Does Private Equity Stir Up Industries?
REINDER LUBBERS S1409360
Supervisor: Dr. W. Westerman
University of Groningen Faculty of Economics and Business
MSc Business Administration Specialization: Finance
2
Does Private Equity Stir Up Industries?
REINDER LUBBERS*
ABSTRACT
Private equity is often the subject of public debate regarding its impact on economies. While several papers have documented the effects of private equity on a firm level, the effects of private equity on an industry level seem to be overlooked. This paper analyzes the influence of private equity on industry performance across twelve European countries. I find that the relative investment level of private equity positively influences the industries’ productivity, operating income, number of employees and average wage level. Causality tests shows that there exists no reverse causality: the level of private equity investments causes the change in annual growth rates of the industries and not vice versa.
I. Introduction
As the private equity market has grown steadily since the 1970s in size and number of company buyouts, the average industry holding share of private equity (PE) across the globe, has been growing ever since1. PE firms seem able to attract an increasing amount of funding from large funds like pension and insurance funds, due to their capability to often earn higher operating returns compared to hedge funds and non-buyout firms (Strömberg (2008)). The continuing rise of private equity has attracted attention from researchers, the media and policy makers. Policy makers and labor unions have criticized private equity involvement, for its supposed excessive leverage, short investment horizon and cost cutting. Often, the debate is
* I thank Dr. Wim Westerman for being as good of a supervisor as I could wish for, Bart Veldman for helpful comments, my parents Steven Lubbers and Lieke van Noortwijk for their support, Leo and Johanna Boersen for their support and Arieke Boersen for being there for me.
1
3
filled with individual examples and less with scientific research (Rasmussen (2008)). The subprime crisis has drawn even more criticism to private equity as regulators are turning their attention to the private equity market, recently resulting in adoption of more stringent regulations regarding the private equity market by the European Commission. The effect of PE investments on firm level is well documented by several papers, which often find positive results (Kaplan and Strömberg (2009)), but the effect of PE investments on industry level has, to the best of my knowledge, only been thoroughly analyzed in one paper (Bernstein et al. (2010)). As the private equity market is growing, it makes sense to widen our research scope from firm level to industry level. Intra-industry spillover effects which would benefit the entire industry, could well exist when we combine two prior observations: First, firms that are acquired by PE funds become better managed than their competitors (Bloom et al. (2009)). Second, competitors of acquired firms copy the governance mechanisms of their acquired rivals (Oxman and Yildrim (2008)). If these two ideas hold, they could open up new paths for scientific research. Policy makers could also benefit from the results of scientific research on the effect of PE investment on industry level.
The aim of this paper is to analyze whether private equity investments have a beneficial effect on an industry as a whole. To do so, I analyze a number of industry variables across 12 countries for 35 different industries between 2001 and 2009. I use data on merger and acquisition deals, financed by private equity, to construct a level of PE investment, relative to the industry size. I then run ordinary least squares (OLS) regressions with the annual growth rates of the industries, on productivity, operating income and employment variables as dependent variable, and the level of PE investment as the independent variable. Unlike Bernstein et al. (2010), I focus solely on continental Western Europe, for which I analyze a different set of industry variables on a more detailed level of industry, with the crisis years of 2008 and 2009 included.
4
average wage level is an exception however: all PE industries experience a higher annual growth in wages than non-PE industries, but there are no differences found between the industries with various levels of PE investment. The main results differ from the results from Bernstein et al. (2010), who find that PE industries experience higher annual growth in production and employment measures than non-PE industries. However, they do not find differences in annual growth between different levels of PE investment. I analyze the crisis years of 2008 and 2009 separately as a sensitivity test: industries with PE investments seem to perform worse than non-PE industries, but all results herein are insignificant.
In the analyses in this paper, the annual growth rates are adjusted by country-industry, country-year and industry-year fixed effects, so that the growth rates measured are relative to their country, industry and year. This way I control for economic fluctuations between the countries, industries and years. The level of PE investment is calculated by dividing the total deal size of all PE investments, invested in that industry in the four previous years, by the enterprise value of the industry in that specific year and country.
The results do not seem to be the result of reverse causality. Granger causality tests show that the level of PE investments causes the change of the industries’ growth rates and that it is not vice versa.
The paper proceeds as follows. Section II describes the available literature regarding this research question. Section III presents the methodology used. Section IV presents the data. Section V presents the results. I conclude and provide recommendations in Section VI.
II. Literature review
5
the ability of these private equity firms to create value. This can be done in numerous ways, for instance by bringing in outside debt in order to lower the cost of capital, aligning management incentives properly, active monitoring, bringing in human capital to improve operations etc.
Following the first LBO boom of the 1980s, Jensen (1989) argued that publicly held corporations had outlived their usefulness in many sectors of the economy because of conflicting interests between owners and managers over free cash flow. LBO’s would implement superior governance mechanisms, which would align the conflicting interests and benefit the firm, alongside improved capital structures and corporate governance. Jensen’s theory proved premature: several high-profile LBO’s of the 1990s turned into failures and investment bank Drexel Burnham Lambert filed for bankruptcy, which caused the PE market to shrink drastically. The internet bubble of 2000 halted a revival of private equity as several telecommunication companies, with PE funds invested in them, filed for bankruptcy. Following the internet bubble, the US and Euribor interest rates were decreased in attempts to boost the economy2. This made borrowing attractive again and the private equity market recovered to experience a second LBO boom in the mid 2000’s (Kaplan and Strömberg 2008). This wave of increased private equity activity was again halted, this time by the subprime crisis of 2008.
Both LBO periods have been analyzed mostly on a firm level: a vast number of papers have studied the effect of PE investment on the performance of acquired firms, but the effect of PE investment on industry performance has, to the best of my knowledge, only been thoroughly analyzed by Bernstein et al. (2010). In order to analyze possible effects, most papers analyze a variety of measures like productivity, operating income and employment (Kaplan and Strömberg (2009)). Tables I, II and III present overviews of papers that analyze productivity, operating income and employment measures respectively.
As you can see from table I, all papers find positive changes in productivity following LBOs on firm level (Amess (2002), Harris et al. (2005), Meuleman et al. (2009), Jelic and Wright (2011)). Unfortunately, all of these papers analyze the UK. Bernstein et al. (2010) find that industries in which PE funds invest, experience higher growth rates on production, but
2
6 Table I
Overview of papers on private equity and productivity
Authors Countries Period Analysis level Findings
Amess (2002) UK 1986-1997 Firm Production increases via monitoring
Harris et al. (2005) UK 1994-1998 Plant Production increases after LBO Meuleman et al.
(2009)
UK 1993-2003 Firm Production increases depending on type of buyout
Bernstein et al. (2010)
Global 1991-2007 Industry Industries in which PE funds invest experience higher growth rates in
productivity. Growth rates are not influenced by the level of PE investment
Jelic and Wright (2011)
UK 1980-2009 Firm Production increases depending on type of buyout
Table II
Overview of papers on private equity and operating income
Authors Countries Period Analysis level Findings
Kaplan (1989) US 1980-1986 Firm Operating income increases to 36% above
industry median via improved incentives Desbrières and
Schatt (2002)
France 1988-1994 Firm Operating incomes decreases after LBO
Cressy et al. (2007) UK 1995-2002 Firm Operating income of bought out firms is 4,5% higher than rivals, three years after buyout
Phalippou and Gottschalg (2009)
US 1980-2003 Firm PE funds experience lower operating income than the S&P 500
Brown et al. (2009) US 1980-2001 Firm Operating income increases after LBO Meuleman et al.
(2009)
UK 1993-2003 Firm No subsequent change in operating income following LBO
Guo et al (2011) US 1990-2006 Firm Slight increase in operating income for buyouts vs. non-buyouts. CEO replacement following buyout increases operating income.
Jelic and Wright (2011)
7 Table III
Overview of papers on private equity and employment
Authors Countries Period Analysis level Findings
Lichtenberg and Siegel (1990)
US 1981-1986 Plant Reduction in non-production personnel. No effect on production personnel. Negative effect on wages non-production personnel
Bruining et al. (2005)
UK and the Netherlands
1992-1998 Firm Increase in number of employees and wages
Amess and Wright (2007)
UK 1999-2004 Firm No effect on number of employees,
negative effect on wages.
Wright et al. (2007) UK 1997-2006 Firm Slight positive effect on employment and wages
Bernstein et al. (2010)
Global 1991-2007 Industry Industries where PE funds invest in experience higher growth rates in number of employees and wages. Growth rates are not influenced by the level of PE
investment
Davis et al. (2011) US 1980-2005 Firm Mixed results on number of employees
report no differences between industries with different levels of PE investment.
Findings on operating income following LBO’s are less clear cut. As table II shows, most papers find backed firms to earn higher rates of operating income compared to non PE-backed firms (Kaplan (1989), Desbrières and Schatt (2002), Cressy et al. (2007), Brown et al. (2009), Guo et al (2011)). Meuleman et al. (2009) and Jelic and Wright (2011) however, do not find a positive relationship between PE investment and operating income. Bernstein et al. (2010) do not analyze operational income measures, therefore there are no known results on industry level.
8
declining wages, while Bruining et al. (2005), Wright et al. (2007) and Bernstein et al. (2010) find wages to grow faster following PE-backed buyouts.
As all three tables show, most research on private equity has focused on the UK and the US (Cumming et al. (2007). These are indeed the countries with the highest numbers on buyouts and total deal value worldwide (Gilligan and Wright (2010)). With the exception of some papers that performed analyses at plant level and a single paper that performed analysis on industry level, all papers perform analyses on firm level. All papers show positive effects on productivity, most papers show positive effects on operating income and the results for employment seem to be mixed, with a small majority reporting positive effects on the number of employees and wage level.
What could we expect to see when we shift our focus from firm level to industry level? For one, there is the direct effect of PE investments into firms. If firms benefit directly from PE investment and when a significant part of the industry is owned by private equity, then we should be able to see a direct effect between PE investments and industry performance. The PE average industry share however, is probably not big enough to warrant a significant effect on industry performance (Bernstein et al. (2010)). Then there could be a passive effect: the private equity investments in the industry could cause the competitors of the newly acquired firm to experience increased competition, which could consequently trigger these companies to improve their own operations. Oxman and Yildrim (2008) find that firms change their governance mechanisms to emulate firms that are bought out within their industry. Combining this with the finding by Bloom et al. (2009) that PE-backed firms are better managed than non PE-backed firms, shows that industries could be indirectly affected by PE investments. As this paper measures the industry performance as a whole, it will be difficult to point out the nature and size of possible spillover effects.
9
of the highest quality. Furthermore, they find the PE industries to be less exposed to economic cycles. They also test for reverse causality, to prove that the industries were affected by the PE deals, not vice versa. The results suggest that there are no signs of reverse causality.
A. Hypotheses
I analyze industry performance by measuring productivity, operating income and employment variables. I do not test for capital expenditures, since my industry database does not supply a true measure for it. As nearly all papers focus on the UK and US, I analyze continental Western Europe, consisting of twelve countries, for the years of 2001 to 2009. Taking the previous literature, that focuses on firm level, and Bernstein et al. (2010) into account, I expect to find that PE investments positively influence the industries’ performance and operating income. I am not sure about the effect on employment: Bernstein et al. (2010) do find positive results but the papers that analyze at firm level do not.
This leads to the following hypotheses to be tested,
: Private equity investments have a positive effect on industry performance in the four years after the completion of the deal.
with the null hypothesis being that PE investments do not have a positive effect on industry performance. I will test for reverse causality using Granger causality tests (Bernstein et al. (2010)).
: There is no reverse causality: private equity investments cause a subsequent effect on industry performance and not the other way around.
with the null hypothesis being that reverse causality exists.
10
III. Methodology
I measure the effect of PE investments on the yearly performance of industries. To analyze the performance of industries, I construct country-industry-year observations, for instance: the Netherlands - transport and storage - 2004. I measure the annual growth rate for every variable per observation, by calculating the natural logarithm (ln) of the simple return plus one (Phalippou and Gottschalg (2009), Ivashina and Kovner (2011), Jelic and Wright (2011)). I use natural logarithms in order to diminish the importance of outliers. Then I need to know how much private equity is invested during the observation, so that I can analyze the possible relationship between PE investments in an industry and the subsequent performance.
A. Data sources
I obtain the observations and the corresponding industry data from the OECD’s STAN database3 (STAN) (Bernstein et al. (2010)). STAN provides detailed information about industrial performance at a relatively detailed level of activity. It contains data on measures for productivity, labor input, operating performance, invested capital and more. Through the use of the same industry classification code, it is easy to compare industry data across countries.
I obtain the PE deals from the Zephyr database4 (Amess et al. (2008), Huyghebaert and Luypaert (2010)), which contains detailed information on more than 800.000 M&A deals worldwide, with integrated detailed company information. Compared to the SDC Platinum database of Thomson Financial and Mergerstat, the Zephyr database covers deals of smaller value and has a better coverage of European transactions (Huyghebaert and Luypaert (2010)).
In order to calculate the relative influence of private equity on the country-industry-year observations, the total deal value of the PE deals needs to be divided by the total enterprise value of the observation (Bernstein et al. (2010)). STAN however, does not provide data on industry enterprise value. We could divide the total deal value by the industry’s revenue but as Baker and Ruback (1999) show, each industry has different enterprise value/revenue
3
See https://stats.oecd.org/Index.aspx
4
11
ratio’s, as industries differ in for instance levels of leverage and operating margins. Therefore, I construct a moderating variable per industry which is calculated by dividing enterprise value by revenue. I obtain the necessary data from the Orbis database5 and construct industry-year variables, as there is too little data to construct a reliable multiple for each country. Multiplying this multiple with an observation’s revenue will provide us with its enterprise value, with which the total deal value needs to be divided by:
(1)
Where PEI is the level of PE investment per observation, TDEAL is the total PE deal value invested in the industry, EVREV is the industry’s enterprise value/revenue ratio and REV is the revenue per observation from STAN.
TDEAL consists of PE deals that are completed in the four years prior to the observation (Kaplan (1989), Desbrières and Schatt (2002), Cressy et al. (2007)), due to the limited amount of years that PE funds stay invested, before they look for an exit (Kaplan and Schoar (2005)). It may be conservative to only include the deals completed in the past four years as a 2009 Roland Berger PE survey6 speaks of an average holding period of five years and Strömberg (2008) even speaks of an average of five to seven years. I choose four years however, to be sure that I am not including deal amounts that are already divested by the year of the observation.
B. Industry variables
Table IV shows the differences in STAN variables used in this analysis and in that of Bernstein et al. (2010). Appendix table A1, shows the definitions of the STAN variables.
B.1. Productivity variables
Unfortunately, Bernstein et al. (2010) do not clearly explain why they choose the variables
12 Table IV
STAN variables
This table lists the OECD’s STAN database variables analyzed in this paper, as well as literature that have analyzed the same economic variables regarding PE investments.
Type of variables Variables analyzed in this paper Analyzed in existing literature
1) Productivity - Value added (VALU) - Phalippou and Gottschalg (2009), Bernstein et al. (2010)
- Production (gross output) (PROD)
- Harris et al. (2005), Bernstein et al. (2010), Jelic and Wright (2011)
2) Operating performance - Gross operating surplus & mixed income (GOPS)
- Kaplan (1989), Kaplan and Strömberg (2009), Guo et al. (2011)
- Net operating surplus & mixed income (NOPS)
- Ivashina and Kovner (2011)
- Gross operating surplus & mixed income / Net capital stock (GOPS) / (CPNK)
- Kaplan (1989), Guo et al. (2011)
- Net operating surplus & mixed income / Net capital stock (NOPS) / (CPNK)
- Cressy et al. (2007), Ivashina and Kovner (2011), Jelic and Wright (2011)
3) Employment
- No. of employees (excl. self- employed) (EMPE)
- Wright et al. (2007), Amess et al. (2008), Davis et al. (2011), Bernstein et al. (2010), Jelic and Wright (2011)
- Labor costs (LABR) / No. of employees (excl. self-employed) (EMPE)
- Lichtenberg and Siegel (1990), Wright et al. (2007), Amess et al. (2008), Davis et al. (2011)
13
GDP’s, as measured by the United Nations System of National Accounts (p. 103)7. I will therefore measure value added but also production, so that the results of the latter can be compared to the results find by Bernstein et al. (2010).
B.2. Operating performance variables
Productivity and value added do not tell us whether the growth rate of actual operating income is influenced by the level of PE investment. I measure operating income via two variables: earnings before interest, tax, depreciation and amortization (EBITDA) and earnings before interest and tax (EBIT), since both are of more concern than outright output, as most papers analyzing PE deals show (e.g. Kaplan (1989), Kaplan and Strömberg (2009), Guo et al. (2011), Ivashina and Kovner (2011)). STAN provides two fitting measurements: gross operating surplus & mixed income (GOPS), which resembles EBITDA and net operating surplus & mixed income (NOPS), which resembles EBIT. Both are often denoted as operating income (Kaplan (1989), Cressy et al. (2007), Kaplan and Strömberg, (2009)). Finally, I divide both EBITDA and EBIT by the amount of invested capital, as these are seen as indicators for relative company performance (e.g. Kaplan (1989), Long and Ravenscraft (1993), Guo et al. (2011), Ivashina and Kovner (2011)). The closest aggregate to invested capital STAN provides, is the variable net capital stock (CPNK)8, which reflects the market value of capital assets, excluding financial assets.
B.3. Employment variables
To analyze the effect of PE investments on employment effects, I measure the growth rates of the number of employees (Bernstein et al. (2010), Jelic and Wright (2011)) and the amount of the wages (Amess et al. (2008), Davis et al. (2011)). STAN provides the variables number of employees (EMPE) for the first, and labor costs (LABR), which encompasses all forms of compensation of the employees, for the latter. The amount of wages is calculated by dividing labor costs by number of employees for all observations where data on both variables are available. Measuring the number of employees in full-time equivalents (FTE’s) would
7
See http://unstats.un.org/unsd/nationalaccount/docs/SNA2008.pdf
8
14
negate changing levels of part-time work across years, but this variable is unfortunately only available for five out of twelve sample countries. I therefore use the variable EMPE. Bernstein et al. (2010) analyze the growth rates of both labor costs and number of employees, but not of wages.
C. Method of calculation
To ultimately measure differences in performances between the observations, I divide them into different groups to analyze, according to the PEI of each observation. Observations without any PE investments in the four prior years, are always in the same non-PE group. The observations with PE investments, the PE observations, will be divided into four different groups, based on the relative amount of PEI, in order to measure the difference within these groups and the difference between these groups and the non-PE group. The first three groups are formed the same way as Bernstein et al. (2010) do. First, I combine all of the PE observations into one group (PE). Second, I split the PE group into two groups with an equal amount of observations, PElow and PEhigh, where low and high stand for the level of PEI. Third, I split the PE into four quartiles: PEq1, PEq2, PEq3 and PEq4, again with an equal amount of observations, with PEq4 being the top 25% of observations in terms of PEI. Fourth, I split the PE group into a bottom 90% (PElow90%) and a top 10% (PEhigh10%), to analyze the top 10% in terms of PEI of the PE group.
15
(2)
where is the natural logarithm of (1 + the simple growth rate of the observation from last year’s observation, is the coefficient that stands for the non-PE group, is the coefficient of PE, the total PE group, is the country-industry fixed effect,
is the country-year fixed effect, is the industry-year fixed effect and is the residual error term. If the coefficient is positive in the OLS regression results, then it means that the PE observations grow more (or decline less) than the non-PE observations. When I split the PE group up into multiple PE groups, take for instance the quartiles, then the regression will look like this:
(3)
which contains dummies PEq1, PEq2, PEq3 and PEq4, which are all equal to 1 if the observation falls into that group of PEI, 0 otherwise. Again, the signs of the coefficients ,
, and will show the difference between each group and the non-PE group.
To measure the difference in the growth rate of observations between the PE groups, I will test for the equality of the means using a Wald test (Beck and Levine (2004), Amess and Wright (2007), Bernstein et al. (2010)).
D. Reverse causality
16
IV. Data
A. Integrating database industries
The STAN industries are classified using the International Standard Industrial Classification (ISIC) code. The Zephyr database classifies industries in multiple ways with different codes including the UK SIC code, which resembles the ISIC code used by STAN. I match the Zephyr UK SIC codes with the STAN’s ISIC codes, to appoint all the target firms from Zephyr to their relevant industries. For target companies with multiple UK SIC codes, I hand-pick the most relevant industry. I manually assign companies without a known UK SIC code to their most relevant industries, with the help of the other classification codes and the industry descriptions (Bernstein et al. (2010)).
B. Industry data
I redistribute the target companies into 35 major industries, of which the list and respective ISIC codes can be found in appendix table A2. Bernstein et al. (2010) use 20 industries, but STAN data is available for more industries so the analysis can be performed on a more detailed level.
To be able to make reliable comparisons between observations, I need a group of countries that are comparable with each other in terms of financial development and size, as each observation bears the same weight in the analysis. Including countries that are significantly smaller in size would see their performances get attached too much weight in the benchmark. For this reason, I exclude from the European OECD countries included in STAN: Czech Republic, Estonia, Hungary, Iceland, Luxembourg, Slovak Republic and Slovenia, for there is a significant gap between the national value added of these countries and the next biggest country (Ireland), measured in 2000. I also exclude Austria, Greece, Poland and Portugal, because the amount of PE investment, measured by dividing total deal value by national value added, is significantly less than the other countries, measured in 2000. National value is used as a measure, rather than national production, as it is a clearer measure for true production9.
99
17
The remaining set of countries is Belgium, Denmark, Finland, France, Germany, Ireland, Italy, the Netherlands, Norway, Spain, Sweden and Switzerland. Compared to the FTSE developed Europe index group10, which is one of the few group of countries that are denoted as developed Europe, I have excluded Austria, Greece, Luxembourg and Portugal.
Some countries lack STAN data on a variable or industry: Spain (NOPS), Ireland (CPNK), Switzerland (LABR, NOPS, CPNK) and France (mining and quarrying industry). Since the growth rates of the observations are adjusted by the sample average of all available growth rates, it doesn’t matter much if sample sizes slightly differ in size between the variables (Bernstein et al. (2010)). Furthermore, some countries lack STAN data on years: France (2008 & 2009), Germany (2009) and Switzerland (2009). These observations are therefore excluded from the analysis.
C. Private equity deals
M&A deals are selected from Zephyr, if they are financed with PE, when the target company originates from one of the twelve countries, with a known deal value and are completed between 01-01-1997 and 31-12-2008. The overview of selection criteria and deals per step, is displayed in table V. I exclude financials since these firms operate in a regulated environment and their characteristics differ substantially from non-regulated firms (Berger and Udell (1998), Chen et al. (2010)). I exclude deals from the agricultural industry for the same reason as with the countries: as the industry is by far the smallest in terms of total production as well as amount of PE deals (8) and since the observations are all weighted equally in the analysis, I exclude this industry. The very small number of deals could also suggest that this industry is also mostly operating in a regulated environment. I exclude the France 2007 & 2008, Germany 2008 and Switzerland 2008 deals because of missing STAN data for the subsequent years. I exclude three mining and quarrying deals from France because of missing STAN data and I exclude seven portfolio deals which cannot be attributed to one, single industry. From the 3,471 deals, 106 of them lack a known UK SIC code for the target company. I manually assign these companies to their related industries, as e.g. Bernstein et al. (2010) do.
10
18 Table V
Selection criteria for PE deals
This table presents the amount of PE deals from the Zephyr database after each selection step.
Selection criteria Deals per step
Financed with PE 46,067
From the 12 countries 13,446
With a known deal value 6,298
Time period: 01/01/1997 up to 31/12/2008 4,112
Exclude financials 3,965
Exclude agricultural industry 3,957
Exclude France ’07-’08, Germany ’08 and Switzerland ‘08 3,481
Exclude 3 ‘Mining and quarrying’ deals from France 3,478
Exclude 7 portfolio deals 3,471
Figure I
Deal distribution and total deal value per year
This figure presents a breakdown of the amount of PE deals per year and its total deal value.
19
Figure 1 shows the distribution of deals and the total deal value per year. Clearly visible is the increasing amount of PE deals per year, as well as the increasing deal value per deal, as shown by the differences between the percentages. The drop in amount of deals and total deal value in 2008 and 2009 is partly caused by the exclusion of deals from France, Germany and Switzerland.
Appendix table A3 shows the distribution of deals and the total deal value per country. Again, some interesting observations can be made. First, the high amount of French deals compared to German deals. Second, German and Dutch deals are on average the highest in deal value. Third, Norwegian deals are on average the lowest in deal value.
Appendix table A4 shows the distribution of deals and the total deal value per industry. Several observations can be made: first, the highest number of deals are made in the industry ‘computer related activities’, whereas the highest total deal value is involved in the ‘post and telecommunications’ industry. Second, the average amount per deal is €157.4m, the lowest industry average amount per deal is €18.5m, found in the ’computer related activities’ industry and the highest average amount per deal of €379.1m is measured in the industry ‘post and telecommunications’.
D. Enterprise value/revenue multiple
The industry-year EV/revenue multiple is acquired from the Orbis database, as Zephyr only provides the pre-deal EV/revenue multiples of 88 out of 3,471 deals and Thomson Reuters Datastream11 could not be used as its industry classifications do not match with the UK SIC classifications which STAN uses. For industry-year multiples without Orbis data, I use the average of the other years of that industry across all countries. The average EV/revenue multiple across all industries and years is 1.16. The highest average industry multiple is 2.27 from ‘Pharmaceuticals’ whereas the lowest is 0.38 from ‘Wholesale excl. motor vehicles’.
20
E. Country-industry-year observations
Table VI shows the descriptive statistics of the simple growth returns for all of the STAN variables. The negative skewness and excessive kurtosis show that extreme returns occur more often than would be the case with a normal distribution.12
As shown by the means, PROD and VALU grow on average with 3.6% and 3.1% respectively. The means of the operating income variables are influenced by outliers, but the medians show that EBITDA (GOPS) grows more on average than EBIT (NOPS). Labor costs (LABR) grow with an annual average of 3.2% while the average number of employees (EMPE) stays roughly the same with 0.2% growth. Therefore, the average wages (LABR/EMPE) grow on average with 3.1%.
The average industry share of PE investments, measured as a share of industry production times the EV/revenue multiple, is 3.6% across all PE observations. Pulp and paper is the industry in this sample with the highest average industry PE share (17.5%), with construction averaging the lowest industry PE share of (0.3%). An average of 3.6% may seem small but
Table VI
Summary statistics for growth rates of observations
This table presents the summary statistics of the simple growth rates of the country-industry-year observations.
PROD VALU GOPS NOPS GOPS/CPNK NOPS/CPNK EMPE LABR/EMPE
Mean 0.036 0.031 0.054 -0.174 0.039 -0.042 0.002 0.031 Median 0.039 0.032 0.035 0.024 0.021 0.012 0.002 0.031 Maximum 0.655 2.033 30.000 99.000 12.538 99.691 0.322 0.352 Minimum -0.534 -1.018 -10.800 -267.984 -12.297 -264.311 -0.198 -0.138 Std. Dev. 0.085 0.105 0.775 6.449 0.601 7.095 0.046 0.033 Skewness -0.153 3.422 23.505 -29.980 6.531 -22.590 0.142 0.558 Kurtosis 8.304 79.056 46.063 1307.256 256.921 965.109 6.274 9.949 Observations 3,041 3,040 2,737 2,334 2,315 2,071 2,612 2,601 12
21
there are two factors that should be taken into account: first, the average holding period of PE is slightly larger than 4 years: a 2009 Roland Berger PE survey13 speaks of an average holding period of five years and Strömberg (2008) even speaks of an average of five to seven years. Second, as table V shows, more than half of the deals from Zephyr have an unknown value and are therefore discarded. This is caused by the tendency of most PE firms to provide as little deal information as is required by law and regulations (Gilligan and Wright 2010).
Appendix table A5 shows the correlations between the independent variables. Given the overlap between the three fixed effects of country-industry, country-year and industry, it is logical if some multicollinearity exists. As all fixed effects are considered important in the analysis, I use all three fixed effects (Bernstein et al. (2010)). Brooks (2008) states that the presence of multicollinearity does not affect the consistency, unbiasedness and efficiency of the OLS regression.
Appendix table A6 shows the number of observations and the average annual level of PE investment per analysis group. A couple of things stand out: First, the low annual levels of PE investment in groups PElow (0.25%) and of the two quartiles PElow consists of, PE Q1 (0.08%) and PE Q2 (0.42%). Second, the difference in PE investment levels between PE Q3 (1.49%) and PE Q4 (12.53%). PE high, which consists of both these quartiles has an average PE level of 7.01%. And third, the high average level of PE investment of the highest 10% of PE observations, PE high 10 (24.41%). The rest of the PE observations, pooled into PElow90% scores an average of (1.32%). Bear in mind that the actual levels may be considerably larger in reality, considering that more than half of the Zephyr deals were discarded because of missing information on deal sizes and are therefore not included in the analysis.
V. Results
I present the results per type of variables: productivity first, operating performance second and employment as third and last.
13
22
A. Productivity
Table VII shows the results on the OLS regression coefficients on the PROD variable, which stands for total output, the VALU variable, which stand for value added: total output minus intermediate inputs and the EMPE variable, which stands for number of employees. As argued before, the VALU variable is the superior variable in showing true productivity. However, OLS regression coefficients for both PROD and VALU are very similar: first, we see no difference between the PE 0 and the PE 1 group. Second, the PEhigh group outperforms the PElow group for both variables. The difference between PEhigh and PElow is highly significant for both PROD and VALU. Third, when the PE observations are split up into four quartiles, there is a significant outperformance visible for PE Q4: the variables grow annually nearly 1% more than the non-PE observations. And since there is no real difference between PE 0 and PE, it means that PROD and VALU grow for PE Q4 on average annually with 4.2% and 3.5%, instead of the sample mean of 3.2% and 2.6% respectively. This is significant on a 5% level. And fourth, the highest outperformance belongs to the PEhigh10% group with 1.43% PROD and 1.16% VALU. Both are significant. This shows that the higher the PE activity, the higher the annual growth rate for productivity and value added. Compared to Bernstein et al. (2010), there are some differences when we look at PROD and VALU: first, I find no outperformance of the total PE group versus the non-PE group. Second, I do find PEhigh to outperform PElow significantly. Third, these results show more differences between the quartiles and especially show PEq4 and PEhigh10% to significantly outperform on both variables.
B. Operating performance
23 Table VII
OLS regression coefficients on production variables
This table contains OLS regression coefficients. The endogenous variable is the ln growth rate of the variables production (PROD) and value added (VALU). The exogenous variables are the PE groups and the country-industry, country-year and industry-year fixed effects. The non-PE group is not omitted, so the coefficients show the differences between the PE groups and the non-PE group. Group equality shows significance levels from a Wald test of means of equality for the PE groups per column. *** Indicates statistical significance at the 1% level, ** at the 5% level and * at the 10% level.
Production Production Production Production Value added Value added Value added Value added
24 Table VIII
OLS regression coefficients on absolute operating performance variables
25 Table IX
OLS regression coefficients on relative operating performance variables
This table contains OLS regression coefficients. The endogenous variable is the ln growth rate of the variables gross operating surplus divided by net capital stock (GOPS/CPNK) and net operating surplus divided by net capital stock (NOPS/CPNK). The exogenous variables are the PE groups and the country-industry, country-year and industry-year fixed effects. The non-PE group is not omitted, so the coefficients show the differences between the PE groups and the non-PE group. Group equality shows significance levels from a Wald test of means of equality for the PE groups per column. *** Indicates statistical significance at the 1% level, ** at the 5% level and * at the 10% level.
26 Table X
OLS regression coefficients on employment variables
27
for NOPS, which is significant when you look at the Wald test probabilities. Third, PEq4 outperforms all three other quarters, and PEhigh10% scores the highest coefficients. This implies that the level of PE investments within an observation is linked to a growth increase of EBITDA (GOPS) and EBIT (NOPS). However, this relationship is weak, as shown by the significance levels. We see PEq2 score significantly bad with NOPS and the worst with GOPS as well, but this doesn’t provide us with much information as the (suppressed) PE ownership levels of the four quarters are 0.08%, 0.42%, 1.49% and 12.53% respectively, showing hardly any difference in PE quartiles 1, 2 and 3. The difference between PEq4 and the other quartiles and the difference between PEq3 and the two bottom quartiles PEq1 and PEq2 are possibly of more meaning to this analysis, as the differences in PE investment levels are more clear.
Table IX provides us with two measures of return on invested capital (ROIC) by dividing the industries’ EBITDA (GOPS) by its invested capital (CPNK) and by dividing the industries’ EBIT (NOPS) by its invested capital (CPNK). The average return of the entire sample is 0.9% for GOPS/CPNK and -0.7% for NOPS/CPNK. Not a single coefficient is significant here. The OLS regression coefficients are actually negative for the entire PE sample on both variables, implying that EBITDA on invested capital grows slower and that the EBIT on invested capital shrinks faster for the PE observations. However, the coefficients are not close to being significant, as are the differences between the various groups and quartiles. Furthermore, PEq4 and PEhigh10% both have positive coefficients for both variables, which could imply outperformance by the industries with the most PE activity, although these findings are far from significant.
C. Employment
Table X presents the OLS regression results on the growth rates on number of employees and on the average wage level, calculated by dividing labor costs by the total number of employees. The annual average growth rate for the entire sample of the number of employees is 0.1% whereas it is 3.0% for the wages.
28
positive coefficient for the PE group. The difference between PElow and PEhigh is however highly significant with coefficients of -0.69% and 0.16% respectively. Looking at the quartiles we see that the coefficient is negative for PEQ1-3 but significantly positive for PEQ4 with 0.0%. PEhigh10% scores the highest coefficient with 0.88%, which is significant on a 1% level. Bernstein et al. (2010) find the coefficients for number of employees to decrease when PEI became larger, I find results the other way around: the more PEI, the higher the coefficient.
The results about the growth in wages are very clear: the coefficients are all positive and range between 0.53% and 0.79%. They are all significantly higher than the non-PE group. However, there is no clear difference between all PE groups visible, suggesting that the difference in PE investment levels has no influence on the growth of the wages.
D. Reverse causality
Table XI shows that the level of PE investment, PEI, seems to be causing the change in subsequent growth rates and not vice versa. For three different PEI lag lengths, PEI granger causes nearly all variables. GOPS is the only variable that shows some evidence of two-way causation but the influence of PEI on GOPS is still more significant than the other way around.
Table XI
Pairwise Granger causality tests
This table shows the probability values of the null hypotheses that PEI does not Granger cause the given variable. The number behind PEI shows the lag length. The percentages in brackets are the probability values that the given variable Granger causes PEI. *** Indicates statistical significance at the 1% level, ** at the 5% level and * at the 10% level.
PROD VALU GOPS NOPS GOPS/CPNK NOPS/CPNK EMPE LABR/EMPE
29
E. Sensitivity tests
Next, I conduct a sensitivity test for the years 2008 and 2009 that are affected by the crisis. Appendix table A7 shows the summary statistics. The means show clearly that the sample observations have severely suffered from the crisis, with EBIT (NOPS) showing an average annual decline of 29,9%. The 399 observations are split up into the same PE groups as with the total sample. The results are displayed in the appendix tables A8 to A11. Two observations can be made: first, that all PE groups experience negative coefficients for productivity, absolute and relative operating performance variables. Second, all coefficients are insignificant. A subtle hint of procyclicality can be seen with all PE observations scoring worse than non-PE observations. However, the coefficients and the probabilities of the Wald-test are so insignificant that we really cannot draw any conclusions.
In unreported results, I run the same regressions without the two largest countries in terms of number of deals and total size of deals which are France and Germany, as do Bernstein et al. (2010) with the U.S. and U.K. All results closely mirror the results of the entire sample.
VI. Conclusions and recommendations
30 Table XII
Main results overview
This table shows an overview of the main results. PE vs. non-PE shows how industries in which PE funds invest in compare to industries without PE investments. High PE vs. low PE shows how industries with a level of PE investment higher than the median level score on the variable, versus industries with a level of PE investment lower than the median level. *** Indicates statistical significance at the 1% level, ** at the 5% level and * at the 10% level.
Type of variables Variables analyzed in this
paper PE vs. non-PE High PE vs. low PE
1) Productivity - Value added (VALU) +/- +***
- Production (gross output)
(PROD) +/- +***
2) Operating performance - Gross operating surplus &
mixed income (GOPS) + +**
- Net operating surplus &
mixed income (NOPS) + +**
- Gross operating surplus & mixed income / Net capital stock (GOPS) / (CPNK)
- +
- Net operating surplus & mixed income / Net capital stock (NOPS) / (CPNK)
- +
3) Employment - No. of employees (excl.
self- employed) (EMPE) - +***
- Labor costs (LABR) / No. of employees (excl. self-employed) (EMPE)
+*** +/-
that PE industries suffer more during the crisis than non-PE industries, but the results are insignificant and therefore inconclusive.
Granger causality tests show that the level of PE investments causes the performance of the industries and not vice versa. Therefore, I accept both hypotheses which state that PE investments have a positive effect on industry performance and that reverse causality does not exist. The findings of non reverse causality confirm results find on firm level (Baum and Silverman (2004)) and on industry level (Bernstein et al. (2010)).
31
variables. The results show that the outperformance for the absolute operating performance variables EBITDA and EBIT, is the highest of all variable types measured. Improvements in operating income confirm previous findings (e.g. Cressy et al. (2007), Kaplan and Strömberg (2009), Guo et al. (2011)). The result that industry absolute operating performance benefits most from PE investments is in line with the nature of private equity, where exit valuations are often based on a multiple of operating income, usually EBITDA (Damodaran (2006), Phalippou and Gottschalg (2009), Acharya et al. (2009)). Acharya et al. (2009) find that private equity fund returns are positively related to an increase in the operating income to sales ratio. Therefore, private equity funds are especially encouraged to improve operating income, in order to enhance the exit valuation. All industries in which PE invest, experience faster growing wages which is in line with findings by Wright et al. (2007) and Bernstein et al. (2010), although it is not related with the level of PE investment. The number of employees experiences decreasing growth rates for industries with low levels of PE investment but experiences increasing growth rates for industries with the highest levels of PE investment. These results confirm other contradictory results on the influence of PE investments on employment (Amess and Wright (2007), Amess et al. (2008), Davis et al. (2011)).
This is the second study, to the best of my knowledge, that thoroughly analyzes the effect of PE investments on an industry as a whole. My results show that there is indeed a significant, positive effect of PE investments on industry performance. Looking at the significant results, this research avenue deserves further study, especially as so little research has been done on this subject. It would be very interesting to figure out the kind of spillover effects that could be present and what part of the effect is direct and what part is indirect. It could also be interesting to analyze variables such as R&D expenditures, employment policies, asset sales and disposals and industry specialization.
32
33
Appendix
Table A1
Definitions of STAN variables
Industry variable Description
Production (gross output) (PROD) Value of goods and/or services produced in a year, whether sold or stocked, measured at current prices
Value added (VALU) Production minus the intermediate inputs that are used up into processes of production. Comprises LABR, consumption of fixed capital, taxes less subsidies on production and NOPS.
Gross operating surplus & mixed income (GOPS) Earnings before interest, tax, depreciation and amortization.
Net operating surplus & mixed income (NOPS) Earnings before interest and tax.
Net capital stock (CPNK) Value of all vintages of assets to owners where valuation reflects market prices for new and used assets
Labor costs (LABR) Wages and salaries of employees paid by producers as well as supplements such as contributions to social security, private pensions, health insurance, life insurance and similar schemes.
34 Table A2
Industries and ISIC codes
Industry Starting digits UK SIC code
Basic metals and fabricated metal products 27-28
Chemicals excl. pharmaceuticals 24 excl. 244
Community, social and personal services 90-93
Computer related activities 72
Consultancy and other business activities 74
Construction 45
Education, health and social work 80-85
Electical machinery and apparatus 30-31
Electricity, gas and water supply 40-41
Food products, beverages and tobacco 15-16
Hotels and restaurants 55
Machinery and equipment 29
Manufacturing and recycling 36-37
Medical, precision and optical equipment 33
Mining and quarrying 10-14
Motor vehicle transport equipment 34
Other non-metallic mineral products 26
Other transport equipment 35
Pharmaceuticals 244
Post and telecommunications 64
Printing and publishing 22
Pulp and paper 21
Radio, television and communication equipment 32
Real estate 70
Renting of machinery and goods 71
Research & development 73
Retail of motor vehicles and fuel 50
Retail trade excl. motor vehicles 52
Rubber, plastics and fuel products 23, 25
Textiles, textile products, leather and footwear 17-19
Transport and storage 60-63
Wholesale excl. motor vehicles 51
35 Table A3
Deal distribution and total deal value per country
Country Deals % of deals Total deal value
(in € billions)
% of total deal value
36 Table A4
Deal distribution and total deal value per industry
Industry Deals % of deals
Total deal value ( in € billions)
% of total deal value
Basic metals and fabricated metal products 133 3.8% 15.0 2.7%
Chemicals excl. pharmaceuticals 113 3.3% 31.4 5.7%
Community, social and personal services 136 3.9% 24.6 4.5%
Computer related activities 455 13.1% 8.4 1.5%
Consultancy and other business activities 188 5.4% 32.7 6.0%
Construction 51 1.5% 6.1 1.1%
Education, health and social work 73 2.1% 8.9 1.6%
Electical machinery and apparatus 84 2.4% 7.5 1.4%
Electricity, gas and water supply 44 1.3% 11.0 2.0%
Food products, beverages and tobacco 160 4.6% 22.3 4.1%
Hotels and restaurants 77 2.2% 13.3 2.4%
Machinery and equipment 233 6.7% 28.9 5.3%
Manufacturing and recycling 71 2.0% 5.4 1.0%
Medical, precision and optical equipment 124 3.6% 18.4 3.4%
Mining and quarrying 25 0.7% 5.5 1.0%
Motor vehicle transport equipment 57 1.6% 8.8 1.6%
Other non-metallic mineral products 66 1.9% 17.6 3.2%
Other transport equipment 49 1.4% 12.6 2.3%
Pharmaceuticals 46 1.3% 11.3 2.1%
Post and telecommunications 159 4.6% 60.3 11.0%
Printing and publishing 82 2.4% 22.8 4.2%
Pulp and paper 48 1.4% 10.0 1.8%
Radio, television and communication equipment 118 3.4% 16.0 2.9%
Real estate 81 2.3% 28.1 5.2%
Renting of machinery and goods 30 0.9% 8.1 1.5%
Research & development 135 3.9% 4.0 0.7%
Retail of motor vehicles and fuel 23 0.7% 5.8 1.1%
Retail trade excl. motor vehicles 170 4.9% 28.2 5.2%
Rubber, plastics and fuel products 67 1.9% 13.4 2.5%
Textiles, textile products, leather and footwear 83 2.4% 9.0 1.6%
Transport and storage 123 3.5% 26.1 4.8%
Wholesale excl. motor vehicles 144 4.1% 22.5 4.1%
Wood 23 0.7% 2.0 0.4%
37 Table A5
Matrix of correlations between fixed effects
This table presents the average correlations between the three fixed effects of all eight variables.
Country-industry Country-year Industry-year
Country-industry 1.00 0.14 0.24
Country-year 0.14 1.00 0.35
Industry-year 0.24 0.35 1.00
Table A6
Average level of PEI per analysis group
Observations Average PE share
Non-PE 985 0.0% PE 2,056 3.6% PElow 1,028 0.2% PEhigh 1,028 7.0% PEq1 514 0.1% PEq2 514 0.4% PEq3 514 1.5% PEq4 514 12.5% PElow90% 1,850 1.3% PEhigh10% 206 24.4% Table A7
Summary statistics for returns of observations 2008-2009
This table presents the summary statistics of the ln returns of the country-industry-year observations.
PROD VALU GOPS NOPS GOPS/CPNK NOPS/CPNK EMPE LABR/EMPE
38 Table A8
OLS regression coefficients on production variables 2008-2009
This table contains OLS regression coefficients of a sensitivity analysis of the years 2008 and 2009. The endogenous variable is the ln growth rate of the variables production (PROD) and value added (VALU). The exogenous variables are the PE groups and the country-industry, country-year and industry-year fixed effects. The non-PE group is not omitted, so the coefficients show the differences between the PE groups and the non-PE group. Group equality shows significance levels from a Wald test of means of equality for the PE groups per column. *** Indicates statistical significance at the 1% level, ** at the 5% level and * at the 10% level.
Production Production Production Production Value added Value added Value added Value added
39 Table A9
OLS regression coefficients on absolute operating performance variables
This table contains OLS regression coefficients of a sensitivity analysis of the years 2008 and 2009. The endogenous variable is the ln growth rate of the variables gross operating surplus (GOPS) and net operating surplus (NOPS). The exogenous variables are the PE groups and the industry, country-year and industry-country-year fixed effects. The non-PE group is not omitted, so the coefficients show the differences between the PE groups and the non-PE group. Group equality shows significance levels from a Wald test of means of equality for the PE groups per column. *** Indicates statistical significance at the 1% level, ** at the 5% level and * at the 10% level.
40 Table A10
OLS regression coefficients on relative operating performance variables
This table contains OLS regression coefficients of a sensitivity analysis of the years 2008 and 2009. The endogenous variable is the ln growth rate of the variables gross operating surplus divided by net capital stock (GOPS/CPNK) and net operating surplus divided by net capital stock (NOPS/CPNK). The exogenous variables are the PE groups and the country-industry, country-year and industry-year fixed effects. The non-PE group is not omitted, so the coefficients show the differences between the PE groups and the non-PE group. Group equality shows significance levels from a Wald test of means of equality for the PE groups per column. *** Indicates statistical significance at the 1% level, ** at the 5% level and * at the 10% level.
41 Table A11
OLS regression coefficients on employment variables
This table contains OLS regression coefficients of a sensitivity analysis of the years 2008 and 2009. The endogenous variable is the ln growth rate of the variables no. of employees (EMPE) and the average wage level (LABR/EMPE). The exogenous variables are the PE groups and the country-industry, country-year and industry-year fixed effects. The PE group is not omitted, so the coefficients show the differences between the PE groups and the non-PE group. Group equality shows significance levels from a Wald test of means of equality for the non-PE groups per column. *** Indicates statistical
significance at the 1% level, ** at the 5% level and * at the 10% level.
42
REFERENCES
Acharya, Viral V., Moritz Hahn, and Conor Kehoe, 2009, Corporate governance and value creation: evidence from private equity, working paper, NYU Stern School of Business, University of Munich, and McKinsey and Company
Amess, Kevin and Mike Wright, 2007, The wage and employment effects of leveraged buyouts in the UK,
International Journal of the Economics of Business 14:2, 179-195.
Amess, Kevin, Sourafel Girma, and Mike Wright, 2008, What are the wage and employment consequences of leveraged buyouts, private equity, and acquisitions in the UK?, Business School Research Paper, No. 2008– 01, NUBS.
Baum, Joel A.C. and Brian S. Silverman, 2004, Picking winners or building them? Alliance, intellectual and human capital as selection criteria in venture financing and performance of biotechnology startups, Journal
of Business Venturing 19, 411–436.
Beck, Thorsten and Ross Levine, 2004, Stock markets, banks and growth: panel evidence, Journal of Banking
and Finance 28, 423-442.
Bernstein, Shai, Josh Lerner, Morten Sorensen, and Per Strömberg, 2010, Private equity and industry performance, NBER Working Paper 15632, (Harvard Business School).
Bloom, Nicholas, Raffaella Sadun, and John Van Reenen, 2009, Do private equity owned firms have better management practices?, Centre for Economic Performance, occasional paper 24.
Brooks, Chris, 2008, Introductory Econometrics for Finance, 2nd edition, (Cambridge University Press,
Cambridge).
Brown, David T., C. Edward Fee, and Shawn E. Thomas, 2009, Financial leverage and bargaining power with suppliers: evidence from leveraged buyouts, Journal of Corporate Finance 15, 196–211.
Bruining, Hans, Paul Boselie, Mike Wright, and Nicolas Bacon, 2005, The impact of business ownership change on employee relations: buyouts in the UK and the Netherlands, International Journal of Human Resource
Management 16, 345–65.
Cumming, Douglas, Donald S. Siegel, and Mike Wright, 2007, Private equity, leveraged buy-outs and governance, Journal of Corporate Finance 13, 439–460.
Damodaran, Aswath, 2006, Valuation approaches and metrics: a survey of the theory and evidence.
Findations and Trends in Finance 1 (8), 693-784.
Davis, Steven J., John C. Haltiwanger, Josh Lerner, and Javier Miranda, 2011, Private equity and employment, Working paper, National Bureau of Economic Research
Desbrières, Philippe and Alain Schatt, 2002, The impacts of LBOs on the performance of acquired firms: the French case, Journal of Business Finance & Accounting 29, 695–729.
Gilligan, John, and Mike Wright, 2010, Private Equity Demystified, 2nd edition, ICAEW, London.
Guo, Shouron, Edith S. Hotchkiss, and Weihong Song, 2011, Do buyouts (still) create value, Journal of Finance, 66 (2), 479-517.
Harris, Richard, Donald Siegel, and Mike Wright, 2005, Assessing the impact of management buy-outs on economic efficiency: plant-level evidence from the United Kingdom, The Review of Economics and
Statistics, 87, 148-153.
Huyghebaert, Nancy and Mathieu Luypaert, 2010, Antecedents of growth through mergers and
acquisitions: empirical results from Belgium, Journal of Business Research, 63 (4), April, 392-403. Ivashina, Victoria, and Anna Kovner, 2011, The private equity advantage: leveraged buyout firms
and relationship banking, Review of Financial Studies 24, 2462-2498.
Jelic, Ranko and Mike Wright, 2011, Exits, performance, and late stage private equity : the case of UK management buy-outs, European Financial Management 17, 560–593.
Jensen, Michael C., 1989, The eclipse of the public corporation, Harvard Business Review 67, 61-74.
Kaplan, Steve, and Antoinette Schoar, 2005, Private equity performance: returns, persistence and capital flows.
Journal of Finance 60 (August), 1791-1823.
Kaplan, Steve, and Per Strömberg, 2009, Leveraged buyouts and private equity, Journal of Economic
Perspectives 23, 121-146.
Lichtenberg, Frank R. and Donald S. Siegel, 1990, The effect of leveraged buy-outs on productivity and related aspects of firm behavior, Journal of Financial Economics, 27, 165-194.
43 (December), Center for Economic Studies.
Meuleman, Miguel, Kevin Amess, Mike Wright and Louise Scholes, 2009, Agency, strategic entrepreneurship and the performance of private equity backed buyouts, Entrepeneurship Theory and Practice 33, 213–240. Phalippou, Ludovic and Oliver Gottschalg, 2009, The performance of private equity funds, The Review of
Financial Studies 22, 1747-1776.
Rasmussen, Poul., 2008, Taming the private equity fund “locusts”, Europe Today 8 (Spring), 130-133. Strömberg, Per., 2008, The new demography of private equity, World Economic Forum: The global economic
impact of private equity report 2008.
Wilson, Nick, Mike Wright, Donald S. Siegel, and Louise Scholes, 2011, Private equity portfolio company performance during the global recession, Journal of Corporate Finance 2011.
