (P>|t|) t Coef. (P>|t|) t Sales .0478 3.04*** (0.003) .0247 1.30 (0.195) .0409 2.82(0.005) *** .0360 4.01*** (0.000) Debt .0268 3.00*** (0.003) -.0000 -0.00 (0.998) .0092 (0.332) 0.97 -.0105 (0.074) 1.79* cons -.4528 -2.37** (0.018) -.5753 -2.36** (0.019) -.2836 -1.52 (0.129) -.3185 -2.58(0.010) ** 2004 -.0480 -0.53 (0.594) -.0414 (0.636) -0.47 2005 -.0486 -0.51 (0.611) -.0176 (0.849) -0.19 2006 -.0715 -0.66 (0.508) -.0455 -0.46 (0.646) 2007 -.5610 -5.31*** (0.000) -.5463 -5.60*** (0.000) 2008 -.4532 -3.38*** (0.001) 2009 .2685 2.11** (0.037) -.0901 -0.87 (0.386) 2010 .3220 2.60** (0.010) -.1115 -1.22 (0.224) 2011 .1654 1.32 (0.190) -.2031 -2.14(0.033) ** 2012 .0262 0.29 (0.771) 2013 -.1856 -2.87*** (0.004) -.0904 -1.06 (0.290) 2014 -.3073 -4.16*** (0.000) -.1842 -2.01** (0.045) # obs 454 220 426 1100 Note: *, **, and *** to indicate 10%, 5%, and 1% significance levels. Robust standard errors are used for this regression.
Figure 5 shows the effect of sales on the HPR from Table 5. There is a steeper upward sloping line before and after the crisis. This means that the effect of sales on the HPR is bigger during these periods. Figure 5: The Effect of Sales on HPR Before, During and After Crisis Figure 6 shows the effect of debt on the HPR from Table 5. There is a steep upward sloping line before the crisis. There is a significant positive effect of debt on the HPR. The horizontal line mean there is no effect during the crisis. 2007 2008 2013 -1 -. 5 0 .5 H PR 5 10 15 20 log Sales
Figure 6: The Effect of Debt on HPR Before, During and After Crisis 2007 2008 2013 -1 .4 -1 .2 -1 -. 8 -. 6 -. 4 H PR -2000 -1000 0 1000 2000 Debt
6. Conclusion The level of underpricing is indeed lower during the crisis period tested in our research. This is in line with the research of Loughran and Ritter (2004) where they find a lower level of underpricing during previous economic downturns in 2001-2003. The control variable sales has a negative relationship with underpricing, especially before the crisis. It means in normal times, the cost of IPO is a function of sales. But in crisis, this relationship disappears. On the other hand, sales has a positive effect on the HPR, especially before and after the crisis period. It means in normal times, the HPR is a function of sales. But in crisis, this relationship disappears as well. In both cases, the initial relationship disappears during the crisis period. Underpricing has a negative effect on the HPR. This means that a higher underpricing results in a lower HPR. This effect gets bigger over the measured time periods. Debt has no significant influence on underpricing. But debt has a significant positive influence on the HPR before the crisis. The key message is that there is no significant influence of sales or debt on the underpricing or HPR during the crisis period. In this specific case we see that normal economic expectations by economic theories do not count during the financial crisis. The normal influences of company’s sales and debt before and after the crisis period are not influencing the level of underpricing during the crisis period. In further research it would be interesting to see if there are more variables influencing the level of underpricing. From these other factors it could be investigated whether they have less influence during an economic downturn or not. Another idea for further research is to compare the same variables of sales and debt over multiple historical economic downturns.
Reference list Brav, A., Gompers, P., 1997, Myth or reality? The long-run underperformance of IPOs, Journal of Finance 52, 1791-1821 Loughran,T., Ritter, J., 2004, Why has IPO underpricing changed over time?, Journal of Financial Management, 5-37 Lukas, Deborah, and Robert McDonald, 1990, Equity issues and stock price dynamics, Journal of Finance 45: 1019-1043. Ransley, R.D., 1984, A research project into the operation and development of the unlisted securities market 1980- 1984, Unpublished (London Business School, London). Ritter, J. and Welch, I., 2002, A review of IPO activity, pricing and allocations, Journal of Finance 57: 1795-1828 Ritter, J., 1987, The cost of going public, Journal of Financial Economics 19, 269-281 Ritter, J., 1991, The long-run performance of initial public offerings, Journal of Finance 46, 3-27 Roëll, A., 1996, The decision to go public: an overview, European Economic Review 40,1071-1081
