The Post-Issue Operating Performance of Private Equity-backed
IPOs in the United States: An International Perspective
Master Thesis MSc International Financial Management
by Bas Jan Boekema
*June 2015
Abstract:
This paper studies the differences in post-issue operating performance between PE-backed and non-backed IPOs and between foreign and domestic IPOs, using a sample of 582 companies that went public in the US. Our findings reveal that both PE-backed and foreign IPOs tend to outperform non-backed and domestic IPOs in terms of ROA and ROE in the pre-offering period, but subsequently underperform them during the post-offering years, providing evidence for the window-dressing hypothesis. More importantly, we find that the negative effects of backing do not hold for foreign IPOs, since we find that foreign PE-backed IPOs tend to outperform domestic PE-PE-backed and foreign non-PE-backed IPOs in terms of ROA during the first and second year after flotation.
JEL Classification: G24; G32; and G34
Keywords: Initial Public Offerings (IPOs); Operating Performance; Private Equity
Supervisor: Dr. J. H. von Eije
Co-
assessor: Dr. A. J. Meesters
Page / 2
1. Introduction
This paper studies the post-issue operating performance of Private Equity (PE)-backed Initial Public Offerings (IPOs) in the United States (US). Even though a few studies investigate the post-issue operating performance of IPOs (see e.g., Jain and Kini, 1994; Mikkelson et al., 1997; Khurshed et al., 2005), the vast majority of studies shed light on the issue stock price performance. Research focusing on the post-issue operating performance of PE-backed IPOs is even scarcer. The purpose of this study is to address this gap in the literature.
According to Ernst & Young’s quarterly updates on Private Equity Public Exits (2013 and 2014), 2014 has been the best IPO-year since 2010. As a result of increased macroeconomic stability, improved equity markets and a boost in risk appetite, 1,232 initial public offerings raised a total amount of 259.2 billion dollar2 during 2014. Moreover, PE-backed deals accounted for a record proportion of the aggregate
global IPO activity in terms of proceeds, raising 109.9 billion dollar (42% of total proceeds) across 211 IPOs, an increase of 89% by value and 14% by volume compared to 20133. When excluding the initial
listing of Alibaba Group, the online and mobile commerce company that raised the largest amount through an IPO ever4, PE-backed IPO proceeds still increased 46% compared to 2013.
Surprisingly, although PE-backed IPOs continue to grow, the lack of comparative evidence regarding the operating performance of such deals still remains. Katz (2009) and Levis (2011) are among the few authors who examine the post-issue operating performance of PE-backed IPOs, focusing on the US and United Kingdom, respectively. Others have focused on a particular type of PE such as Leverage Buyouts (LBOs) returning to public status and hence becoming reverse LBOs (see e.g., DeGeorge and Zeckhauser, 1993; Holthausen and Larcker, 1996; Cao, 2010) or Venture Capital (VC)-backed IPOs (see e.g., Rindermann, 2003; Coakley et al., 2007; Chemmanur et al., 2012).
This paper contributes to the literature in several ways. First, by analysing a sample of PE-backed and non-backed IPOs, the study examines the median differences in post-issue operating performance across both groups for the US. According to previous studies conducted by Kaplan (1989) and Smith (1990), PE-firms are able to enhance the productivity and profitability of their portfolio companies.
2 Dollar refers to US dollar ($), the currency used in this paper.
3 2013 has seen the largest proportion of PE-backed IPOs in terms of deals: 181 IPOs accounting for 19% of the total number of IPOs (raising 57 billion dollar, 35% of the total IPO proceeds).
Page / 3 Moreover, Levis (2010) suggests that the management and financial practices applied when a company is under PE control will be maintained at least for some time after the exit as a result of continuing involvement of PE firms (e.g. by retained shareholdings). Therefore, the study investigates whether PE-firms continue to benefit their portfolio companies in the post-exit period, by analysing the post-issue operating performance of PE-backed IPOs and compares it to a sample of non-backed IPOs. We find that PE-backed IPOs tend to outperform non-backed IPOs in terms of ROA and ROE in the pre-offering period, but underperform them after flotation. Secondly, the paper examines the differences in operating performance between foreign and domestic IPOs5. Once companies have decided to go public, they can
choose to raise capital domestically or elsewhere. Similar to PE-backed IPOs, we find that foreign IPOs tend to outperform domestic IPOs in terms of ROA and ROE in the pre-offering period, but underperform domestic IPOs thereafter. Moreover, we find that foreign backed IPOs tend to outperform domestic PE-backed and foreign non-PE-backed IPOs. In addition to Katz (2009) and Levis (2010), we argue that the outperformance in operating performance of PE-backed companies may depend upon the type of IPO. In other words, the operating performance may differs between foreign backed IPOs and domestic PE-backed IPOs. So, the paper provides valuable information for both portfolio managers and managers responsible for funding decisions of international companies.
The remainder of the paper is organised as follows. The next section reviews prior literature on the operating performance of IPOs. Section 3 contains a description of the data and presents the sample characteristics. Section 4 explains the methodology, while Section 5 reports the results. Finally, Section 6 concludes.
2. Theory
2.1 Post-issue operating performance of IPOs
Previous studies on post-issue operating performance of IPOs show deterioration in operating performance subsequent to the IPO.Jain and Kini (1994), the first who study the operating performance of non-backed IPOs, find that IPO companies exhibit lower levels of operating returns on assets and operating cash flows deflated by total assets relative to their pre-offering year. Their findings are
Page / 4 consistent with the window-dressing hypothesis and market-timing hypothesis introduced by DeGeorge and Zeckhauser (1993) and Ritter (1991), respectively. According to the window-dressing hypothesis, managers manipulate companies’ earnings prior to the IPO (called window-dressing) in order to make it an attractive investment for investors, while the market-timing hypothesis suggests that companies go public when they perform well (i.e. display high earnings) or when the market is overconfident (also known as the windows of opportunity hypothesis). Both theories predict a decline in post-issue operating performance, as pre-offering levels of earnings cannot be sustained after flotation.
Furthermore, Jain and Kini (1994) provide evidence suggesting that the decline in post-issue operating performance is related to ownership changes. The dilution of managerial ownership that occurs after companies go public may lead to higher agency costs, described by Jensen and Meckling (1976), as managers have incentives to increase perquisite consumption (e.g. the use of IPO proceeds in non-shareholder value maximising projects).As a result of this non-shareholder value maximising behaviour of managers, post-issue operating performance is likely to deteriorate.Kutsuna et al. (2002) confirm this relationship between operating performance and ownership changes in their study of Japanese companies, while Álvarez and González (2005) and Wang (2005) conclude that ownership is unrelated to operating performance, studying Spanish and Chinese IPOs, respectively.
Page / 5
2.2 Post-issue operating performance of PE-backed IPOs
As stated in the introduction, evidence on the post-issue operating performance of PE-backed IPOs is scarce. Katz (2009) exploring the effect of private companies’ ownership structure on earnings management, reporting conservatism and post-issue performance, finds that the presence of and monitoring by PE-firms restrains upward earnings management (window-dressing) and induces a higher frequency of timely loss recognition. Moreover, she finds that larger PE-firms are positively associated with better post-issue operating and stock price performance by portfolio companies. She argues that this positive post-issue operating and stock price performance can be attributed to less upward earnings management, more timely loss recognition, tighter monitory and control by large PE-firms or those holding a majority stake. On the other hand, she argues that companies backed by PE-firms holding minority stakes exhibit worse post-issue operating and stock price performance than management-owned companies because of the lesser ability of these PE-firms to control and monitor managements.
Page / 6 and the marked reduction in debt immediately after flotation, which leads to better aftermarket performance in the post-issue period.
Meles et al. (2014), examining the operating performance of Italian PE investments exiting both via IPOs and other common ways such as secondary buyouts, also find that PE-backed companies outperform other companies. However, they argue that, among PE-investors, only venture capitalists provide favourable long-term effects for the companies. Moreover, they find that companies that went public show a larger decline in operating performance compared to companies using other exit strategies, whereas Sousa and Jenkinson (2013) show that companies going public outperform companies exiting through a secondary buy-out.
Other studies have focused on the post-issue operating performance of reverse LBOs and VC-backed IPOs. DeGeorge and Zeckhauser (1993) show that reverse LBOs outperform comparison companies (continuing LBOs and public companies) in the pre-offering period but underperform them in the post-issue period. They argue that this decline in post-issue operating performance can be explained by performance manipulation (window-dressing) and timing of reverse LBOs. Moreover, they find that reverse LBOs tend to be larger than the average IPO in terms of gross proceeds raised.Holthausen and Larcker (1996) show that the accounting performance of reverse LBOs is significantly better than their industry counterparts at time of the IPO and in the following four years, though a decline in performance is documented. They show that the decrease in performance is related to a decline in the concentration of equity ownership by management and insiders, but unrelated to changes in leverage. More recently, Cao (2010) finds that reverse LBOs exhibit no significant deterioration in operating performance after flotation. He argues that the decline in operating performance is related to short holding periods by the buyout sponsors. According to him, reverse LBOs with shorter holding periods experience worse operating performance and a higher probability of bankruptcy. Moreover, he finds that holding periods are negatively related to favourable market conditions, confirming the windows of opportunity hypothesis.
VC-Page / 7 backed and non-backed companies in the United Kingdom. He shows that both VC-backed and non-backed companies experience a significant decline in performance during the Dotcom bubble period (1998-2000), while IPOs issued in the pre-bubble years (1985-1997) do not underperform. Finally, Chemmanur et al. (2012) find that VC-backing and high management quality positively affect the companies’ post-issue operating performances.
Summarizing the theoretical and empirical work, there is sufficient evidence on post-issue operating performance of reverse LBOs and VC-backed IPOs, while the numbers of studies investigating PE-backed IPOs are limited. In order to contribute to the existing literature on PE-backed IPOs, this paper studies the following hypothesis:
H1: There are no differences in the post-issue operating performance between PE-backed and
non-backed IPOs.
H1A: There are positive differences in the post-issue operating performance between PE-backed and
non-backed IPOs, i.e. PE-backed IPOs outperform non-backed IPOs in the post-issue period.
Based on the preceding theory, we expect the differences in the post-issue operating performance between PE-backed and non-backed IPOs to be positive. Several authors show that PE-backed IPOs outperform non-backed IPOs in the post-issue period as a result of tighter monitoring, control and longer holding periods by PE-firms, less upwards earnings management, timely loss recognition, and so on.On the other hand, DeGeorge and Zeckhauser (1993) and Ritter (1991) propose two main reasons for deterioration in the post-issue operating performance, namely window-dressing and market-timing. Therefore, we expect PE-backed IPOs to outperform non-backed IPOs unless they engage in window-dressing or market-timing activities. We do not expect underperformance to be a result of ownership changes as the evidence of ownership changes on operating performance is rather mixed and Levis (2010) argues that management and financial practices applied when a company is under PE control will be maintained during the post-exit period. Moreover, we control for ownership changes in the multiple regressions conducted later on in this study.
2.3 Post-issue operating performance of foreign IPOs
Page / 8 American Depositary Receipts (ADR) programs, Charitou and Louca (2009) find, among other things, that capital-raising cross-listed companies experience improvements in their operating performance during the post-issue period relative to their pre-offering period. They argue that cross-listing on US stock exchanges potentially helps companies to raise the capital needed to support their growth opportunities, resulting in operating performance improvements after the listing. Moreover, Hail and Leuz (2005), find that cross-listing on an US stock exchange reduces companies’ cost of capital. They show that these effects are larger for companies from countries with weaker institutional structures. In addition, they show substantial cash flow effects, suggesting that cross-listing in the US improves companies’ abilities to exploit and generate growth opportunities. Hence, we investigate the following hypothesis in order to examine the differences in operating performance between foreign and domestic IPOs:
H2: There are no differences in the post-issue operating performance between foreign and domestic
IPOs.
H2A: There are positive differences in post-issue operating performance between foreign and
domestic IPOs, i.e. foreign IPOs outperform domestic IPOs in the post-issue period.
Again, we expect the differences in post-issue operating performance to be positive as companies cross-listing in the US have more options to attract capital and to exploit their growth opportunities, which in turn may result in better operating performance.
3. Data
3.1 Sample selection
Using Standard & Poor’s (S&P) Capital IQ database we select all IPOs that took place on the major US Stock Exchanges6 between 2000 and 2010.Similar to previous studies, we exclude companies belonging to the
financial sector, due to intrinsic differences in the nature of their operations and accounting information (Pagano, 1998), and utilities, because utility offerings tend to be different from those of other operating companies (Loughran and Ritter, 1995).Furthermore, we exclude issues with multiple listings at once and
Page / 9 IPOs backed by sponsors in which no PE-firm is involved7. Finally, companies with insufficient accounting
data for pre-offering levels are omitted from the sample. The total sample consists of 582 IPOs.
3.2 Sample description
Table 1 provides an overview of the number of IPOs per year, company nationality and S&P sector classification as well as some sample characteristics. As is shown in Panel A, almost one third of the total number of IPOs took place in 2000 at the time of the Dotcom bubble, when stock markets experienced a rapid rise in internet-based companies that went public.Furthermore, the majority of companies with PE-backing went public between 2004 and 2010, while most non-backed IPOs took place between 2000 and 2006. Additionally, the right-hand side of the table presents the distribution of foreign and domestic issues.The table shows much activity in 2000 and between 2004 and 2006, but less activity during 2008 when markets where hit by the financial crisis.
Panel B presents the sample characteristics. The mean (median) gross proceeds raised by these companies are 201.75 million (95.16 million) dollar.The companies’ mean (median) age at flotation is 15.7 years (8.0 years), calculated as the difference between the IPO-year and the company’s year of incorporation.The mean (median) number of primary shares offered in the IPO as a percentage of the pre-offering outstanding shares is 29.12% (23.59%). Finally, the mean (median) first day return, calculated as the first day price of the stock minus the offering price as a proportion of the offering price, is 50.24% (12.50%), indicating that the IPOs are on average underpriced.
Panels C and D provide the number of IPOs per company nationality and S&P Capital IQ sector classification, respectively. As can be observed from the table, 92 of the 582 companies that went public are incorporated outside the United States. Remarkably, the majority of companies incorporated in British Virgin Island and Cayman Islands have their headquarters in China, indicating the use of tax havens8 by
those companies. Furthermore, Panel D exhibits that a great number of IPOs is occurring in the Information Technology sector.The majority of those IPOs (109, all non-backed) took place during the Dotcom bubble in 2000, mentioned earlier in this paper. Moreover, Panel D shows that PE-backed IPOs are predominantly active in the Information Technology sector, while Levis (2010) find that the majority
Page / 10 of PE backed IPOs occurred in the Consumer Goods and Services sector. Finally, we observe that most foreign IPOs take place in the Information Technology sector, while the majority of domestic IPOs occur in the Consumer Discretionary, Healthcare and Information Technology sector.
3.3 Description of the subsamples
In order to investigate the differences between the types of IPOs, we split the total sample into four subsamples based on S&P Capital IQ classifications.
Table 2 shows the differences in characteristics between PE-backed and non-backed IPOs and between foreign and domestic IPOs. As we can observe from the Panel A, PE-backed IPOs are larger than non-backed IPOs in terms of gross proceeds, age at flotation and pre-offering ROA and ROE, while they are smaller in terms of primary shares offered at the IPO. The median difference in gross proceeds is 38.39 million dollar and the median difference in age at flotation is 2.0 years. The median differences in pre-offering ROA and ROE are 5.32% and 14.79%, respectively. Finally, the median difference in primary shares offered is -4.51%. All the median differences are significantly different from zero at the .01 level.
Panel B presents the differences between foreign and domestic IPOs. Foreign IPOs are larger in terms of gross proceeds raised through the IPO and pre-offering ROA and ROE, while non-backed IPOs are larger in terms of primary shares offered and first day return. The median difference in gross proceeds is 20.80 million dollar and the median differences in ROA and ROE are 6.77% and 20.62%, respectively. The median differences in primary shares offered and first day return are -6.46% and -8.68%, respectively. All the median differences are significantly different from zero at the 0.5 or .01 level.
In addition, Panels A and B provide the median differences for leverage and the change in leverage in the post-offering years. Both panels exhibit some small median differences, although not significantly different from zero.
4. Methodology
Page / 11 effectiveness and the ability of a company to generate earnings from its investments.EBITDA, total assets and total equity, all measured at the end of the fiscal year, are used to calculate these operating metrics. However, in contrast to Jain and Kini (1994), we calculate ROA as EBITDA9 divided by the average total
assets instead of just total assets, in order to account for varying assets totals throughout the year. In other words, ROA is calculated as EBITDA divided by the sum of total assets at the end of current fiscal year plus total assets at the end of the preceding fiscal year divided by two. For the same reason, ROE is calculated as EBITDA divided by average total equity.All companies are required to have accounting data available at S&P Capital IQ for the fiscal year prior to the IPO.However, we do not require them to have accounting data available for the three fiscal years following the IPO. Hence, we provide the numbers of observations in the tables as the number of companies in the sample may differ across time windows.
We measure the median change in operating performance as the median change in ratios, i.e. the median value of [ROA𝑖(𝑡) − ROA𝑖(−1)], where i represents the company, -1 represents the fiscal year prior to the IPO, and t represents a post-offering fiscal year end (ranging from +1 to +3).We follow the same procedure to measure the change in ROE.Like previous studies (see e.g. Kaplan, 1989; Jain and Kini, 1994; Pereira and Sousa, 2015) we use the medians rather than means as a measure of central tendency, because operating performance measures may be skewed and the mean is particularly sensitive to outliers. Moreover, we exclude results for the IPO-year as those results include pre- and post-offering data, which makes it difficult to separate pre- and post-exit operating performance. Furthermore, in contrast to Jain and Kini (1994), we do not adjust the median changes for sector performance as we control for it in the multiple regression analysis conducted later on in this study. Finally, in order to analyse the differences in operating performance between PE-backed and non-backed IPOs and between domestic and foreign IPOs, we split the total sample, repeat the procedure explained above and compare the differences.
Similar to Jain and Kini (1994), Levis (2010) and Pereira and Sousa (2015), the significance tests for median changes are based on the Wilcoxon signed-rank test, while the significance levels of median differences are based on a two-sample Wilcoxon rank-sum test (Mann-Whitney U-test).
Page / 12
5. Results
5.1 Operating performance of the total sample
Table 3 presents the median changes in ROA and ROE expressed in a percentage for the different time windows.The median changes in operating performance are measured relative to the pre-offering fiscal year (year -1). In contrast to the main literature, we observe an increase in ROA in the post-issue period. Panel A reports a median change in ROA of +2.67%, +1.25%, and +1.19% (all significantly different from zero at the .01 level) for year +1, +2, and +3 relative to year -1.Panel B reports a median change in ROE of -2.44%, -5.02%, and -6.74% for year +1, +2, and +3 relative to year -1.Although the declines in ROE are consistent with previous articles investigating operating performance of IPOs, the median changes in ROE are not significantly different from zero.
5.2 Operating performance of the sample split by PE-backed and non-backed IPOs
In order to investigate the median differences in post-issue operating performance between PE-backed and non-backed IPOs, we split the sample into two subsamples based on the S&P Capital IQ classification as explained in the sample selection.
Panel A of Table 4 (and Figure 1) presents the median changes in ROA for both PE-backed and non-backed IPOs.The median change in ROA for PE-backed IPOs is +1.07%, +0.07%, and -0.83% for year +1, +2, and +3 relative to year -1, but not significantly different from zero. Non-backed IPOs experience a positive median change in ROA of +2.87%, +1.50%, and +1.89% (all significantly different from zero at the .01 level) for year +1, +2, and +3 relative to the pre-offering fiscal year. Moreover, Panel A of Table 4 shows that PE-backed IPOs underperform non-backed IPOs in terms of ROA during the post-issue period, although the differences in median changes between both groups are only significant for year +1 and +3 (at .10 and .05, respectively).
Page / 13 period, although the differences in median changes between both groups are only significant for the second year after the offering.
5.3 Operating performance of the sample split by foreign and domestic IPOs
In addition to PE-backed and non-backed IPOs, we analyse the median differences in post-issue operating performance between foreign and domestic IPOs. We split the sample into two subsamples based on the companies’ country of incorporation in order to perform our analyses.
The median changes in ROA for foreign and domestic IPOs are presented in Panel C of Table 4 (and Figure 2). Foreign IPOs exhibit a negative median change in ROA of -1.91%, -1.59%, and -2.65% for year +1, +2, and +3 relative to year -1, but the median changes are not significantly different from zero. On the other hand, domestic IPOs experience positive median changes in ROA (+3.34%, +1.73%, and +1.79%) which are all significantly different from zero at the .01 level. Moreover, the results show that foreign IPOs underperform domestic IPOs in terms of ROA, as the differences in median changes between both groups are statistically significant at .01 for the whole time window.
Panel D of Table 4 (and Figure 2) shows the median changes in ROE for both IPO groups. Both foreign and domestic IPOs experience negative median changes in the years following the IPO. The median changes for foreign IPOs are -12.61%, -11.92%, and -14.19% for year +1, +2, and +3 relative to year -1. All median changes are significantly different from zero at the .05 level. Domestic IPOs also exhibit negative median changes (-1.11%, -3.71%, and -5.03%), but these are not significantly different from zero. Again, the results imply that foreign IPOs underperform domestic IPOs. Indeed, Panel D of Table 4 shows that all differences in median changes between both groups are statistically significant at the 0.1 level.
1.07% 0.07% -0.83% -8.38%* -10.43%** -9.99%** 2.87%*** 1.50%*** 1.89%*** -1.52% -3.18% -6.17% Median Change in
ROA from -1 to +1 ROA from -1 to +2Median Change in ROA from -1 to +3Median Change in ROE from -1 to +1Median Change in Median Change inROE from -1 to +2 Median Change inROE from -1 to +3
Figure 1: Median Change in ROA and ROE of the Subsamples - PE-backed and non-backed IPOs
Page / 14
5.4 Summary of initial findings
Our results reveal some interesting differences between the different types of IPOs. First of all, in contrast to the main literature that documents a decline in operating performance measures (see e.g., Jain and Kini, 1994; Mikkelson, 1997; Pereira and Sousa, 2015), our results for the total sample show positive median changes in ROA during the post-issue period compared to pre-offering levels. Secondly, in contrast to our expectation (H1), the results show that PE-backed IPOs tend to underperform non-backed IPOs both in terms of ROA and ROE during the post-issue period (although not all the negative differences in median changes are statistically significant). Furthermore, in contrast to our second hypothesis (H2), the results imply that foreign IPOs underperform domestic IPOs both in terms of ROA and ROE. They display worse median changes in ROA and ROE compared to domestic IPOs and all the differences in median changes between both groups are statistically significant. The results are remarkable because both PE-backed and foreign IPOs have significant better ROA and ROE in the pre-offering period compared to non-backed and domestic IPOs, respectively. Indeed, the outperformance of PE-backed and foreign IPOs in the pre-offering period and the subsequent underperformance in the post-offering period provide evidence for the window-dressing or market-timing hypothesis.
5.5 Multiple regression analysis
The differences in post-issue operating performance may also be a result of other factors such as the company characteristics. In order to control for these effects, we estimate the following equation:
(1) 𝑦𝑖= 𝛼0+ 𝛼1𝑃𝐸𝑖+ 𝛼2𝑓𝑜𝑟𝑒𝑖𝑔𝑛𝑖+ 𝛼3𝑠𝑖𝑧𝑒𝑖+ 𝛼4𝑙𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖+ 𝛼5𝑎𝑔𝑒𝑖+ 𝛼6𝑠ℎ𝑎𝑟𝑒𝑠𝑖+ 𝜀𝑖 -1.91% -1.59% -2.65% -12.61%** -11.92%** -14.19%** 3.34%*** 1.73%*** 1.79%*** -1.11% -3.71% -5.03% Median Change in
ROA from -1 to +1 ROA from -1 to +2Median Change in ROA from -1 to +3Median Change in ROE from -1 to +1Median Change in Median Change inROE from -1 to +2 Median Change inROE from -1 to +3
Figure 2: Median Change in ROA and ROE of the Subsamples - Foreign and Domestic IPOs
Page / 15 where,
𝑦𝑖 is the dependent variable, which is the median change in ROA or ROE in the first, second or third year after the IPO. Subscript i represents the company and 𝜀𝑖is the error term. The variables on the right-hand
side of the equations are explanatory variables. 𝑃𝐸𝑖 and 𝐹𝑜𝑟𝑒𝑖𝑔𝑛𝑖 are dummy variables that equal 1 if the company is PE-backed and incorporated outside the US, respectively. We predict the dummy variables’ coefficient signs to be positive, as we expect that PE-backed and foreign IPOs positively influence the changes in ROA and ROE. Similar to previous studies, we control for 𝑠𝑖𝑧𝑒𝑖, which is the company’s size in the pre-offering year calculated as the natural logarithm of average total assets, and 𝑎𝑔𝑒𝑖, which is the company’s age at flotation calculated as the natural logarithm of age. Again, we predict the coefficient signs of size and age to be positive, since Mikkelson et al. (1997) and Pereira and Sousa (2015) find that larger and older companies tend to perform better during the post-issue period. According to Levis (2010), PE-backed IPOs have higher debt levels at time of flotation compared to non-backed IPOs. Therefore, we also control for 𝑙𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖, which is the company’s leverage ratio in the pre-offering year and calculated as average total debt divided by average total equity. We predict the coefficient sign of leverage to be positive, since we expect that leverage positively affect the ROE as the cost of debt is usually lower than the cost of equity. Finally, we control for 𝑠ℎ𝑎𝑟𝑒𝑠𝑖, which is the number of primary shares offered in the IPO as a percentage of the pre-offering outstanding shares. The evidence of ownership changes on operating performance is rather mixed (see e.g., Kutsuna et al., 2002; Wang, 2005), so we do not predict the sign of this control variable.
In addition, we include four dummy variables in order to control for sector-related performance and performance differences across time periods. The first sector dummy represents the Information Technology sector, while the second sector dummy controls for both the Consumer Discretionary and Consumer Staples. In the regression, both dummies benchmark their operating performance against all other sectors. The third and fourth dummy variable control for the differences over time and therefore control for the market-timing hypothesis. The dummy variables for period 2000-2003 and period 2008-2010 benchmark their performance against period 2004-2007.
Page / 16 independent variables equal to their 1st and 99th percentile values. We also correct the standard errors of
the coefficient estimates by applying White’s heteroskedasticity method in the multiple regression analysis.
Table 5 provides an overview of the multiple regressions estimating the relationship between the change in ROA, the change in ROE and the various explanatory variables. In contrast to our expectation (H1), we find no evidence that the changes in ROA are related to being a PE-backed company. Moreover, the coefficient signs for PE-backed IPOs are negative for the second and third year after flotation, although not statistically significant. Regarding foreign IPOs, the table shows that the change in ROA in year +1 and +2 are negatively associated with being a foreign company. Contrary to our hypothesis (H2), this implies that foreign IPOs underperform in terms of ROA during the first and second post-offering year. In addition, the table shows that size and age are negatively associated with the changes in ROA, implying that larger and older companies tend to underperform in terms of ROA during the post-issue period. This is an interesting result, since Mikkelson et al. (1997) and Pereira and Sousa (2015) find that age and size are positively related to the ROA. In addition, the table shows a positive significant relationship between the first period dummy and the changes in ROA, suggesting that companies that went public between 2000 and 2003 (the Dotcom bubble period) performed better in terms of ROA.
Page / 17
5.6 Robustness checks
In order to test the robustness of our multiple regressions, we estimate two additional equations. Holthausen and Larcker (1996) find that the post-issue operating performance of reverse LBOs (measured by several cash flow variables deflated by total assets) is unrelated to changes in leverage. In addition, we are interested in the relationship between the changes in ROE and the changes in leverage during the post-issue period. Alternatively, like Pereira and Sousa (2015) we control for underpricing as this feature can be used by companies to signal their quality to the market in presence of asymmetric information (see e.g., Allen and Faulhaber, 1989; Grinblatt and Hwang, 1989; Welch, 1989). This signalling theory predict that companies underpricing their stock exhibit better operating performance in comparison to those that do not underprice their stock. We estimate the following equation to test the impact of changes in leverage and underpricing:
(2) 𝑦𝑖= 𝛼0+ 𝛼1𝑃𝐸𝑖+ 𝛼2𝑓𝑜𝑟𝑒𝑖𝑔𝑛𝑖+ 𝛼3𝑠𝑖𝑧𝑒𝑖+ 𝛼4𝑙𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖+ 𝛼5𝑎𝑔𝑒𝑖+ 𝛼6𝑠ℎ𝑎𝑟𝑒𝑠𝑖+ 𝛼7𝑐ℎ𝑎𝑛𝑔𝑒 𝑙𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖+
𝛼8𝑢𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔𝑖+ 𝜀𝑖
where,
𝑐ℎ𝑎𝑛𝑔𝑒 𝑙𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖 is the difference in leverage ratio for the first, second and third year after the IPO relative to the pre-offering year. More precisely, the change in leverage is measured each time during the period in which the changes of ROA and ROE are measured. We predict the coefficient signs of changes in leverage to be positive because an increase in leverage normally positively affects the change in ROE if the ROA remains unchanged. 𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔𝑖 is the first-day return (also known as the initial return), calculated as the first-day closing price of the stock minus the offering price as a proportion of the offering price. We expect the coefficient sign of underpricing to be positive, since the signalling theory argues that companies who underprice tend to outperform those that do not underprice during the post-issue period (see e.g., Jain and Kini, 1994).
Page / 18 underperform in terms of ROA during the second post-offering year. Moreover, the table does not provide evidence that the changes in ROE are related to being a PE-backed or foreign company. Changes in leverage are positively related to the changes in ROE, implying that companies increasing their leverage ratios perform better in terms of ROE. In line with the signalling theory, we find that underpricing is positively related to changes in ROA, suggesting that companies who underprice their stock perform better in terms of ROA during the post-issue period.
Additionally, we add an extra dummy variable that controls for foreign PE-backed IPOs. By this means, we are able to see how our results are influenced by the interaction between a company being backed and incorporated outside the US. More precisely, we test if the unexpected negative effects of PE-backing also hold for foreign PE-backed IPOs. We estimate the following equation in order to address this question:
(3) 𝑦𝑖= 𝛼0+ 𝛼1𝑃𝐸𝑖+ 𝛼2𝑓𝑜𝑟𝑒𝑖𝑔𝑛𝑖+ 𝛼3𝑃𝐸 ∗ 𝑓𝑜𝑟𝑒𝑖𝑔𝑛𝑖+ 𝛼4𝑠𝑖𝑧𝑒𝑖+ 𝛼5𝑙𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖+ 𝛼6𝑎𝑔𝑒𝑖+ 𝛼7𝑠ℎ𝑎𝑟𝑒𝑠𝑖+
𝛼8𝑐ℎ𝑎𝑛𝑔𝑒 𝑙𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖+ 𝛼9𝑢𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔𝑖+ 𝜀𝑖
where,
𝑃𝐸 ∗ 𝑓𝑜𝑟𝑒𝑖𝑔𝑛𝑖 is a dummy variable that equals 1 if the company is PE-backed and incorporated outside the US.
Page / 19
6. Conclusion
Using a sample of 582 IPOs in the US, we study the differences in post-issue operating performance between PE-backed and non-backed IPOs and between foreign and domestic IPOs. Contrary to the main literature, we find an increase in median ROA during the post-issue period compared to pre-offering levels for the sample as a whole.
Subsequently, we split our sample in PE-backed and non-backed IPOs and find that PE-backed IPOs are larger in terms of gross proceeds raised, older at the time of flotation and exhibit better ROA and ROE compared to non-backed companies in the fiscal year before flotation. Moreover, PE-backed IPOs tend to underperform in terms of ROA and ROE during the post-issue period compared to non-backed IPOs, although we do not find that this underperformance is related to being a PE-backed company.
In addition, we split the sample in foreign and domestic IPOs and find that foreign IPOs are larger in terms of gross proceeds raised and pre-offering ROA and ROE compared to domestic IPOs. Moreover, we find that foreign IPOs tend to underperform domestic IPOs both in terms of ROA and ROE during the post-issue period. Indeed, the multiple regression analysis shows that the post-issue changes in ROA are related to foreign companies. This outperformance of foreign IPOs in terms of ROA in the pre-offering period and subsequent underperformance in the post-issue period provides evidence for the window-dressing hypothesis.
More importantly, we find that the negative effects of PE-backing do not hold for foreign IPOs. Our results show that foreign backed IPOs exhibit significant better changes in ROA than domestic PE-backed and foreign non-PE-backed IPOs during the first and second post-offering year.
Page / 20
7. References
Allen, F., and Faulhaber, G., 1989. Signalling by Underpricing in the IPO Market. Journal of Financial Economics, Vol. 23, No. 2, 303-323.
Álvarez, S., González, M., 2005. Performance of Spanish Firms Going Public: Windows of Opportunity and the Informative Effect. Applied Financial Economics, Vol. 15, 1283-1297.
Cao, J., 2010. IPO Timing, Buyout Sponsors’ Exit Strategies and Firm Performance of RLBOs. Working Paper. Singapore Management University.
Charitou, A., Louca, C., 2009. Cross-Listing and Operating Performance: Evidence from Exchange-Listed American Depositary Receipts. Journal of Business Finance and Accounting, Vol. 36, No. 1/2, 99-129.
Chemmanur, T., Simonyan, K., Tehranian, H., 2012. Management Quality, Venture Capital Backing, and Initial Public Offerings. Working Paper. Carroll School of Management, Boston College.
Coakley, J., Hadass, L., Wood, A., 2007. Post-IPO Operating Performance, Venture Capital and the Bubble Years. Journal of Business Finance and Accounting, Vol. 34, No. 9/10, 1423-1446.
DeGeorge, F., Zeckhauser, R., 1993. The Reverse LBO Decision and Firm Performance: Theory and Evidence. Journal of Finance, Vol. 48, No. 4, 1323-1348.
Ernst & Young, 2013. Private Equity Public Exits Q4 2013.
Ernst & Young, 2014. Private Equity Public Exits Q4 2014.
Grinblatt, M., Hwang, C., 1989. Signalling and the Pricing of New Issues. Journal of Finance, Vol. 44, No. 2, 393-420.
Hail, L., Leuz, C., 2005. Cost of Capital and Cash Flow Effects of US Cross-Listings. Working Paper. The Wharton School, University of Pennsylvania.
Holthausen, R., Larcker, D., 1996. The Financial Performance of Reverse Leveraged Buyouts. Journal of Financial Economics, Vol. 42, 293-332.
Page / 21 Jensen, M., Meckling, W., 1976. Theory of the Firm: Managerial Behavior, Agency Costs and Ownership
Structure. Journal of Financial Economics, Vol. 3, No. 4, 305-360.
Kaplan, S., 1989. The Effects of Management Buyouts on Operating Performance and Value. Journal of Financial Economics, Vol. 24, 217-254.
Katz, S., 2009. Earnings Quality and Ownership Structure: The Role of Private Equity Sponsors. The Accounting Review, Vol. 84, No. 3, 623-658.
Khurshed, A., Paleari, S., Vismara, S., 2005. The Operating and Share Price Performance of Initial Public Offerings: The UK Experience. Working Paper. Division of Accounting Finance, Manchester Business School.
Kutsuna, K., Okamura, H., Cowling, M., 2002. Ownership Structure Pre- and Post-IPOs and the Operating Performance of JASDAQ Companies. Pacific-Basin Finance Journal, Vol. 10, 163-181.
Levis, M., 2011. The Performance of Private Equity-Backed IPOs. Financial Management, Spring 2011, 253-277.
Loughran, T., Ritter, J., 1995. The New Issues Puzzle. Journal of Finance, Vol. 1, No. 1, 23-51.
Meles, A., Monferrà, S., Verdoliva, V., 2014. Do the Effects of Private Equity Investments on Firm Performance Persist Over Time?. Applied Financial Economics, Vol. 24, No. 3, 203-218.
Mikkelson, W., Megan Partch, M., Shah, K., 1997. Ownership and Operating Performance of Companies That Go Public. Journal of Financial Economics, Vol. 44, No. 3, 281-307.
Pagano, M., Panetta, F., Zingales, L., 1998. Why Do Companies Go Public? An Empirical Analysis. Journal of Finance, Vol. 53, No. 1, 27-64.
Pereira, T., Sousa, M., 2015. Is There Still a Berlin Wall in the Post-Issue Operating Performance of European IPOs?. School of Economics and Management, University of Porto.
Rindermann, G., 2003. Venture Capitalist Participation and the Performance of IPO Firms: Emperical Evidence From France, Germany and the UK. Working Paper. Department of International Business, University of Münster.
Page / 22 Smith, A., 1990. Corporate Ownership Structure and Performance: The Case of Management Buyouts.
Journal of Financial Economics, Vol. 27, 143-164.
Sousa, M., Jenkinson, T., 2013. Keep Taking The Private Equity Medicine? How Operating Performance Differs Between Secondary Deals and Companies That Go To Public Markets. Working Paper. University of Oxford.
Wang, C., 2005. Ownership and Operating Performance of Chinese IPOS. Journal of Banking and Finance, Vol. 29, 1835-1856.
Page / 23
Table 1: Summary Statistics of the Total Sample
Panel A presents an overview of the number of IPOs per year, while Panel B provides some sample characteristics. Panels C and D report the number of IPOs per company nationality and S&P Capital IQ sector classification, respectively.The sample is retrieved from the S&P Capital IQ database according the criteria mentioned earlier in this paper.Missing values for country- and year of incorporation are obtained from ORBIS (Bureau van Dijk).Gross Proceeds presents the proceeds raised by the companies through the
IPO.Age at Flotation is the difference between the IPO-year and the year of incorporation of the company. Shares Offered as % of Outstanding Shares represents the number of primary shares offered in the IPO as a percentage of the pre-offering outstanding
shares. The First-Day Return (or initial return) is calculated as the first-day closing price of the stock minus the offering price as a proportion of the offering price.
Panel A: Overview IPOs per Year
Year PE-backed non-backed Foreign Domestic Total
2000 1 168 25 144 16910 2001 1 53 8 46 54 2002 4 31 2 33 35 2003 4 11 0 15 15 2004 19 70 10 79 89 2005 16 59 10 65 75 2006 12 52 15 49 64 2007 24 8 8 24 32 2008 1 5 1 5 6 2009 11 5 4 12 16 2010 20 7 9 18 27 Total 113 469 92 490 58211
Panel B: Sample Characteristics
Variable Mean Median Std. Dev. Min Max Obs.
Gross Proceeds ($ million) 201.75 95.16 484.67 0.74 8680.00 582
Age at Flotation (years) 15.7 8.0 21.3 0.0 183.0 582
Shares Offered % of Outstanding Shares (%) 29.12 23.59 19.72 0.41 100.00 385
First-Day Return (%) 50.24 12.50 534.89 -91.18 12666.67 565
10 109 out of 168 took place in the Information Technology sector.
Page / 24 Panel C: Overview IPOs per Company Nationality
Country of Incorporation Headquarters
Country PE-backed non-backed Total Country PE-backed non-backed Total
Argentina 0 1 1 Argentina 0 1 1
Australia 0 1 1 Bermuda 1 4 5
Bermuda 2 7 9 Brazil 0 1 1
Brazil 0 1 1 Canada 1 6 7
British Virgin Islands 1 4 512 Cayman Islands 2 1 3
Canada 1 8 9 China 12 25 37
Cayman Islands 14 22 3613 Germany 0 1 1
Channel Islands 1 0 1 Greece 0 1 1
China 0 1 1 Hong Kong 1 3 4
Hong Kong 0 2 2 Ireland 1 2 3
Ireland 1 1 2 Israel 1 8 9
Israel 1 9 10 Japan 0 1 1
Japan 0 1 1 Luxembourg 0 1 1
Luxembourg 0 1 1 Netherlands 0 1 1
Marshall Islands 0 1 1 New Zealand 0 1 1
Netherlands 0 1 1 Norway 0 1 1
New Zealand 0 1 1 Singapore 1 1 2
Norway 0 1 1 South Korea 1 1 2
Singapore 1 2 3 Switzerland 0 1 1
South Korea 1 1 2 Taiwan 1 3 4
Switzerland 0 2 2 United Kingdom 0 1 1
Taiwan 0 1 1 United States 91 404 495
United States 90 400 490
Total 113 469 582 Total 113 469 582
Panel D: Overview IPOs per S&P Capital IQ Sector Classification
PE-backed non-backed Domestic Foreign Total
Industry Classification No. % No. % No. % No. % No. %
Consumer Discretionary 21 18,6% 62 13,2% 72 14,7% 11 12,0% 83 14,3% Consumer Staples 3 2,7% 9 1,9% 9 1,8% 3 3,3% 12 2,1% Energy 4 3,5% 44 9,4% 40 8,2% 8 8,7% 48 8,2% Healthcare 13 11,5% 111 23,7% 118 24,1% 6 6,5% 124 21,3% Industrials 15 13,3% 45 9,6% 53 10,8% 7 7,6% 60 10,3% Information Technology 53 46,9% 182 38,8% 183 37,3% 52 56,5% 23514 40,4% Materials 3 2,7% 14 3,0% 12 2,4% 5 5,4% 17 2,9% Telecommunication Services 1 0,9% 2 0,4% 3 0,6% 0 0,0% 3 0,5% Total 113 100,0% 469 100,0% 490 100,0% 92 100,0% 582 100,0% 12 All companies are headquartered in China. 13 28 out of 36 companies are headquartered in China.
Page / 25
Table 2: Summary Characteristics of the Subsamples
Panel A presents the sample characteristics split by PE-backed and non-backed IPOs, while Panel B presents the sample characteristics split by foreign and domestic IPOs. Gross Proceeds presents the proceeds raised by the companies through the IPO.
Age at Flotation is the difference between the year of the IPO and the year of incorporation of the company. Shares Offered as % of Outstanding Shares represents the number of primary shares offered in the IPO as a percentage of the pre-offering outstanding
shares. The First-Day Return (or initial return) is calculated as the first-day closing price of the stock minus the offering price as a proportion of the offering price. ROA and ROE are defined as EBITDA divided by average total assets and EBITDA divided by average total equity, respectively. Leverage is calculated as average total debt divided by average total equity. Change Leverage (t) is the difference in leverage ratio between the post-offering year (+1, +2 or +3) and the fiscal year before flotation. Year -1 is the fiscal year preceding the year during the company goes public. Significance levels of median differences are based on a two-sample Wilcoxon rank-sum test (Mann-Whitney U-test). The significance levels .10, .05 and .01 are indicated by *, ** and ***, respectively.
Panel A: Sample Characteristics - PE-backed and non-backed IPOs
PE-backed non-backed
Variable Median Mean (Obs.) Median Mean (Obs.) Median Difference
Gross Proceeds ($ million) 124.66 224.58 113 86.27 196.25 469 38.39***
Age at Flotation (years) 10.0 19.0 113 8.0 14.9 469 2.0***
Shares Offered % of Outstanding Shares (%) 20.21 24.24 91 24.72 30.64 294 -4.51***
First-Day Return (%) 13.33 18.96 113 12.35 58.06 452 0.98
Return on Assets (ROA) (Year -1) (%) 13.98 14.32 113 8.66 -15.76 469 5.32*** Return on Equity (ROE) (Year -1) (%) 32.60 75.60 113 17.81 -96.00 469 14.79***
Leverage (Year -1) 0.97 4.26 113 0.87 0.85 469 0.10
Change Leverage (Year +1) -0.37 -3.35 103 -0.43 -1.18 417 0.06
Change Leverage (Year +2) -0.41 -4.08 101 -0.38 4.14 386 -0.03
Change Leverage (Year +3) -0.36 -3.95 94 -0.41 -2.54 351 0.05
Panel B: Sample Characteristics – Foreign and Domestic IPOs
Foreign Domestic
Variable Median Mean (Obs.) Median Mean (Obs.) Median Difference
Gross Proceeds ($ million) 110.80 355.99 92 90.00 172.79 490 20.80**
Age at Flotation (years) 9.0 15.6 92 8.0 15.8 490 1.00
Shares Offered % of Outstanding Shares (%) 17.46 21.35 42 23.92 30.08 343 -6.46***
First-Day Return (%) 4.65 29.36 86 13.33 53.99 479 -8.68**
Return on Assets (ROA) (Year -1) (%) 16.25 12.76 92 9.48 -14.17 490 6.77*** Return on Equity (ROE) (Year -1) (%) 39.09 46.09 92 18.47 -83.11 490 20.62**
Leverage (Year -1) 0.83 -1.6615 92 0.90 2.11 490 -0.07
Change Leverage (Year +1) -0.41 -0.13 85 -0.43 -1.90 435 0.02
Change Leverage (Year +2) -0.43 -0.07 82 -0.37 2.94 405 -0.06
Change Leverage (Year +3) -0.42 -0.09 74 -0.40 -3.39 371 -0.02
Page / 26
Table 3: Post-Issue Operating Performance of Total Sample
Panels A and B show the median absolute change of ROA and ROE expressed in percentage for the total sample of IPOs that occurred between 2000 and 2010. The change in operating performance is measured relative to the pre-offering fiscal year (-1). ROA and ROE are defined as EBITDA divided by average total assets and EBITDA divided by average total equity, respectively. All companies have accounting data available at S&P Capital IQ for the fiscal year prior to the IPO, but are not required to have accounting data available for the three fiscal years following the IPO. The significance tests are based on the Wilcoxon signed-rank test. The significance levels .10, .05 and .01 are indicated by *, ** and ***, respectively.
Year Relative to Completion of IPO
From -1 to +1 From -1 to +2 From -1 to +3
Measure of Operating Performance Median Mean Median Mean Median Mean
Panel A: Return on Assets (ROA) Median Level Year -1 (%)
Sample of IPO Companies: 10.12%
Change (%) 2.67*** 10.91 1.25*** 9.30 1.19*** 9.21
Number of Observations 520 486 445
Panel B: Return on Equity (ROE) Median Level Year -1 (%)
Sample of IPO Companies: 21.99%
Change (%) -2.44 56.75 -5.02 129.01 -6.74 78.11
Page / 27
Table 4: Post-Issue Operating Performance of Subsamples
Panels A and B show the median absolute change of ROA and ROE expressed in percentage for the sample split by PE-backed and non-backed IPOs, while Panels C and D show the median absolute change of ROA and ROE expressed in percentage for the sample split by foreign and domestic IPOs. The change in operating performance is measured relative to the pre-offering fiscal year (-1). ROA and ROE are defined as EBITDA divided by average total assets and EBITDA divided by average total equity, respectively. All companies have accounting data available at S&P Capital IQ for the fiscal year prior to the IPO, but are not required to have accounting data available for the three fiscal years following the IPO. The significance tests for median changes are based on the Wilcoxon signed-rank test, while significance levels of median differences are based on a two-sample Wilcoxon rank-sum test (Mann-Whitney U-test). The significance levels .10, .05 and .01 are indicated by *, ** and ***, respectively.
Year Relative to Completion of IPO
From -1 to +1 From -1 to +2 From -1 to +3
Measure of Operating Performance Median Mean Median Mean Median Mean
Panel A: Return on Assets (ROA) - PE-backed and non-backed IPOs PE-backed Change (%) 1.07 0.48 0.07 -1.31 -0.83 -1.78 Number of Observations 103 101 94 Non-backed Change (%) 2.87*** 13.48 1.50*** 12.09 1.89*** 12.15 Number of Observations 417 385 351 Difference -1.80* -1.43 -2.72**
Panel B: Return on Equity (ROE) - PE-backed and non-backed IPOs PE-backed Change (%) -8.38* -45.46 -10.43** -65.98 -9.99** -63.77 Number of Observations 103 101 94 Non-backed Change (%) -1.52 82.00 -3.18 180.17 -6.17 115.89 Number of Observations 417 385 353 Difference -6.86 -7.25* -3.82
Panel C: Return on Assets (ROA) – Foreign and Domestic IPOs Foreign Change (%) -1.91 -3.46 -1.59 -3.93 -2.65 -1.60 Number of Observations 85 81 74 Domestic Change (%) 3.34*** 13.71 1.73*** 11.95 1.79*** 11.37 Number of Observations 435 405 371 Difference -5.25*** -3.32*** -4.44**
Page / 28
Table 5: OLS Estimation Change in ROA and ROE
The table reports the multiple regressions on the change in ROA and ROE for year +1, +2 and +3 relative to the pre-offering year (dependent variable). The independent variables include two dummy variables, PE and Foreign, that equal 1 if the company is PE-backed and incorporated outside the US, respectively. Size represents the company’s size in the pre-offering year calculated as the natural logarithm of average total assets. Leverage is the company’s leverage ratio in the pre-offering year calculated as average total debt divided by average total equity. Age is the company’s age at flotation
calculated as the natural logarithm of age, while Shares represents the number of primary shares offered in the IPO as a percentage of the pre-offering outstanding shares. Sector 1 is a dummy variable that equals 1 if the company is from the Information Technology sector, while Sector 2 is a dummy variable that equals 1 if the company is from the Consumer Discretionary or Consumer Staples sector. Finally,
Period 1 and Period 2 are dummy variables for period 2000-2003 and period 2008-2010, respectively. All regressions are estimated using the OLS method. The sample is winsorised by setting the top and
bottom 1% values of both the dependent variable and independent variables equal to their 1st and 99th percentile values. The numbers in the parentheses are White’s (1980) heteroskedasticity-consistent t-statistics. The significance levels .10, .05 and .01 are indicated by *, ** and ***, respectively (see bold values).
OLS Estimation Change in ROA and ROE
Variables Change ROA +1 Change ROA +2 Change ROA +3 Change ROE +1 Change ROE +2 Change ROE +3
Page / 29
Table 6: Robustness of OLS Estimation Change in ROA and ROE (including Underpricing and Change Leverage)
The table reports the multiple regressions on the change in ROA and ROE for year +1, +2 and +3 relative to the pre-offering year (dependent variable). The independent variables include two dummy variables, PE and Foreign, that equal 1 if the company is PE-backed and incorporated outside the US, respectively. Size represents the company’s size in the pre-offering year calculated as the natural
logarithm of average total assets. Leverage is the company’s leverage ratio in the pre-offering year calculated as average total debt divided by average total equity. Age is the company’s age at flotation calculated as the natural logarithm of age, while Shares represents the number of primary shares offered in the IPO as a percentage of the pre-offering outstanding shares. Change Leverage (t) is the difference in leverage ratio between the post-offering year (+1, +2 or +3) and the fiscal year before flotation. Underpricing represents the first-day return (also known as the initial return), calculated as the first-day closing price of the stock minus the offering price as a proportion of the offering price. Sector 1 is a dummy variable that equals 1 if the company is from the Information Technology sector, while
Sector 2 is a dummy variable that equals 1 if the company is from the Consumer Discretionary or Consumer Staples sector. Finally, Period 1 and Period 2 are dummy variables for period 2000-2003 and
period 2008-2010, respectively. All regressions are estimated using the OLS method. The sample is winsorised by setting the top and bottom 1% values of both the dependent variable and independent variables equal to their 1st and 99th percentile values. The numbers in the parentheses are White’s (1980) heteroskedasticity-consistent t-statistics. The significance levels .10, .05 and .01 are indicated by *, ** and ***, respectively (see bold values).
OLS Estimation Change in ROA and ROE including Underpricing and Change Leverage (t)
Variables Change ROA +1 Change ROA +2 Change ROA +3 Change ROE +1 Change ROE +2 Change ROE +3
Page / 30
Table 7: Robustness of OLS Estimation Change in ROA and ROE (including PE * Foreign, Underpricing and Change Leverage)
The table reports the multiple regressions on the change in ROA and ROE for year +1, +2 and +3 relative to the pre-offering year (dependent variable). The independent variables include three dummy variables, PE, Foreign and PE * Foreign, that equal 1 if the company is PE-backed, incorporated outside the US, and both, respectively. Size represents the company’s size in the pre-offering year calculated as
the natural logarithm of average total assets. Leverage is the company’s leverage ratio in the pre-offering year calculated as average total debt divided by average total equity. Age is the company’s age at flotation calculated as the natural logarithm of age, while Shares represents the number of primary shares offered in the IPO as a percentage of the pre-offering outstanding shares. Change Leverage (t) is the difference in leverage ratio between the post-offering year (+1, +2 or +3) and the fiscal year before flotation. Underpricing represents the first-day return (also known as the initial return), calculated as the first-day closing price of the stock minus the offering price as a proportion of the offering price. Sector 1 is a dummy variable that equals 1 if the company is from the Information Technology sector, while
Sector 2 is a dummy variable that equals 1 if the company is from the Consumer Discretionary or Consumer Staples sector. Finally, Period 1 and Period 2 are dummy variables for period 2000-2003 and
period 2008-2010, respectively. All regressions are estimated using the OLS method. The sample is winsorised by setting the top and bottom 1% values of both the dependent variable and independent variables equal to their 1st and 99th percentile values. The numbers in the parentheses are White’s (1980) heteroskedasticity-consistent t-statistics. The significance levels .10, .05 and .01 are indicated by *, ** and ***, respectively (see bold values).
OLS Estimation Change in ROA and ROE including PE * Foreign, Underpricing and Change Leverage (t)
Variables Change ROA +1 Change ROA +2 Change ROA +3 Change ROE +1 Change ROE +2 Change ROE +3
Page / 31
Appendix 1: Correlation Explanatory Variables
The table reports the correlation coefficients of the explanatory variables. PE, Foreign and PE * Foreign are dummy variables that equal 1 if the company is PE-backed, incorporated outside the US, and both, respectively. Size represents the company’s size in the pre-offering year calculated as the natural logarithm of average total assets. Leverage is the company’s leverage ratio in the pre-offering year
calculated as average total debt divided by average total equity. Age is the company’s age at flotation calculated as the natural logarithm of age, while Shares represents the number of primary shares offered in the IPO as a percentage of the pre-offering outstanding shares. Change Leverage is the difference in leverage ratio for the first, second and third year after the IPO relative to the pre-offering year.
Underpricing represents the first-day return (also known as the initial return), calculated as the first-day closing price of the stock minus the offering price as a proportion of the offering price. Sector 1 is a
dummy variable that equals 1 if the company is from the Information Technology sector, while Sector 2 is a dummy variable that equals 1 if the company is from the Consumer Discretionary or Consumer Staples sector. Finally, Period 1 and Period 2 are dummy variables for period 2000-2003 and period 2008-2010, respectively.
Correlation Explanatory Variables
Variables PE For. PE * For. Size Lev. Age Shares C.L. +1 C.L. +2 C.L. +3 Und. Sec. 1 Sec. 2 Per. 1 Per. 2
PE 1.000
Foreign (For.) 0.091 1.000
PE * Foreign (PE * For.) 0.351 0.590 1.000
Size 0.201 0.162 0.184 1.000
Leverage (Lev.) -0.122 0.096 0.017 0.091 1.000
Age 0.095 0.030 -0.006 0.368 -0.019 1.000
Shares -0.132 -0.153 -0.115 -0.035 0.090 0.048 1.000
Change Lev. (Year +1) (C.L. +1) 0.119 -0.053 0.006 0.001 -0.789 0.018 -0.076 1.000
Change Lev. (Year +2) (C.L. +2) 0.048 -0.097 -0.025 0.054 -0.783 0.097 -0.056 0.584 1.000
Change Lev. (Year +3) (C.L. +3) 0.093 -0.028 0.004 -0.121 -0.646 -0.079 -0.056 0.615 0.442 1.000
Underpricing (Und.) -0.091 -0.047 -0.046 -0.243 -0.005 -0.154 -0.081 -0.019 -0.025 0.013 1.000
Sector 1 (Sec. 1) 0.151 0.091 0.141 -0.260 -0.037 -0.155 -0.247 0.003 -0.066 0.032 0.230 1.000
Sector 2 (Sec. 2) 0.055 0.014 0.025 0.216 -0.031 0.178 -0.071 0.039 0.104 -0.038 -0.066 -0.367 1.000
Period 1 (Per. 1) -0.313 -0.116 -0.114 -0.307 0.010 -0.154 0.068 0.004 -0.094 0.008 0.272 0.172 -0.034 1.000
