Do firms that were Private-Equity owned pre-IPO perform operationally better post-IPO than other IPOs? Thesis for MSc Finance

Do firms that were Private-Equity owned pre-IPO

perform operationally better post-IPO than other

IPOs?

Thesis for MSc Finance

Name: Friso de Jong Student number: s2188937

Abstract

This research is focused on whether there is a difference in operational performance of firms after an IPO, if they were Private-Equity owned prior to the IPO or not. Based on IPO data and financial data from 2002-2016, the results of the analyses show that there is a significant difference in operational performance after controlling for leverage, firm size, geography and industry. Firms that were Private-Equity owned prior to an IPO tend to perform better in the short-term (1-2 years) after the IPO. Also in the long-term (3-5 years) the performance of these firms tends to be higher, however the difference in performance clearly decreases over time. These findings prove that Private-Equity ownership leads to an increase in the operational performance of firms.

1. Introduction

Does Private-Equity (PE) ownership lead to an increased operational performance of firms in the long-term? The aim of this research is to answer this question. To answer this question, a research is designed that focusses on whether there is a difference in operational performance of firms after an IPO, based on IPO ownership. The distinction between pre-IPO ownership is: PE owned versus other forms of ownership. This distinction is chosen, because PE ownership results in a certain type of governance form that is different from other forms of firm ownership. Next to the difference in governance form, PE is also trying to improve the operational performance of a firm. These differences may lead to differences in operational performance of firms. However, these differences in operational performance have to be sustainable in order to create value. To check the sustainability of the operational performance, it is measured from 1-5 years after the IPO. The operational performance of firms is measured with an adjusted return on assets (AROA), in which the book value of the assets is adjusted by subtracting the cash and cash equivalents. The dataset consisted of 5,855 IPOs in Europe and the US in the period of 2002-2011.

This leads to regressions of a PE dummy variable on the AROA, controlling for leverage, industry and firm size. The results are that the coefficient of the PE in t+1 = 0.036, t+2 = 0.025, t+3 = 0.020, t+4 = 0.015 and in t+5 = 0.003 (insignificant). These coefficients indicate that the operational performance of a firm that is PE owned pre-IPO is significantly higher in the short-term (1-2 years) after the IPO. For the long-term (3-5 years) the effect is also significantly higher, however the difference in performance decreases over time. These findings prove that PE improves the operational performance of firms. This contributes to current literature and debate about the effect of PE ownership on firms. To understand what this debate is about, a deeper understanding of what PE owners exactly are and how they operate is required.

overhang problem is that the debt burden of a company is so large, that the company is not able to take on anymore debt, even if this debt would be used to invest in profitable projects. This may lead to a downward spiral, which may eventually end in bankruptcy or forgiveness of debt by the creditors. On the other hand, a higher leverage ratio increases the returns for the investor. This can be explained with a simple example: When someone buys a house, he or she generally has to take a mortgage. The price of the house is 200,000 euros, the buyer puts in 50,000 euros of his own money and has to borrow 150,000 euros. A few years later the buyer wants to sell the house and its price has increased to 250,000 euros. Assuming no interest payments had to be made, the seller now has 100,000 euros in cash after paying off the mortgage. The price of the house increased with 25%, but the own capital of the seller has increased with 100%. This mechanism means that the PE investors often want to have a very high leverage ratio, because this increases their returns. Another reason for having large levels of debt, is the use of tax shields because this increases the value of the firm according to Modigliani & Miller (1958, 1963) in their discussion on the theory of capital structure.

According to Opler & Titman (1993) two main questions arise with regard to the debate about LBOs, namely: 1) Do LBOs create wealth or do they merely redistribute wealth? 2) Does the leverage ratio taken on in LBOs cause problems in periods of economic distress? This second question is answered by Opler & Titman (1994) themselves, because in their research they find that high leverage ratios create indirect distress costs that are significant. The first question however, has remained a hot topic over time. Politicians are still pushing for legislation that limits the harming effects of LBOs. In the eyes of the politicians the harming effects of these LBOs mainly concern with how PE investors are shaping companies. Making the companies more efficient can be done in a lot of ways, but it may involve cutting jobs. Which is in the eyes of the politicians a harmful effect, especially when the investors earn a lot of money with these actions.

that was previously owned by PE investors performs worse than its peers, it may indicate that the effects of LBOs are harmful.

2. Theory

directors and contracts with owners and creditors to limit the waste of free cash flow. This is very much in line with what theory says about maximizing value. More recently Harris et al. (2005) have found empirical evidence that in the short-term as well as the long-term there is a substantial increase in productivity after a buyout has taken place. And Cumming et al. (2007) underwrite this by saying that, after having reviewed recent empirical evidence, the incentive and governance mechanisms following from buyouts enhance the performance of firms.

The previous section primarily focused on agency theory and how buyouts function as governance and control mechanism. The following part will focus more on the strategic entrepreneurship view on LBOs. Much of the debate about LBOs has been about the short-term focus of the buyers. Meaning that they tend to cut spending on R&D and capital expenditures in order to increase the short-term gains. This is interesting, because this means that in the short-term it may create value but in the long-run this may lead to value loss. Companies that cut on R&D and capital expenditures may miss critical investments that allow them to remain competitive over time. Eventually this cutting can lead to value destruction in the long-run, which is one of the main critics on LBOs in the debate about whether they create value or not. In this research it is important to be aware of this, because firms that were previously owned by PE are being researched. Thus the PE investor is not the owner of the company anymore. In order for the investor to make a good return on the exit, the PE might focus on increasing short-term profits at the expense of long-term profits. Therefore, according to this reasoning it is likely that after the exit the firm will lose value. However, in order for PE to grow the business during the period in which they are the owners, it is important to invest in the right projects. According to Meuleman et al. (2009) PE can play a major role in fostering this growth, due to various reasons. One reason is for example that the PE has experience in growing businesses from prior investments and thus has more experience in selecting and managing the right projects as discussed by De Clercq & Dimov (2008). This may indicate that PE simply is better at creating value with the right investments than management of these firms. Therefore, cutting on R&D and capital expenditures may not necessarily be bad, if this means that with less expenses PE and management are able to achieve more growth and value.

likely it is that the firm will go public and the less likely it is that the firm will end in bankruptcy or financial restructuring. The research in my paper focusses on IPO exits and the post-IPO performance of companies. Regarding IPO-exits, Strömberg (2007) found that these IPO-exits only accounted for 13% of all the LBO-exits and that this amount has decreased over time. Thus 87% of the PE acquisitions does not lead to an IPO. The most important conclusion following from the research by Strömberg is that he does not find support for the claim that PE ownership leads to short-termism and financial failure, because of the holding periods and relatively low bankruptcy rates. The findings of Strömberg confirm what Jensen (1989) argued, which is that ultimately the LBO organizational form is an optimal governance structure for many firms.

more likely to go public than poor performers. To check if these explanations are valid, they looked at whether the public anticipated this by correcting in the share price, which is the case, because although the RLBOs stocks outperform their peers, the difference in performance is not statistically significant.

The aforementioned research, theories and the ongoing debate lead to the research question: does PE ownership lead to an increased operational performance of firms in the long-term? This is translated into the following hypotheses, in which the first hypothesis is focused on the short-term (1-2 years) and the second hypothesis is focused on the long-term (3-5 years).

Hypotheses:

1. IPOs following from a PE exit perform better than other IPOs in the short-term (1-2 years) after the IPO

2. IPOs following from a PE exit perform better than other IPOs in the long-term (3-5 years) after the IPO

3. Methodology

This section will describe how the research was setup and why it was setup in this way. First the required data will be discussed and how it will be derived. Second the preparation of this data will be described. Third the research design will be discussed and elaborated on, which assumptions have been made and why this is the case. Fourth the necessary calculations are explained. And finally the rationale behind the statistical tests used and how data should be analyzed will be discussed. The data will be collected from the CapitalIQ database, which is in line with research performed by Strömberg (2007) on the demography of PE. The time period for which data will be collected is from 2002-2011 for the IPO data and from 2002-2016 for the financial data.

3.1 Variables

The operational performance of the firm can be measured in many different ways. Barber & Lyon (1996) discussed these different ways of measuring operating performance in their paper extensively, by looking at what different methods were used by researchers when they were measuring operating performance. They conclude that they favor the use of operating income over earnings, because operating income is not influenced by special items and tax considerations. Moreover, the operating income is not influenced by capital structure changes, that may affect the interest expense and thus the earnings. This operating income will have to be scaled in order to be able to compare the operating performance across firms. Again this can be done in many different ways. Scaling it on equity is not an option, because the capital structures of firms may differ significantly and thus may lead to skewed results. Barber & Lyon (1996) conclude in their research that it should be scaled on the average of the beginning-of-the-year and end-of-year book value of the total assets, however many other studies use end-of-period assets. The problem with the use of total assets is that it includes operating and non-operating assets and therefore the measure is not fully accurate. Thus, Barber & Lyon (1996) argue that the total assets should be adjusted by deducting the cash and marketable securities, because these can have a big influence on the total assets. Although some cash is necessary for the operations, most of it is not. The cash and marketable securities will be captured in a variable called cash and cash equivalents (C&CE), because marketable securities are securities that are liquid and thus have the same characteristics as cash. In the research in this paper the end-of-year book value of the assets deducted by cash and cash equivalents will be used. The end-of-year book value of the assets is chosen, because in this way the assets that were used over the year to generate revenue are certain to be incorporated. This leads to a measurement of firm performance being called adjusted return on assets (AROA):

𝐴𝑅𝑂𝐴 = _{'* +,,-.,/0&0&}&'() (1)

Another proxy for the operating performance is the return on sales or ROS. This measure is proposed by Barber & Lyon (1996), because it can detect reductions in selling, general and administrative expenses or improvements in production efficiency. The ROS is measured by:

Next to the main variables it is important to control for certain variables that may influence this AROA. Barber & Lyon (1996) found that many researchers choose to compare firms on industry, which is often done by two- or four-digit SIC codes. It may be required to control for industry, because industries have certain characteristics. These characteristics vary between industries and hence lead to for example differences in AROA among industries. This could influence the analyses and thus dummy control variables for industry are included, based on a one-digit SIC code. This one-digit SIC code will be used, because the level of detail is appropriate regarding the sample size, which is expected to be limited due to the nature of IPOs. IPOs are limited in their occurrence, because there are very demanding requirements for firms to be listed on an index. Moreover, not all firms that are eligible for an IPO are willing to do this. Both of these reasons thus result in a limited amount of IPOs each year. Thus making a less detailed distinction between industries more appropriate, hence the choice for a one-digit SIC code versus a two-digit SIC code. Next to this the researchers also correct for size, since Fama & French (1995) found that smaller firms tend to have lower earnings to book equity than big firms have. There are various arguments on why this is the case. For example: economies of scale, which may lead to larger firms having higher earnings. Or high entry barriers in certain industries, which is in favor of the earnings of larger firms. Size will be measured with the natural log of the book value of the total assets in the IPO year. Controlling for both size and industry is also in line with research performed by Degeorge & Zeckhauser (1993), Kaplan (1989) and Denis & Denis (1993).

Furthermore, another factor that might influence the AROA could be the geographical location of the firm. There are multiple reasons for why the geographical location has an influence on the performance of firms. One reason is for example the economic state of a country. This differs among countries and thus a company from the US might show a difference in performance with a company from Europe. Also access to raw materials or a very good supporting infrastructure, think of universities that deliver highly educated personnel, may influence the firm performance. Therefore, a control variable will be used in this research with regard to the geographical location of the firm. Since data will be chosen from Europe and the US a dummy variable can be included for geographical location, showing a 1 when the firm is from the US and a 0 if the firm is from Europe.

Type of variable Name Measure

Dependent AROA EBIT / (BV Assets – C&CE)

ROS EBIT / Sales

Independent PE Dummy 1 (PE) – 0 (Other)

Leverage Debt / (Debt + Equity)

Firm Size ln (BV Assets)

Geography Dummy 1 (US) – 0 (Europe)

Industry One-digit SIC code dummies

Table 1: Variables overview

3.2 Analyses

The research is based on the influence of the independent dummy variable PE on the dependent variable AROA. The relationship between these variables and the control variables and the dependent variable is expected to be linear, based on theory and the characteristics of the data. Therefore, the main statistical method that will be used in this research is a multiple linear regression model based on OLS. The Newey-West robust standard errors method will be used to help assure that the model is as valid as possible. The following model will be the basis for the analysis:

𝐴𝑅𝑂𝐴_. = 𝐶 + 𝛽₉∗ 𝑃𝐸 + 𝛽₌∗ 𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒 + 𝛽_D∗ 𝐹𝑖𝑟𝑚 𝑆𝑖𝑧𝑒 + 𝛽_I∗ 𝐺𝑒𝑜𝑔𝑟𝑎𝑝ℎ𝑦 + 𝛽_O∗ 𝑆𝐼𝐶1 + 𝛽_R∗ 𝑆𝐼𝐶2 + 𝛽_T∗ 𝑆𝐼𝐶3 + 𝛽_V∗ 𝑆𝐼𝐶4 + 𝛽_X∗ 𝑆𝐼𝐶5 + 𝛽_9Z∗ 𝑆𝐼𝐶6 + 𝛽₉₉∗ 𝑆𝐼𝐶7 +

𝛽₉₌∗ 𝑆𝐼𝐶8 + 𝛽_9D∗ 𝑆𝐼𝐶9 + 𝛽_9I∗ 𝑃𝑜𝑠𝑡 𝑐𝑟𝑖𝑠𝑖𝑠 + 𝜀 (3)

Next to this, variations in the use of variables will be made in order to check the sensitivity of the model. A different proxy for the operating performance is the return on sales (ROS). This will result in the following model, where AROA is replaced by ROS:

𝑅𝑂𝑆_. = 𝐶 + 𝛽₉∗ 𝑃𝐸 + 𝛽₌∗ 𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒 + 𝛽_D∗ 𝐹𝑖𝑟𝑚 𝑆𝑖𝑧𝑒 + 𝛽_I∗ 𝐺𝑒𝑜𝑔𝑟𝑎𝑝ℎ𝑦 + 𝛽_O∗ 𝑆𝐼𝐶1 + 𝛽R∗ 𝑆𝐼𝐶2 + 𝛽T∗ 𝑆𝐼𝐶3 + 𝛽V∗ 𝑆𝐼𝐶4 + 𝛽X∗ 𝑆𝐼𝐶5 + 𝛽9Z∗ 𝑆𝐼𝐶6 + 𝛽99∗ 𝑆𝐼𝐶7 + 𝛽9=∗ 𝑆𝐼𝐶8 +

3.3 Data preparation

The data obtained for this research contained a significant amount of outliers. In order to perform appropriate analyses, the data has to be trimmed. Trimming in this case means removing the outliers. The observations were removed if they were either <-1.0 or >1.0 for the dependent variable AROA. Thus meaning that all observations that have an AROA above 100% or below -100% are removed. These trimming borders are chosen, because values that fall outside of these borders are considered to be non-normal from a theoretical point of view since a firm has to be doing exceptionally well or dramatically bad if the AROA is outside of these borders. Trimming at these border values results in normally distributed AROAs. However, the trimming reduced the sample size from 8785 to 5855, which is a significant difference. Next to this dataset, there are also two other datasets with outliers removed at <-2.0 or ><-2.0 and at <-0.5 or >0.5. These additional datasets will be used for sensitivity analyses. In addition to trimming, the initial dataset has also been winsorized at the 5th and 95th percentile. This method proved to be inadequate, because this did not result in an acceptable normal distribution of the AROA. Thus the results of the analyses that were performed on the dataset were not significant and meaningful. Therefore, the trimmed datasets are chosen for the analyses in the research.

3.4 Data description

Year IPOs PE IPO Overview Other US Europe

2002 184 21 163 150 34 2003 243 29 214 201 42 2004 304 55 249 249 55 2005 453 50 403 247 206 2006 597 78 519 302 295 2007 606 52 554 321 261 2008 429 46 383 237 192 2009 884 89 795 450 434 2010 1088 139 949 515 573 2011 1067 177 890 551 516 Total 5855 736 5119 3223 2608

Table 2: IPO data overview

Description: The first column shows the number of IPOs for each year. The second and third column show the split of PE owned IPO versus Other IPO. Finally the fourth and fifth column show the split of IPOs in the US versus IPOs in Europe

SIC IPOs 0 Agriculture, Forestry & Fishing 46

1 Mining & Construction 998

2 Manufacturing 759

3 Manufacturing 1036

4 Transportation & Public Utilities 673

5 Wholesale & Retail Trade 408

6 Finance, Insurance & Real Estate 956

7 Services 736

8 Services 208

9 Public Administration 35

Total 5855

Table 3: IPO overview per industry 1-digit SIC code Description: IPOs per industry of the dataset from 2002-2011

Table 3 shows an overview of the number of IPOs per industry based on a 1-digit SIC code. As expected the manufacturing industry is an industry in which a lot of IPOs occur. Also the services and mining & construction industries have a large share in the total number of IPOs. Table A3 in the Appendix shows a more detailed split on the 2-digit SIC code level. It is interesting to see that some industries account for a large amount of the IPOs. For example, the Oil and Gas Extraction industry (SIC code 13) and the Business Services industry (SIC code 73) have a large share in the total amount of IPOs.

Graph 1: AROA Through Time based on the data sample

When looking at the leverage ratios of the companies in Appendix B2, it can be seen that over the years the average and median Debt-to-Capital ratios have remained fairly stable. This means that there is no significant influence of the financial crisis on the leverage ratio of companies in the post-crisis period. Thus the low levels of interest in the post-crisis period did not lead to more debt financing in this period, according to the dataset.

In Appendix B3, it can be seen that the firm sizes were fairly stable over the time period of the sample. This is the case for the medians and the averages. However, for the 5th percentile there is a difference in firm size between the pre-crisis period of 2002-2007 and the post-crisis period of 2008-2011. It seems that the 5th percentile firm size border in the post-crisis period is lower than in the pre-post-crisis period, which indicates that there were relatively more small firms having an IPO than in the pre-crisis period. This observation may support the earlier statements made about more IPOs and lower AROAs, because more small firm IPOs may hint at a search for yield by investors. These findings among the data are interesting for future research, which could help in clarifying the exact reasons behind this.

4. Results

The first analyses were done by taking each individual IPO year and running a separate regression for this specific year. However, the results coming from these analyses were not consistent. The significance varied a lot over the years, which made it difficult to draw conclusions from. One possible explanation for these inconsistent results may be that the sample sizes for the individual years are inadequate. In order to increase the sample size, all

0.00 0.02 0.04 0.06 0.08 0.10 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 AR OA Year

AROA Through Time

data was put together. This implies that the AROA in t+3 from one observation may be in 2007 and from a different observation in 2010, which is also the case for the Leverage variable. As mentioned in the methodology part the economic state of a country may distort the results. In this case the economic state of different years could also mean that results will be distorted. The results of the second analyses however prove otherwise, because the significance and adjusted R^2 increased a lot.

AROA t0 t+1 t+2 t+3 t+4 t+5 C 0.013 -0.017 -0.007 0.012 0.008 0.015 PE 0.036*** 0.036*** 0.025*** 0.020*** 0.015** 0.003 Leverage t -0.008 -0.030*** -0.038*** -0.018** -0.020*** -0.026*** Firm Size 0.023*** 0.022*** 0.020*** 0.018*** 0.017*** 0.016*** Geography -0.005 0.004 0.010** 0.009** 0.014*** 0.010*** SIC 1 -0.141*** -0.099*** -0.095*** -0.108*** -0.112*** -0.116*** SIC 2 -0.103*** -0.064*** -0.056*** -0.069*** -0.058*** -0.057*** SIC 3 -0.126*** -0.084*** -0.083*** -0.097*** -0.089*** -0.075*** SIC 4 -0.111*** -0.076*** -0.070*** -0.082*** -0.072*** -0.061*** SIC 5 -0.061*** -0.036** -0.033** -0.050** -0.040** -0.030 SIC 6 -0.109*** -0.080*** -0.073*** -0.086*** -0.075*** -0.066*** SIC 7 -0.081*** -0.047*** -0.050*** -0.059** -0.057*** -0.047** SIC 8 -0.043** -0.016 -0.029* -0.053** -0.033* -0.031 SIC 9 -0.181*** -0.152*** -0.129*** -0.131*** -0.133*** -0.129*** Post-crisis -0.044*** -0.032*** -0.024*** -0.024*** -0.029*** -0.027*** Adj. R^2 0.167 0.163 0.153 0.138 0.142 0.124 Observations 5740 5792 5766 5741 5721 5574

* Significant at 0.10 probability level ** Significant at 0.05 probability level *** Significant at 0.01 probability level Table 4: AROA coefficients

4.1 PE ownership results

The first reason being that firms may be window-dressing the company in order to make it appear better for the IPO. Second, the firm actually shows a better operational performance as a result of the experience and capabilities the PE owner is bringing to the company. This is addressed by Meuleman et al. (2009) and De Clercq & Dimov (2008) in their research.

The main focus of this research however, is the focus on the period after the IPO. In table 4 from t+1 up to t+4 the coefficient of the PE dummy variable is positive and significant. Implying that a firm that is being offered to the public by PE shows a better operational performance than other firms over a longer period of time after the ownership has changed. At t+1 the coefficient is equal to t0, which may hint that the aforementioned window-dressing does not occur on average, because there is not a negative coefficient after the IPO. Also at t+2, t+3 and t+4 the coefficient remains positive, meaning that PE pre-IPO ownership results in a significant higher performance after the IPO. At t+5 the coefficient is positive, but very small and insignificant. Thus at t+5 there is no relationship between PE ownership IPO and the operational performance 5 years after the IPO. The effect of pre-IPO PE ownership on the post-pre-IPO operational performance also reduces gradually from t+1 to t+4. This is in line with the expectations formed in theory. However, the effect is stronger in the short-term than in the long-term. Therefore, we accept the first hypothesis, which states that IPOs following from PE exits have a better operational performance than other IPOs in the short-term (1-2 years) after the IPO. The second hypothesis can only be partly accepted, because there is a positive and significant longer-term effect in t+3 and t+4, but there is no relationship in t+5. Due to the decrease in coefficient and the lack of relationship in t+5 it seems likely that in t+6, t+7, etc. there will also be no relationship between pre-IPO PE ownership and the operational performance after the IPO.

maintain their higher ROIC for a longer period of time are more likely to have a ROIC that will remain above the median ROIC. However, the ROICs will move more towards the median over time. This is consistent with the findings in this research. Finally, there is also the possibility that most of the AROAs in the t+4 and t+5 are in the post-crisis period. As can be seen in graph 1, the median as well as the average of the AROA in the post-crisis period decreased. This may also have had an influence on the coefficients in table 4, although this was the case for all IPOs.

4.2 Control variables

In table 4 the coefficients for the control variables can be found. If we look at the control variable Leverage, at t0 (the year of the IPO) the Leverage coefficient is insignificant meaning that there is no effect of leverage on the AROA. From t+1 up to t+5 there is a clear negative and significant effect of leverage on the AROA. This negative coefficient is in line with the expectations formed in the methodology section, which state there might be a negative effect of leverage on the AROA, because leverage may lead to an increase in the adjusted assets. The other effect of leverage, as discussed in the theory, may be that capital expenditures are being postponed in order to be able to meet short-term interest payments and debt payments. By postponing capital expenditures, the company may run the risk of decreasing operational efficiency and thus its operational performance may go down. This last effect is interesting, because the criticism on PE is that its focus on short-term gains, by cutting on capital expenditures, is hurting the company in the long run. However, in this regression it is unclear whether the effect of leverage on the denominator in the AROA calculation or the effect of leverage on the capital expenditures is causing the negative coefficient of the leverage. The ROS proxy, which is also used to determine the operational performance of the companies, may provide a better understanding of why this is the case, since the ROS is not affected by a possible change in the denominator due to leverage. Therefore, the ROS could help determine which effect is causing the leverage coefficient to be negative. This will be discussed later on, after the analyses with the ROS as dependent variable.

Another control variable is the geography. What is interesting, is that the dummy variable (1 for US, 0 for Europe) is insignificant at t0 and t+1, but becomes significant and positive from t+2 to t+5. This means that in the IPO year and the year after the IPO there is no difference in performance because of geographical location. However, for the following years it seems that the performance of US firms is slightly better than European firms. Why this exactly is the case requires further research, but a possible explanation may be that the bond market in the US provides a better climate for firms to obtain financing than the European bond market. This financing can then be used to invest in projects, resulting in a higher operational performance. However, it takes time for these projects to have an effect, which is why this may be a possible explanation for the results seen in the analyses.

The control variables for the industries all show fairly significant results, except for SIC 8. This may be due to the limited amount of observations in this industry, as can be seen in table 3. Furthermore, the coefficients mostly display a stable pattern over the years, in which the coefficients revolve around a certain a mean. What is interesting is the difference in coefficients between industries. For example, the coefficient of SIC 1 revolves around -0.100, whereas the coefficient of SIC 7 revolves around -0.050. This can be explained by looking at the characteristics of the industries. SIC 1 is the mining & construction industry, which is an asset intensive industry. SIC 7 is the services industry, which does not require a large amount of assets. The difference in asset intensity of the industries, which is a characteristic for the industries, has an influence on the AROA measurement and thus the coefficients differ.

4.3 Sensitivity

To check if the model was robust, changes in the use of variables and different trimming boundaries were used. The results of these analyses were all in line with the earlier performed analyses, as can be seen in Appendix C. Furthermore, the plots of the residuals from the regressions can be found in Appendix D.

4.4 ROS results

Next to the AROA measure, the ROS measure was used as a proxy for operational performance. The results of this analysis can be found in table 5.

ROS t0 t+1 t+2 t+3 t+4 t+5 C 0.064** 0.022 0.033 0.051** 0.013 0.030 PE 0.007 0.013* 0.005 -0.003 -0.002 -0.018** Leverage t -0.006 -0.025*** -0.034*** -0.027*** -0.030*** -0.029** Firm Size 0.024*** 0.021*** 0.018*** 0.015*** 0.016*** 0.015*** Geography 0.000 0.007 0.013** 0.019*** 0.025*** 0.021*** SIC 1 -0.108*** -0.029 -0.015 -0.018 -0.046 -0.093*** SIC 2 -0.142*** -0.079** -0.063*** -0.064*** -0.031 -0.043 SIC 3 -0.186*** -0.110*** -0.104*** -0.110*** -0.079*** -0.084*** SIC 4 -0.096*** -0.032 -0.021 -0.024 0.006 -0.001 SIC 5 -0.161*** -0.108*** -0.099*** -0.105*** -0.069** -0.075** SIC 6 0.075** 0.149*** 0.165*** 0.174*** 0.205*** 0.199*** SIC 7 -0.121*** -0.066* -0.055** -0.058*** -0.025 -0.030 SIC 8 -0.094*** -0.039 -0.037 -0.057** -0.024 -0.032 SIC 9 -0.180*** -0.152*** -0.117*** -0.103*** -0.086** -0.110** Post-crisis -0.024** -0.011 -0.003 -0.004 -0.014 -0.014 Adj. R^2 0.222 0.225 0.231 0.248 0.245 0.238 Observations 5086 5133 5137 5137 5147 5153

* Significant at 0.10 probability level ** Significant at 0.05 probability level *** Significant at 0.01 probability level Table 5: ROS coefficients

IPO PE ownership and post-IPO operational performance according to the ROS measure. Only at t+5 there is a negative and significant relationship between the PE dummy and the ROS, but this coefficient and thus the effect is very small. Thus according to this analysis we would have to reject both of the hypotheses. This result is in sharp contrast with the result found in the analysis with the AROA as dependent variable. Meaning that scaling of the EBIT on sales results in a different outcome than scaling on the adjusted assets. This is very interesting, because these results imply that firms that were PE owned pre-IPO are better at utilizing the operational assets than other IPOs. However, these firms show similar performance in converting sales to EBIT. This could mean that there is no influence of PE ownership on the margins and that these margins may thus be more determined by industry boundaries. Further research is required to determine the exact reasons for why this may be the case.

The control variables Leverage, Firm Size and Geography show very similar results for the ROS as for the AROA. The industry control variables on the other hand do not display comparable results, the significance is much lower and the coefficients are not stable throughout the years. Furthermore, the Post-crisis dummy is also insignificant and thus based on this analysis there is no difference in ROS between the pre-crisis period and the post-crisis period.

5. Conclusion

To have a better understanding of where this difference comes from, a closer look on both measurements is required. The AROA is a measurement that scales the EBIT from the income statement on the adjusted assets from the balance sheet. The ROS is a measurement that scales the EBIT from the income statement on the sales from the same income statement. Thus the AROA measurement focusses on the balance sheet whereas the ROS focusses on the income statement. This implies that PE-ownership leads to differences in the balance sheet in contrast to no differences in the income statement, according to the results of the research. As mentioned in the methodology, the goal was to scale the EBIT on the operating assets, which are essentially: receivables, inventories and property, plant and equipment (PPE). For the AROA to be higher, the denominator in the calculation has to be lower. Thus the adjusted assets have to be lower. For PPE it is really specific whether it can be brought down, because the machinery and the plants are required for the production. It can be brought down by cutting on the capital expenditures, but this will hurt the operations in the long-term since critical maintenance expenditures may be reduced in this way. Receivables and inventories however, are part of the working capital. By streamlining the operations, the inventory levels can be lowered and by putting pressure on the receivable policies this amount can also be lowered. These are areas that can be made more efficient with the help of experience and expertise, which is what PE ownership can bring to a company as argued by De Clerq & Dimov (2008).

The reason why PE is focused on this, is because of the following simple equation:

𝑂𝑝𝑒𝑟𝑎𝑡𝑖𝑛𝑔 𝐴𝑠𝑠𝑒𝑡𝑠 − 𝑂𝑝𝑒𝑟𝑎𝑡𝑖𝑛𝑔 𝐿𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑖𝑒𝑠 = 𝐼𝑛𝑣𝑒𝑠𝑡𝑒𝑑 𝐶𝑎𝑝𝑖𝑡𝑎𝑙 = 𝐷𝑒𝑏𝑡 + 𝐸𝑞𝑢𝑖𝑡𝑦 (5)

PE wants to minimize the invested capital, because this means that they have to invest less in the company, which results in higher rates of return. PE achieves this by focusing on the working capital of a company. Meaning that they focus on lowering the operating assets and increasing the operating liabilities. Increasing the operating liabilities is done by increasing the accounts payable via longer payment terms with suppliers. An important aspect of this is management, because they are key in achieving this. Thus explaining the vital role that the governance system, as discussed by Jensen (1989), plays in PE ownership.

goods sold and the selling, general and administrative expenses. This may imply that the critics in the current debate on PE, about PE cutting costs too extensively, are not supported by the findings in this research.

This leads to the conclusion being that pre-IPO PE ownership leads to an increased operational performance post-IPO, because the AROA is significantly higher in the short-term although its effect decreases over time. The sustainability of the increased performance however, is questionable. This implies that PE ownership is required to maintain the increased performance over the long-term. Therefore, PE may be regarded as a better form of firm ownership than public ownership.

References

Barber, B. M., & Lyon, J. D. (1996). Detecting abnormal operating performance: The empirical power and specification of test statistics. Journal of Financial Economics, 41(3), 359-399.

Bargeron, L. L., Schlingemann, F. P., Stulz, R. M., & Zutter, C. J. (2008). Why do private acquirers pay so little compared to public acquirers? Journal of Financial Economics, 89(3), 375-390.

Cumming, D., Siegel, D. S., & Wright, M. (2007). Private equity, leveraged buyouts and governance. Journal of Corporate Finance, 13(4), 439-460.

De Clercq, D., & Dimov, D. (2008). Internal knowledge development and external knowledge access in venture capital investment performance. Journal of Management Studies, 45(3), 585-612.

Degeorge, F., & Zeckhauser, R. (1993). The reverse LBO decision and firm performance: Theory and evidence. The Journal of Finance, 48(4), 1323-1348.

Denis, D. J., & Denis, D. K. (1993). Managerial discretion, organizational structure, and corporate performance: A study of leveraged recapitalizations. Journal of Accounting and Economics, 16(1-3), 209-236.

Fama, E. F., & French, K. R. (1995). Size and book-to-market factors in earnings and returns. The Journal of Finance, 50(1), 131-155.

Fama, E. F., & Jensen, M. C. (1983). Separation of ownership and control. The journal of Law and Economics, 26(2), 301-325.

Harris, R., Siegel, D. S., & Wright, M. (2005). Assessing the impact of management buyouts on economic efficiency: Plant-level evidence from the United Kingdom. Review of

Holthausen, R. W., & Larcker, D. F. (1996). The financial performance of reverse leveraged buyouts. Journal of Financial Economics, 42(3), 293-332.

Jain, B. A., & Kini, O. (1994). The post-issue operating performance of IPO firms. The Journal of Finance, 49(5), 1699-1726.

Jensen, M. C. (1989). Eclipse of the public corporation. Harvard Business Review 67(5), 61-74

Kaplan, S. (1989). The effects of management buyouts on operating performance and value. Journal of Financial Economics, 24(2), 217-254.

Koller, T., Goedhart, M., & Wessels, D. (2010). Valuation: measuring and managing the value of companies (Vol. 499). John Wiley and Sons.

Meuleman, M., Amess, K., Wright, M., & Scholes, L. (2009). Agency, strategic

entrepreneurship, and the performance of private equity-backed buyouts. Entrepreneurship Theory and Practice, 33(1), 213-239.

Modigliani, F., & Miller, M. H. (1958). The cost of capital, corporation finance and the theory of investment. The American Economic Review, 261-297.

Modigliani, F., & Miller, M. H. (1963). Corporate income taxes and the cost of capital: a correction. The American Economic Review, 53(3), 433-443.

Myers, S. C. (1977). Determinants of corporate borrowing. Journal of financial economics, 5(2), 147-175.

Opler, T., & Titman, S. (1993). The determinants of leveraged buyout activity: Free cash flow vs. financial distress costs. The Journal of Finance, 48(5), 1985-1999.

Rajan, R. G., & Zingales, L. (1995). What do we know about capital structure? Some evidence from international data. The Journal of Finance, 50(5), 1421-1460.

Appendix A

IPO Overview

Year IPOs PE Other US Europe

2002 186 21 165 151 35 2003 247 30 217 201 46 2004 311 55 256 251 60 2005 467 51 416 253 214 2006 611 80 531 307 304 2007 625 53 572 333 268 2008 454 53 401 252 202 2009 922 96 826 469 453 2010 1130 147 983 524 606 2011 1105 185 920 567 538 Total 6058 771 5287 3308 2726

Table A1: IPO data overview (trimmed at 200%)

Description: The first column shows the number of IPOs for each year. The second and third column show the split of PE owned IPO versus Other IPO. Finally the fourth and fifth column show the split of IPOs in the US versus IPOs in Europe

IPO Overview

Year IPOs PE Other US Europe

2002 178 20 158 146 32 2003 233 25 208 192 41 2004 287 52 235 237 50 2005 429 43 386 233 196 2006 563 74 489 290 273 2007 570 48 522 308 240 2008 402 42 360 227 175 2009 829 81 748 430 399 2010 1009 129 880 484 525 2011 986 156 830 516 470 Total 5486 670 4816 0 3063 2401

Table A2: IPO data overview (trimmed at 50%)

A3 – IPO per industry based on a two-digit SIC code (2002-2011)

SIC IPOs

01 Agricultural Production - Crops 21

02 Agricultural Production - Livestock and Animal Specialties 0

07 Agricultural Services 1

08 Forestry 0

09 Fishing, Hunting and Trapping 1

10 Metal Mining 293

12 Coal Mining 42

13 Oil and Gas Extraction 506

14 Mining and Quarrying of Nonmetallic Minerals, Except Fuels 23 15 Construction - General Contractors & Operative Builders 63 16 Heavy Construction, Except Building Construction, Contractor 49

17 Construction - Special Trade Contractors 22

20 Food and Kindred Products 146

21 Tobacco Products 11

22 Textile Mill Products 17

23 Apparel, Finished Products from Fabrics & Similar Materials 34

24 Lumber and Wood Products, Except Furniture 24

25 Furniture and Fixtures 15

26 Paper and Allied Products 47

27 Printing, Publishing and Allied Industries 64

28 Chemicals and Allied Products 347

29 Petroleum Refining and Related Industries 54

30 Rubber and Miscellaneous Plastic Products 37

31 Leather and Leather Products 11

32 Stone, Clay, Glass, and Concrete Products 51

33 Primary Metal Industries 76

34 Fabricated Metal Products 67

35 Industrial and Commercial Machinery and Computer Equipment 219 36 Electronic & Other Electrical Equipment & Components 269

37 Transportation Equipment 121

38 Measuring, Photographic, Medical, & Optical Goods, & Clocks 154

39 Miscellaneous Manufacturing Industries 31

40 Railroad Transportation 22

41 Local & Suburban Transportation 20

42 Motor Freight Transportation 17

43 United States Postal Service 0

44 Water Transportation 84

45 Transportation by Air 62

46 Pipelines, Except Natural Gas 11

47 Transportation Services 19

48 Communications 221

49 Electric, Gas and Sanitary Services 217

SIC IPOs

51 Wholesale Trade - Nondurable Goods 76

52 Building Materials, Hardware, Garden Supplies & Mobile Homes 16

53 General Merchandise Stores 28

54 Food Stores 10

55 Automotive Dealers and Gasoline Service Stations 35

56 Apparel and Accessory Stores 37

57 Home Furniture, Furnishings and Equipment Stores 10

58 Eating and Drinking Places 55

59 Miscellaneous Retail 64

60 Depository Institutions 8

61 Nondepository Credit Institutions 16

62 Security & Commodity Brokers, Dealers, Exchanges & Services 84

63 Insurance Carriers 152

64 Insurance Agents, Brokers and Service 13

65 Real Estate 210

67 Holding and Other Investment Offices 473

70 Hotels, Rooming Houses, Camps, and Other Lodging Places 18

72 Personal Services 13

73 Business Services 554

75 Automotive Repair, Services and Parking 21

76 Miscellaneous Repair Services 0

78 Motion Pictures 35

79 Amusement and Recreation Services 95

80 Health Services 92

81 Legal Services 0

82 Educational Services 14

83 Social Services 5

84 Museums, Art Galleries and Botanical and Zoological Gardens 0

86 Membership Organizations 0

87 Engineering, Accounting, Research, and Management Services 97

88 Private Households 0

89 Services, Not Elsewhere Classified 0

91 Executive, Legislative & General Government, Except Finance 0

92 Justice, Public Order and Safety 0

93 Public Finance, Taxation and Monetary Policy 0

94 Administration of Human Resource Programs 0

95 Administration of Environmental Quality and Housing Programs 0

96 Administration of Economic Programs 0

97 National Security and International Affairs 0

99 Nonclassifiable Establishments 35

Appendix B

B1 – AROA Tables

Description: Each AROA table is based on the IPO year indicated in the table. The AROAs are displayed from t0, which is the IPO year, up to t+5. The tables are based on the dataset that is trimmed at <-1.0 and >1.0.

B2 – Leverage Tables

Description: Each Leverage table is based on the IPO year indicated in the table. The Leverages are displayed from t0, which is the IPO year, up to t+5. The tables are based on the dataset that is trimmed at <-1.0 and >1.0.

B3 – Firm Size Tables

Description: The first table shows the firm size of the firms at t+1 based on their IPO year. The second table shows this for IPO year periods.

Appendix C Sensitivity tables AROA t0 t+1 t+2 t+3 t+4 t+5 C -0.020 -0.042** -0.028* -0.008 -0.015 -0.004 PE 0.032*** 0.034*** 0.024*** 0.018** 0.012* 0.001 Leverage t -0.011 -0.032*** -0.040*** -0.018** -0.022*** -0.027*** Firm Size 0.023*** 0.022*** 0.020*** 0.018*** 0.017*** 0.016*** Geography 0.000 0.008** 0.014*** 0.013*** 0.018*** 0.013*** SIC 1 -0.131*** -0.096*** -0.094*** -0.107*** -0.110*** -0.115*** SIC 2 -0.089*** -0.057*** -0.052*** -0.065*** -0.053*** -0.053*** SIC 3 -0.112*** -0.077*** -0.079*** -0.093*** -0.083*** -0.071*** SIC 4 -0.097*** -0.069*** -0.066*** -0.078*** -0.066*** -0.056*** SIC 5 -0.047*** -0.029* -0.029* -0.046* -0.034** -0.025 SIC 6 -0.096*** -0.072*** -0.069*** -0.082*** -0.069*** -0.061*** SIC 7 -0.068*** -0.041*** -0.046*** -0.055** -0.052*** -0.043** SIC 8 -0.025 -0.007 -0.024 -0.048* -0.026 -0.024 SIC 9 -0.163*** -0.142*** -0.123*** -0.124*** -0.124*** -0.122*** Crisis t -0.029*** 0.001 0.010 0.008 0.005 0.011 Adj. R^2 0.155 0.153 0.147 0.132 0.132 0.116 Observations 5740 5792 5766 5741 5721 5574

* Significant at 0.10 probability level ** Significant at 0.05 probability level *** Significant at 0.01 probability level

AROA t0 t+1 t+2 t+3 t+4 t+5 C -0.092*** -0.085*** -0.075*** -0.068*** -0.068*** -0.054*** PE 0.046*** 0.045*** 0.032*** 0.027*** 0.023*** 0.014** Leverage t 0.002 -0.023*** -0.032*** -0.015** -0.012* -0.020** Firm Size 0.023*** 0.022*** 0.020*** 0.018*** 0.018*** 0.016*** Geography -0.010** 0.000 0.007 0.006 0.008** 0.003 Post-crisis -0.047*** -0.034*** -0.026*** -0.025*** -0.032*** -0.031*** Adj. R^2 0.141 0.143 0.134 0.119 0.117 0.093 Observations 5740 5792 5766 5741 5721 5574

* Significant at 0.10 probability level ** Significant at 0.05 probability level *** Significant at 0.01 probability level

Table C2: Results for the AROA regression without industry control variables AROA t0 t+1 t+2 t+3 t+4 t+5 C -0.064 -0.039** -0.038** -0.017 -0.003 0.022 PE 0.023** 0.030*** 0.019* 0.018** 0.008 0.002 Leverage t 0.002 -0.037*** -0.051*** -0.023* -0.030*** -0.031** Firm Size 0.035*** 0.033*** 0.028*** 0.026*** 0.023*** 0.023*** Geography -0.024*** -0.013** -0.002 -0.002 0.006 0.002 SIC 1 -0.129*** -0.143*** -0.113*** -0.131*** -0.136*** -0.172*** SIC 2 -0.123** -0.132*** -0.096*** -0.102*** -0.093*** -0.126*** SIC 3 -0.139*** -0.146*** -0.113*** -0.125*** -0.117*** -0.141*** SIC 4 -0.112** -0.123*** -0.088*** -0.103*** -0.091*** -0.112*** SIC 5 -0.046 -0.070*** -0.042** -0.060** -0.054*** -0.077** SIC 6 -0.098** -0.119*** -0.086*** -0.098*** -0.089*** -0.113*** SIC 7 -0.065 -0.084*** -0.052*** -0.061** -0.057*** -0.083** SIC 8 -0.044 -0.056** -0.048** -0.078*** -0.056*** -0.081** SIC 9 -0.209*** -0.244*** -0.197*** -0.181*** -0.177*** -0.221*** Post-crisis -0.059*** -0.042*** -0.032*** -0.028*** -0.036*** -0.036*** Adj. R^2 0.161 0.162 0.143 0.139 0.137 0.128 Observations 5926 5985 5954 5923 5900 5749

* Significant at 0.10 probability level ** Significant at 0.05 probability level *** Significant at 0.01 probability level

AROA t0 t+1 t+2 t+3 t+4 t+5 C 0.040** 0.002 0.007 0.013 0.027* 0.033* PE 0.027*** 0.029*** 0.018*** 0.014*** 0.011** 0.004 Leverage t -0.013** -0.023*** -0.022*** -0.014** -0.018*** -0.014** Firm Size 0.015*** 0.015*** 0.013*** 0.012*** 0.012*** 0.010*** Geography 0.008** 0.011*** 0.016*** 0.018*** 0.019*** 0.014*** SIC 1 -0.111*** -0.069*** -0.067*** -0.071*** -0.092*** -0.101*** SIC 2 -0.058*** -0.022 -0.018 -0.021 -0.028* -0.028 SIC 3 -0.073*** -0.036** -0.039*** -0.046*** -0.056*** -0.052*** SIC 4 -0.081*** -0.047*** -0.045*** -0.047*** -0.055*** -0.050*** SIC 5 -0.039** -0.014 -0.011 -0.017 -0.026* -0.023 SIC 6 -0.088*** -0.057*** -0.055*** -0.056*** -0.064*** -0.060*** SIC 7 -0.059*** -0.024* -0.025** -0.026 -0.036** -0.033* SIC 8 -0.036** -0.009 -0.015 -0.022 -0.028* -0.024 SIC 9 -0.116*** -0.092*** -0.080*** -0.072*** -0.087*** -0.081*** Post-crisis -0.031*** -0.023*** -0.016*** -0.016*** -0.020*** -0.017*** Adj. R^2 0.163 0.158 0.145 0.144 0.148 0.134 Observations 5385 5433 5411 5392 5374 5243

* Significant at 0.10 probability level ** Significant at 0.05 probability level *** Significant at 0.01 probability level

Appendix D

Residual analysis

Graph 1: Residual histogram at t0

Graph 2: Residual histogram at t+1

Graph 3: Residual histogram at t+2

Graph 4: Residual histogram at t+3

Graph 5: Residual histogram at t+4

Graph 6: Residual histogram at t+5