Debt Structure, Private Equity Reputation, and Performance in Leveraged Buyouts

Transcription

1 Debt Structure, Private Equity Reputation, and Performance in Leveraged Buyouts Chen Liu May 5, 2014 Abstract This paper provides a comprehensive study of deal characteristics and participants involvement in leveraged buyouts (LBOs) and their impact on target firms performance. Using a sample of 501 U.S. LBOs completed between 1986 and 2011, I find that better post-buyout operating performance are associated with larger amount of leverage added during the LBO process, tighter LBO loan covenants, and equity contribution of target firms management. LBOs are more likely to exit through an IPO or a sale if they use more bank debt with tighter covenants and are sponsored by private equity firms of high reputation. These relations are robust to credit market conditions and aggregated LBO activities. The evidence suggests that the main source of value creation in LBOs is the reduced agency costs through the disciplining effect of debt, closer monitoring by lenders, and the better aligned management incentives. Private equity firms reputation is also important in ensuring successful deal outcomes. Keywords: leveraged buyout, debt structure, debt contracting, private equity JEL Classification: G34, G32, G21, G24 Queen s School of Business, Queen s University, cliu@business.queensu.ca. I am grateful to my supervisor Lynnette Purda for comments that substantially improved this paper. I appreciate comments and suggestions from Edwin Neave, Wei Wang, Usha Mittoo, Michael King, Fatma Saryal, Nichlas Cain, Richard Lord, and seminar and conference participants at Queen s University, Bank of Canada, the 2014 Midwest Finance Association Annual Meetings, the 2014 Eastern Finance Association Annual Meetings, and the 2014 Financial Management Association Applied Finance Conference. All remaining errors are my own.

2 1. Introduction In a leveraged buyout (LBO), a company is acquired using a relatively small portion of equity and a relatively large portion of outside debt financing. 1 Jensen (1989) argued that the LBO structure of highly leveraged capital structures, active corporate governance, concentrated ownership stakes, and well-aligned managerial incentives make the LBO form superior to widely held public corporations. Early empirical work supported the merits of this structure with papers by Kaplan (1989a) and Smith (1990) finding improvements in performance for firms undergoing an LBO in the 1980s. However, more recent studies by Guo, Hotchkiss, Song (2011) and Cohn, Mills, and Towery (2011) find few improvements to operating performance in LBOs completed in the 1990s and first half of the 2000s. Therefore, I am motivated to examine how recent LBO deals differ from the earlier ones and whether these differences are responsible for smaller performance enhancements in these later deals. To do this, I seek to identify the primary drivers of operational improvement with a particular focus on (1) leverage, (2) debt structure and terms, and the roles of different type of lenders, (3) the involvement of private equity (PE) investors, and (4) the participation of target firms incumbent management in the buyouts. I also examine whether performance is related to credit market conditions, aggregated LBO activities, LBO loan spreads, and the price paid in the LBO transactions. Figure 1 presents the structure of a typical LBO transaction. Equity investors in LBOs are mainly PE firms, incumbent management of target firms, or a combination of PE firm and management. Financiers of these transactions include banks, institutional investors, high-yield bond issuers, and mezzanine funds. 2 Traditionally, banks were heavily involved in financing LBO deals. However, the role of banks may have changed in LBOs, as banks are in general more involved in securitizing loans that they previously would have held (Bord and Santos, 1 A management buyout (MBO) is a form of LBO when incumbent management team takes over the firm. This paper includes MBOs in the sample and uses the general term LBO. 2 Bank lenders typically consist of commercial banks, savings and loan institutions, finance companies, and the investment banks serving as arrangers of the syndicated loans that finance the LBO transaction. Institutional investors are mutual funds, hedge funds, pension funds, insurance companies, structured vehicles such a collateralized debt obligation funds (CDOs), and other proprietary investors. High-yield bond issuers generally include high-yield mutual funds, hedge funds, pension funds, insurance companies, distressed debt funds, and CDOs. 1

3 2012). This motivates me to begin my examination of LBO performance drivers by investigating how the structure of these deals and the participants involved have changed over time. Another motivation for my study is the need for the PE industry to refocus on operational improvement. Based on the traditional LBO model presented by Figure 1, returns to PE s investment is realized through target firms operating improvement for a 3-5 year period before PE firms exit the deal through an IPO or a sale. However, anecdotal evidence from the PE industry shows that PE firms achieve high returns through timing the market to purchase the target and to exit and through financial engineering. Guo et al. (2011) find that besides operating performance, PE firms realized returns also come from changes in industry valuation multiples and the tax benefits from increased deb. Since the financial crisis, it has been argued that the PE industry needs to refocus on operational improvement. The Private Equity Investment (PEI) magazine (2012) states that There has been a realization, post-finance crisis, that private equity needs to return to its roots: creating value through operational improvement rather than financial engineering. Guo et al. (2011, p514) also argues that without consistent operating gains, it is unlikely that the returns we document can persist under less favorable credit and general market conditions. This need for PE firms to return to its roots motivates me to carefully examine operational improvement and its drivers. To undertake this examination, I construct a comprehensive dataset of 501 public-to-private U.S. LBO transactions completed between January 1, 1986 and December 31, 2011 from Capital IQ and SDC. I require all transactions to have financing details from LPC s Dealscan and pre- and post-buyout financial data from Compustat or Capital IQ, and missing data are filled from SEC filings. This dataset has the following merits. First, it is to my knowledge the most comprehensive U.S. LBO sample with large number of deals that have post-buyout data available. 3 As the target firms become private after the buyout, LBO studies on operating performance are restricted by data availability. By hand-collecting financial 3 Kaplan (1989a) s sample includes 76 management buyouts between 1980 and Guo et al. (2011) s study of LBOs during has 196 buyouts, 96 of which have post-buyout data available. 2

4 information of target firms that have publicly traded debt or subsequently file an IPO, I am able to construct a larger LBO that allows me to explore the heterogeneity among these LBO deals and generate results with better statistical properties in the cross-sectional analysis. Second, the sample period covers the cyclicality of LBOs starting from its first wave of the late 1980s and early 1990s, the slight recovery and decline in the 1990s, and the most recent boom and bust in the 2000s. This sample period makes this paper one of the first studies that directly examines how LBO deal characteristics, performance, and their relation have changed over time. 4 Moreover, this paper presents the most up-to-date sample that allows me to examine LBOs completed during and after the financial crisis, while most other studies include buyouts completed before the crisis. Using this data, I first measure post-buyout operating performance of target firms. Following Kaplan (1989a) and Guo et al. (2011), I calculate the percentage changes in EBITDA and net cash flows scaled by total assets or sales from the last fiscal year before the LBO to the first three years after the buyout completion, adjusted by industry medians. I find that performance change is largely positive for LBOs in the 1980s and 1990s but almost insignificant for the deals completed in the 2000s. For example, during the period of , the median industry-adjusted percentage increases in net cash flow to sales are significant at 32.7%, 28.2%, and 31.5% in the first three years after the buyout. Between 1994 and 2011, these increases are still significant, but by a lesser extent at 18.5%, 13.7%, and 29.8%. However, from 2002 to 2011, only the increase in the first year after the buyout is significant at 13.3%, while changes in the second and third years become insignificant. I next examine how LBO deal characteristics have changed over time that may be responsible for the documented decreasing performance. I find three important changes. First, LBOs in the 1990s and 2000s do not use as much leverage as the ones in the 1980s. For deals in the late 1980s, leverage, calculated as debt-to-asset ratio, increased by a median of 45% to a post-buyout leverage of 74% in the first full year after the buyout completion. However, the 4 Guo et al. (2011) studies deal characteristics and performance changes by comparing the deal pricing and financing details calculated in their paper with the results presented in Kaplan and Stein (1993). 3

5 median leverage increase was only 22% and post-buyout leverage was 57% for deals in the 2000s. Second, there is a structural change in the composition of the LBO debt. The proportion of bank debt in total LBO debt has decreased from a median of 85% in the late 1980s to a median of 34% in the 2000s. In the meanwhile, institutional investors have become more important in the LBO market with institutional loans financing a median of 63% of total LBO debt in the 2000s. In addition, covenants within the LBO loans have become less restrictive. Third, PE firms have become more important in LBO transactions. The proportion of deals sponsored by PE firms increased from 68% in the late 1980s to 96% in the second half of the 2000s. In addition, there are more club deals in recent years, leading to mega LBOs with large transaction values between 2005 and Having documented a decline in post-buyout operating performance and a shift in deal structure and participants, the second part of the paper seeks to identify what aspects of a deal s structure and the role of participants are associated with its performance. Possible drivers of performance that I consider are (1) changes in leverage, (2) monitoring by lenders, (3) involvement of PE firms, and (4) better aligned incentives through management participation. In examining these performance drivers, I control for pre-buyout characteristics of target firms, credit market conditions, aggregated LBO activities, LBO loan spread, and the price paid for the target firm. Regression results show higher industry-adjusted improvement in operating performance occurs when leverage is increased by a larger amount through the buyout process, LBO loan covenants are more restrictive, and when incumbent management of target firms contributes equity and participates in the buyout. PE reputation, however, is not significantly related to changes in operational improvement. In addition, I do not find evidence that links performance to credit market conditions, LBO loan spreads, or the buyout price. Overall, these results suggest that the main source of value creation is the reduced agency costs in the post-buyout firms through the discipline effect of debt, closer monitoring by lenders, and the better aligned management incentives. These results help us to understand the observed reduced performance 4

6 enhancement in the more recent LBOs as they use less leverage and less restrictive loan covenant, which are important drivers for performance improvement. Another way to examine LBO success is to look at the outcome of each deal whether it goes bankrupt or exits through an IPO or a sale to financial or strategic buyer. Using IPOs or sales as an indicator of LBO success, I find that LBOs are more likely to succeed if they use more bank debt and tighter covenants, experience no CEO change, and are sponsored by highly reputable PE firms. LBOs are more likely to fail if the buyers are subsidiary of banks that are also financiers of the deals. These results are consistent with the lenders monitoring and PE firms reputation as sources of value creation in LBOs. I also find that LBOs completed during the time when interest rates are lower than their historical average are less likely to succeed, providing some evidence for the market timing behavior of LBO buyers that they overinvest in unprofitable deals under favorable credit market conditions. Contributions of this paper are as follows. First, this paper contributes to the literature on value creation of LBOs by examining the primary drivers of performance improvement and successful deal outcomes. To the best of my knowledge, this is the first paper that studies the effects of detailed LBO financing structure and its contractual features, and PE reputation on post-buyout operation performance. Results of this paper will further our understanding of when and how an LBO may be successfully employed to improve firm performance. By doing so, it facilitates our understanding as to why recent LBOs seem to be less successful than previous transactions. In addition, this is one of the first large sample LBO studies with a sample period that covers the entire cyclicality of the LBO history. Second, this paper contributes to the literature on debt structure and debt contracting in the setting of LBOs. This paper finds that loan covenants are important drivers of operating performance and instrumental to ensure successful outcomes. This result has important implications for practitioners as well as policy makers in that they should focus on covenants to reduce risks and to improve performance of target firms. The proportion of bank debt is also important for LBOs to exit through an IPO or a sale, suggesting that composition of LBO debt 5

7 needs to be carefully structured. Third, this paper contributes to the literature on private equity by being one of the first studies that investigate how PE reputation affects performance. The finding that, controlling for target and deal characteristics, PE reputation is not related to operating performance in the first three years after the buyout but is important in ensuring successful deal outcomes provides some indirect evidence that PE firms create value through later stage of LBOs. Findings of this paper will motivate future studies in investigating when and how PE firms create value in LBOs. The rest of the paper proceeds as follows. Section 2 reviews related literature and develops hypotheses. Section 3 describes the sample and provides evidence on post-buyout operating performance. Section 4 presents the changing characteristics and participants of LBO deals over time. Section 5 examines the drivers of post-buyout performance. Section 6 conducts robustness analyses. Section 7 concludes. 2. Literature Review and Hypotheses 2.1 Measuring Value Creation in LBOs Previous studies have examined value creation in LBOs in two ways: returns to LBO investors and post-buyout performance improvement in LBO target firms. In the first approach, value creation is measured as the returns to invested debt and equity capital from the time of buyout to a subsequent IPO, sale of the firm, or bankruptcy. Studies on LBO deal level returns suggest significant value creation through LBOs as evidenced by positive returns to investors. For example, Kaplan (1989a) estimates a median market-adjusted return of 28% (mean 42%) for investors in 25 MBOs in the 1980s that went public after an average of 2.7 years. Also, Guo et al. (2011) finds a median market- and risk-adjusted return to pre-buyout capital of 68.7% (mean 94.7%) for a sample of 70 LBOs completed from 1990 to On the LBO fund level, literature provides mixed evidence. Kaplan and Schoar (2005) investigate returns for 160 LBO funds between 1980 and 2001 and find that the median fund 6

8 underperformed stock market index, generating only 80% of the return on the S&P 500. Higson and Stucke (2012) find that the buyout funds in their sample have significantly outperformed the S&P 500, with funds liquidated in the period generating excess returns of on average 4.5% per year. The different results of these two studies are mainly due to sample selections as both studies find large heterogeneity in returns across funds. In addition, both studies show that PE performance is persistent and suggest that different LBO sponsors may have different skills in managing their portfolio companies. The heterogeneity in fund returns and performance persistence motivate me to examine characteristics of PE firms to determine whether and how they are associated with the performance of target firms. The second way to examine value creation in LBOs is to focus on the post-buyout operating performance of target firms. 5 Kaplan (1989a) studies 48 MBOs from 1980 to 1986 and finds that industry-adjusted ratios of EBITDA to sales increased by 21.3% and cash flow to sales increased by 28.3% during a three-year period following the buyout. Smith (1990) reports a significant increase in operating cash flow per employee and per dollar of operating assets from the year prior to the buyout to one year post-buyout for 58 MBOs between 1977 and Lichtenberg and Siegel (1990) study 193 LBOs between 1981 and 1986 with a total of 1,132 plants and show that plant total factor productivity increases more than the industry average in the years following a buyout. In contrast to the significant performance enhancement documented in the early studies, the evidence of performance improvement is weaker for more recent LBOs. Guo et al. (2011) find a median of only 2.25% industry-adjusted increase in operating margins and a 12.54% decrease in cash flow margins for 94 LBOs completed between 1990 and Cohn et al. (2011) also find little evidence of performance enhancement using corporate tax return data for 317 LBOs from 1995 to This paper follows the second approach and examines post-buyout operating performance for the following reasons. First, this paper defines value creation as the real effects of buyouts, 5 Some studies, for example Guo et al. (2011), use changes in operating perofrmance as an explanation for returns to investors. 7

9 such as increased efficiency and reduced costs, rather than purely financial returns. 6 For this purpose, operating performance is a cleaner measure of value creation compared with the returns to investors. This is because returns to investors do not necessarily reflect the real value creation as they are usually calculated upon the exit of the deal, therefore depending on market conditions and investors market timing ability. Second, one of the goals of this paper is to construct a comprehensive and up-to-date LBO database that includes deals completed during and after the credit crunch. Most of these deals have not reached their outcome yet so no returns to investors are available for these deals. However, I can still examine the value creation and its drivers through operating performance change. One problem with measuring performance using cash flow variables from target firms financial statements, as mentioned in Cumming, Siegel, and Wright (2007), is that they are in general subject to managerial manipulation. However, as all the performance ratios in the paper are industry-adjusted and assuming all firms in the same industry are subject to managerial manipulation in the similar ways, I expect the effect from manipulation to be small although the incentives for LBOs to show improved performance are probably greater than the average firms. 2.2 How LBOs Create Value: Hypotheses In this subsection, I develop hypotheses on the sources of post-buyout performance changes of LBO target firms. I hypothesize that operational improvement in target firms is driven by the disciplining effect of increased leverage, closer monitoring by lenders, active involvement of PE firms, and better aligned management incentives The Disciplining and Monitoring Effect of Debt The first key ingredient in a buyout transaction is leverage. Jensen (1986, p325) states that many of the benefits in going-private and leveraged buyout transactions seem to be due to the control function of debt. Leverage creates pressure on managers not to waste money, as they 6 According to KKR founder Henry Kravis, private equity firms create value in LBOs over the long-term as managers, not merely as financial engineers. Kravis said that We only make money because we improve the operations of the newly acquired company. Source: Merger Talk - LBO firms rush to exits with quick flips. Reuters News, December 30,

10 must make interest and principal payments. This pressure reduces the free cash flow problem described in Jensen (1986) where entrenched managers dissipate the free cash flows and overinvest in negative-npv projects. Also, the increased risk of financial distress associated with higher leverage motivates managers to operate the firm efficiently and to increase profit. Therefore, I expect that target firms that have increased their leverage by a greater amount through the LBO process perform better. Hypothesis 1.1(Debt Disciplining Hypothesis): Firms with higher level of leverage increase have better post-buyout performance. In addition to the disciplining effect of debt, I examine whether lenders monitoring associated with LBO debt leads to additional disciplining effect and therefore generating better performance of target firms. At the center of this examination is the conflict of interest between shareholders and bondholders that has negative impact on the value of the firm s outstanding debt as well as the total value of the firm (Bradley and Roberts, 2004). Lenders monitoring on managers behavior can help to mitigate these conflicts and reduce the attendant agency costs. To study the monitoring effect, I first examine the proportion of bank debt in total LBO loans. This is because banks are generally thought to have more incentives and comparative advantages in monitoring borrowers (Diamond 1984, 1993; Park 2000). Therefore, I hypothesize that LBOs funded with a larger proportion of bank debt would perform better as these deals are more closely monitored by banks. Another way to examine the monitoring effect is through the investigation of LBO loan covenants. Chava and Roberts (2008) suggest that covenants increase firm value in two ways. First, covenants that monitor and control managers behavior mitigate the reduction in firm value from the conflicts of interest between shareholders and debtholders and managers acting on behalf of shareholders to expropriate bondholders wealth. For example, covenants can restrict borrowers use of cash flows and require them to repay with proceeds of excess cash flow, asset sale, or debt and equity issuance. 7 These requirements mitigate the free cash flow 7 Based on Miller (2012), in a typical syndicated loan contract, 100% of net proceeds from asset sales and debt issuance and 50% to 75% of excess cash flow are required to prepay the loans. 9

11 problems described in Jensen (1986). Second, covenants define the circumstances under which creditors are permitted to intervene in management. This threat of transfer of control rights from borrowers to creditors serves as additional discipline mechanism for managers. Therefore, I expect that LBOs with more restrictive covenants tend to perform better. Besides covenants, the maturity structures of LBO loans are also important. When LBOs are financed with short-term loans, the incentive effects of debt described by Jensen (1986) tend to be stronger. In particular, a shorter maturity increases required debt service payments, thus increasing the incentives for mangers to work harder to generate cash and avoid wasting resources in the earlier stages of the LBOs. In combination, these LBO loan characteristics form Hypothesis 1.2. Hypothesis 1.2 (Lenders Monitoring Hypothesis): LBOs with more bank loans, tighter loan covenants, and shorter loan maturity tend to generate better post-buyout operating performance Private Equity Involvement Another possible source of value creation in LBOs may be the involvement of PE firms. As equity investors in LBOs, PE firms are incentivized to actively engage in the target firms management. Also, general partners (GPs) of PE funds are paid a management fee of 2% on the fund s capital and receive a carried interest of 20% of the profits above a certain benchmark realized by the fund. Therefore, GPs have incentives to closely monitor their portfolio firms. As described by KKR s founder Henry Kravis, PE firms generally aren t board members who show up once a month... Most of us in the industry live with these companies on a day-to-day basis. 8 However, it is hard to directly observe PE firms involvement in management, as target firms become private after the buyouts and are therefore not required to disclose corporate 8 Source: Merger Talk - LBO firms rush to exits with quick flips. Reuters News, December 30,

12 governance information. 9 As a result, I use PE firms reputation as a proxy for their experience and skills to manage the target firms, where reputation of each PE firm is measured by its years of experience or its market share based past deals. I hypothesize that LBOs sponsored by highly reputable PE firms perform better. Hypothesis 2.1 (Private Equity Reputation Hypothesis): LBO deals sponsored by PE firms of high reputation experience higher performance improvement. A recent trend in LBOs is club deals, where two or more PE firms pool their assets to acquire target firms and manage them collectively. Club deals can be beneficial as each PE firm may bring different expertise to target firms. For example, when KKR teamed up with Bain Capital and Vornado Realty Trust to acquire Toys "R" Us, the New York Times stated that it was clear what each firm brought to the table. Kohlberg Kravis has a good reputation in the retail business, Bain has a good record doing turnarounds, and Vornado clearly knows real estate. 10 However, as the number of PE firms in the club gets larger, it may become harder to make timely operational and management decisions. For example, Jeffrey Walker, a managing partner of CCMP Capital, argued that it was difficult to manage an LBO that has more than two or three investors. 11 Also, based on the experience of the venture capital industry, which is closely similar to the private equity industry, the ideal size of the consortium is two PE firms. To sum up, anecdotal evidence suggests that club deals may improve performance through each PE firm offering valuable management advice. However, as the size of the club gets larger, this benefit may decrease. Hypothesis 2.2 (Club Deal Hypothesis): Club Deals perform better than LBOs sponsored by a single PE firm. However, this advantage tends to decrease as the number of PE firms 9 Some studies look at the board composition of target firms using the Dash dataset that s only available for U.K. firms. For example, Cornelli and Karakas (2011) examine the board structure for 88 U.K. LBOs from 1998 to 2003 and find significant changes in board size and composition when a firm goes private. Board size generally decreases and the presence of outside directors is drastically reduced, as they are replaced by individuals employed by the LBO sponsors. 10 Source: Do Too Many Cooks Spoil the Takeover Deal, the New York Times, April 3, Source: Buyout Veterans Have Questions about Club Deals, Dow Jones Newswires, January 24,

13 participating in a deal gets larger. Recent studies in LBOs have investigated the relationship between banks and PE firms and how it affects returns to PE investors at the exit of LBOs. Fang, Ivashina, and Lerner (2012) find that LBOs sponsored by PE firms that are subsidiaries of banks (the bank-affiliated deals) exhibit worse equity returns if the deal is completed during the peaks of the credit market. Ivashina and Kovner (2012) find that bank relationship formed through repeated interactions between banks and PE firms lead to more favorable loan terms and higher equity return to the PE firms. Empirical results on the effect of banking relation on returns to PE firms are mixed, depending on the nature of the relation and the motivation behind it. I follow Fang et al. (2012) and examine bank-affiliated deals, as there may be some distinct features of these deals. First, bank-affiliated PEs may have better access to funds provided by their parent banks should an LBO opportunity rise and these PE firms are better able to take advantage of the favorable credit market conditions. Second, bank-affiliated deals provide parent banks with cross-selling opportunities (such as potential M&A advisory work, cash-management services, etc.) that increase their fee income. As a result, buyout decisions may not be based on PE s expectation on efficiency improvement of target firms, but to take advantage of these cross-selling opportunities. Hypothesis 2.3 (Bank Affiliation Hypothesis): Bank-affiliated LBOs perform worse than stand-alone deals Management Participation When incumbent managers of target firms contribute equity and participate in a buyout, they become the equity investors and their incentives are well-aligned with other shareholders. As a result, agency costs are minimized. Hypothesis 3 (Management Participation Hypothesis): Management-participated LBOs tend to have better performance due to better-aligned management incentives that reduce agency costs. 12

14 Some studies look at management turnover and consider it as a way to measure PE firms control over target firms. Gong and Wu (2011) find that 51% of incumbent CEOs are replaced within two years of the LBO announcement. Acharya and Kohoe (2008) find that for U.K. LBOs, one third of the CEOs are replaced within the first 100 days and two-thirds are replaced over a four-year period. However, management turnover can be a noisy measure. First, it may not be clear that the management change is due to PE firms unless it is explicitly indicated in the proxy statement. Second, even it is confirmed that PE firms replace the CEO or CFO, management change may not necessarily indicate increased control from the PE firms. On the one hand, management turnover can be consistent with replacing bad managers by the good ones. On the other hand, the same managers running the company before and after the buyout may indicate low pre-buyout agency problem and therefore there is no need to replace managers. As a result, I expect the effect of management turnover on performance to be ambiguous. 3. Sample, Data, and Post-Buyout Performance Measures This section describes the construction of the LBO sample and calculates measures of the post-buyout operating performance of the LBO target firms and examines how they have changed over time The Buyout Sample The LBO sample of this paper is constructed from the Standard and Poor s Capital IQ and the Securities Data Company s (SDC) U.S. Mergers and Acquisitions Database. Compared with SDC, Capital IQ has the advantage as it allows me to link the LBO transaction details to target firms financials and information on LBO buyers (private equity firms, management teams, or another corporation). However, one problem with Capital IQ is its limited coverage of earlier deals. Stromberg (2008) compares the 1980s LBOs from Capital IQ and other data sources and estimates that Capital IQ covers between 70% and 85% of the LBO sample for this period. As one of the goals of this paper is to compare deal characteristics of LBOs during the period of 13

15 1986 to 2011, I also collect LBOs from SDC which has better coverage of earlier. I manually combine LBO transactions from the two sources, track for name changes, and eliminate duplicate observations. Each LBO transaction in my sample meets the following criteria: (1) the transaction is flagged as an LBO, MBO, or secondary LBO and completed between January 1, and December 31, 2011; (2) the target is a publicly traded U.S. company; and (3) the transaction value is $10 million or larger. The minimum deal value of $10 million is lower than that in some other studies, such as Kaplan (1989a) and Guo et al. (2011). It is chosen to avoid biasing against earlier time periods when small deals were more common. This initial screening yields a total of 1,586 LBO transactions. To reconstruct the financial structure of each deal accurately, I require that all transactions have financing details available from Reuter s LPC Dealscan loan database. 13 I match the Dealscan data with the buyout sample by borrower name, deal active date, and primary loan purpose. 14 I then reconstruct the financial structure for each deal using the tranche level data of LBO loans from Dealscan and the mezzanine debt from Capital IQ. This reduces the sample to 885 observations. In addition, I require that all target firms have preand post- buyout financial information from COMPUSTAT or Capital IQ, and missing data are filled from SEC filings. 15 This drops 384 transactions, the majority of which are the buyouts in the 1980s, as Capital IQ mainly provides financial statement information for the 1990s and the 2000s. My final sample consists of 501 LBO transactions. Table 1 Panel A compares the mean and median inflation-adjusted transaction values of 12 My sample starts from 1986 because the loan information from Dealscan starts from Restructuring LBO deals across databases requires matching by names of target firms. Target firms mostly appear under their old names in SDC, Dealscan, and Compustat, while Capital IQ uses only the most recent names. I keep track of all name changes using a text search in Company Tearsheet of CapitalIQ. I also use the Wall Street Journal for name changes if the Tearsheet is ambiguous. 14 If the primary loan purpose is LBO, borrower name matches the LBO target firm s name, and the deal active date is around LBO announcement or effective date, I confirm Dealscan and LBO sample as matched. If the primary loan purpose is takeover, I match name and date and go through 10K and 8K to confirm the loan is to fund the buyout of the target firm. 15 Capital IQ provides financials for private companies that belong to one of the following categories: (1). Private companies with publicly traded debt, (2). M.A. targets filing financials in 8-K/A SEC forms; (3). D&B Financials; (4). U.S. bank subsidiaries filing with various regulatory bodies in the U.S., such as FFIEC, CUA, OTS. My sample contains samples of cases (1) and (2). 14

16 the 501 LBOs with the 1,586 public-to-private LBOs originally collected from Capital IQ and SDC. Overall, 31% (501 out of 1,586) of the LBOs use syndicated loans documented by Dealscan and have post-buyout financials available. The median size of the 501 LBOs ($ million) is significantly larger than the overall sample ($ million). This is due to the fact that firms with public debt financing and/or subsequently filing for an IPO, therefore reporting post-buyout financials, are typically larger. I am well aware that the requirement for LBOs to have deal financing information from Dealscan and post-buyout financials available induces a sample selection and survivorship bias. As a consequence, my results may not be representative for the entire LBO population. However, as this is the only way to get this level of details in my data, choosing between representativeness and getting insight from details, I opt for the latter. I will keep in mind this sample bias when interpreting my findings. Figure 2 presents the total number of LBO deals (left y-axis, solid line) and the inflation-adjusted total transaction values (right y-axis, bar) by LBO effective year. 16 The first LBO boom occurred in the late 1980s, with a total transaction value increasing from $24 billion in 1987 to a peak of $81 billion in The largest deal during the time was KKR s buyout of RJR Nabisco in 1989 with a transaction value of $39 billion. The first wave of LBOs ended with the recession in when the high yield bond market collapsed. The buyout market started to resume in 1996 but crashed with the bursting of the tech bubble in In the mid-2000s, LBOs reappeared in a third buyout boom. Total transaction value increased sharply from $5.4 billion in 2002 to $65 billion in 2005 and reached a historical high of $273 billion in The years 2006 and 2007 observed the surge of the mega-buyouts, including the acquisitions of TXU ($42 billion, later renamed as Energy Future Holdings Corp ), HCA Holdings, Inc ($33 billion), Kinder Morgan ($28 billion), and First Data ($27 billion). Table 1 Panel B breaks down the sample by industry grouping based on SIC codes. Target firms are from eight broad industries but are concentrated in manufacturing with approximately 44.5% of the sample coming from this industry. Firms in the service industry and the wholesale 16 LBO effective year is the year an LBO is complete. This is shown as the deal closed date or effective date in Capital IQ and SDC. 15

17 and retail industry are the next biggest groupings. In the late 1980s and early 1990s, almost 50% of the buyout transactions were from the manufacturing industry. Since the year 1997, relatively more firms come from the service and the wholesale/retail industries. Overall, the sample shows an increased industry scope for LBOs over time. I next examine how buyout prices have varied over time. Following Kaplan and Stein (1993), I measure the buyout price, referred to as total capital, as the sum of the market value paid for the target firm s equity, the value of the firm s outstanding debt, and the fees paid in the transaction, minus any cash removed from the firm to finance the buyout. Column (1) of Table 2 presents the sample medians of total capital by LBO effective year. For the full sample of 501 transactions, the median buyout price is $ million. There is a trend towards larger deals in later years, significant at 1% level based on the nonparametric trend test. The median capital for LBOs in the 1990s is $382 million. It increased to $1,956 million in 2007 due to the mega-buyouts and reached $2,223 million in 2011 after the LBO markets recovered from the crisis Evidence on Post-Buyout Operating Cash Flows In this subsection, I calculate measures of post-buyout operating performance and examine how they have changed over time Methodology To document the post-buyout operating performance, I use the operating income as measured by EBITDA and net cash flow (NCF). EBITDA measures the cash generated from buyout firms operations before depreciation, interest, or taxes, and therefore is not affected by the level of interest payments this serves the purpose of my analysis that examines the overall income to the post-buyout firm before incomes being divided to shareholders and debtholders. In addition, since EBITDA is before tax, it excludes the tax-shield effects of LBOs as in Jensen (1989b), allowing me to focus on the operational effect. I use NCF as it is the primary component of the numerator in a net present value analysis to value a buyout company. A 16

18 permanent increase in NCF, therefore, should lead to an increase in firm value. EBITDA and NCF are scaled by sales and average assets for each fiscal year. Performance change is calculated as the percentage changes of these cash flow measures in the first three full years after the year of LBO completion (year +1, +2, and +3) compared to the last fiscal year before the buyout completion (year -1). To control for pre-buyout firm characteristics, I also look at the percentage changes of the cash flow measures from two years before LBO (year -2) to the last fiscal year (year -1). In order to evaluate the economic and statistical significance of pre- to post-buyout changes in performance, I follow Kaplan (1989a) and Guo et al. (2011) and calculate the industry-adjusted performance measures. The industry-adjusted change equals the percentage change in the cash flow variables for the target firms minus the median percentage change over the relevant period for all Compustat firms in the same industry. Firms in the same industry as the target firms are those that have the same four-digit SIC code. Comparisons are made at the three-digit level and then at the two-digit level if fewer than five industry matches are found Evidence on Changes in Operating Performance Table 3 summarizes the medians of unadjusted and market-adjusted percentage changes in operating performance for the last full year prior to completion of the buyout year (year -2 to year -1) and from year -1 to up to 3 years after the buyout. Panel A shows the median changes over the entire sample period and Panel B presents the medians in each LBO wave and the time trend. The table also shows the number of observations for the overall sample and in each wave, as I exclude firms when they exit through IPOs, sales, or bankruptcies my results are not affected by firms that have exited. Panel A of Table 3 shows that the industry-adjusted percentage increases in EBITDA to sales are significant at 7.0%, 6.9%, and 8.9% in years +1, +2, and +3 relative to year -1. Changes in NCF to sales adjusted by industry medians are also positive and significant, with medians of 18.5%, 13.9%, and 19.6% in years +1, +2, and +3. In contrast to the increased cash 17

19 flow variables that are scaled by sales, there are 12.2% and 9.8% significant decreases in EBITDA over average assets in year +1 and +2 relative to year -1. The median changes become insignificant in year +3. The industry-adjusted changes in the net cash flow as a proportion of average assets decreased by 5.7% in the first year after the buyout (significant at 10%), the median changes then become insignificant in year +2 and +3. The significant decrease in the first year after buyout can be explained by Kaplan (1989b) that buyout accounting leads to a change (usually an increase) in the book value of the assets, representing the difference between the market value of equity and the book value. This may also lead to an underestimates of operating improvement when cash flow measures are scaled by total assets. I next divide the sample period into three sub-periods, , , and , based on the cyclicality of the LBO market presented by Figure 2. Panel B shows the median changes of performance in each sub-period and the time trend. The nonparametric trend test results show that there is less performance enhancement in the more recent deals for all four measures. For example, during the period of , the industry-adjusted percentage increases in NCF to sales are significant at 32.7%, 28.2%, and 31.5% in the first three years after the buyout. The increases between 1994 and 2011 are still significant, but by a lesser extent at 18.5%, 13.7%, and 29.8%. From 2002 to 2011, only the increase in the first year after buyout is significant at 13.3% and the changes in years +2 and +3 are insignificant. To conclude, results in this section show some evidence of post-buyout performance improvement. However, this is mainly driven by LBOs in the early years, as there is a significant trend of decreased post-buyout performance improvement in the more recent deals. 4. LBO Deal Characteristics and Participants Involvement Having documented a decreasing trend of improvements to operating performance, this section studies factors that are expected to be important drivers for performance based on the debt disciplining, lenders monitoring, private equity reputation, and management participation hypotheses developed in section 2.2. Specifically, I focus on the pre- and post-buyout leverage 18

20 and its change, the composition of LBO debt and its contractual features, private equity firms reputation and their bank affiliation, and the trend in club deals Leverage, Debt Structure, and Contractual Features The debt disciplining hypothesis and the lenders monitoring hypothesis state that firms that have larger amount of leverage added during the LBOs and are more closely monitored by lenders will experience more improvement to operating performance. Having documented a decreasing trend of performance improvement, I now examine whether and how leverage, bank debt proportion, loan covenants and maturity structure have changed over time Leverage Panel A of Table 4 reports the capital structure changes from LBOs. Columns (1)-(3) show the sample medians of the pre- and post-buyout leverage and leverage changes, calculated using the financial data from Compustat, Capital IQ, and SEC Filings. Prior to the buyout, firms have a median leverage ratio of 31.92%. Leverage increased significantly to a sample median of 64.35% after buyout, with a median percentage increase in leverage of 30.55%. Column (4) shows the median leverage change of all Compustat firms. Comparing with the Compustat population, I find that leverage increase is unique to LBO firms. Over time, both the leverage change and the post-buyout leverage ratio have been decreasing. The decreasing trend of leverage change, in combination with the declined post-buyout performance, provides some preliminary evidence for the debt disciplining hypothesis LBO Debt Structure I next study the composition of LBO debt and its contractual features, using tranche (or facility ) level data from Dealscan and Capital IQ. LBO debt is syndicated through different tranche. According to Miller (2012), revolving credit facilities and Term A loans are usually 19

21 packaged together and syndicated to banks. 17 The term B, C, and D loans are structured specifically for institutional investors. Therefore, I consider the revolving credit facility and the term A loan tranche in an LBO debt package as bank debt and term B, C, and D loans as institutional debt. I also consider notes that are sold to institutional investors as institutional debt. The bridge loan tranche from Dealscan and the mezzanine debt from Capital IQ that are subordinated to bank and institutional debt are considered the junior debt. 18 Panel B of Table 4 presents the structure of LBO debt over the years. For each category of debt, I calculate the ratio of the amount of debt to total LBO debt and the percentage of LBO deals that use this type of debt. Columns (1) and (2) demonstrate the use of revolving credit facilities and term A loans in financing LBO debt. The ratio of revolving credit facilities to total LBO debt has decreased over time, from 61% in 1986 to 13.7% in The ratio of term A loans to total LBO debt has also declined, from its peak at 76.9% in 1993 to 1.2% in Although most LBOs continue to use revolving credit facilities, the proportion of LBOs that uses term A loans has significantly declined from 47.1% in 1993 to 5.7% in As a result, the ratio of bank debt to total LBO debt and the percentage of LBOs that use bank debt have decreased significantly (Column (3)), suggesting the declining importance of banks in financing LBOs. With the decrease of bank debt, there has been an increasing use of institutional debt, as presented in Column (4). The proportion of institutional debt has increased from 1.2% in 1989 to 58.4% in 2011 and the nonparametric trend test shows a significant increase in the proportion at 1% level. Since 1998, the proportion of institutional debt in total LBO debt has exceeded that of bank debt, suggesting that institutional investors have become more important in the LBO market. Column (5) of Panel B demonstrates the use of junior debt. The use of junior debt is 17 A revolving credit facility allows borrowers to draw down, repay, and re-borrow capital over time. Term A loan (also referred to as amortization term loan or TLa) is an installment loan that requires a complete withdrawal of funds at inception and substantial principal repayment throughout the life of the loan. 18 A bridge loan facility is interim, committed financing provided to the borrower to bridge to the issuance of permanent capital and it usually takes the form of an unsecured term loan. Mezzanine debt is a form of financing that is part debt and part equity and is senior only to common equity. It is usually unsecured and bears a higher interest rate than secured debt and often gives the lender a stake in the equity of the borrowing firm.investors of the mezzanine debt are typically insurance companies and the mezzanine funds. 20

22 mainly prevalent in the late 1980s and the 2000s. The large proportion of junior debt in the total LBO debt in the late 1980s corresponded to the use of high-yield junk bond whereas since 2002 the use of mezzanine debt in LBO financing has increased dramatically. In summary, this subsection shows that the importance of banks as financiers of LBO transactions has been decreasing and that institution debt has become more important in financing LBOs. These trends may provide some explanations for the decreasing performance of post-buyout firms as bank monitoring has been considered as instrumental in reducing agency costs of debt and therefore creating value in LBOs (Diamond 1984, 1993; Park 2000). This relation will be tested in the multivariate analysis in the next section LBO Loan Spread, Maturity, and Covenants I next examine the spread and maturity of bank debt verses institutional debt and the LBO debt covenant at package level. Results are presented in Panel C of Table 4. Columns (1)-(3) show the medians of all-in-drawn interest spread of bank debt, institutional debt, and their differences. All-in-drawn spreads (over 6-month London Interbank Offered Rate (LIBOR)) for each tranche are from Dealscan and include both the interest costs and fees associated with borrowing. There is a significant increasing trend of bank debt spread, as shown by Column (1). Institutional debt spreads are higher than bank debt spreads but the institutional-bank difference reported in Column (3) narrows over time with the increased usage of institutional debt in LBO financing. This is consistent with Miller s (2012) argument that the spread difference between institutional loans and bank loans narrows when the institutional demand for syndicated loans is high. Columns (4)-(6) show the median maturity (in months). There is no significant change in the maturity of bank debt. However, the maturity of institutional debt has decreased throughout the sample period, from 120 months in 1992 to 79 months in I also examine the covenants of LBO debt as they provide specific requirements and restrictions on management behavior that reduce the agency costs associated with the conflict of interests between shareholders and debt holders. Covenant information is obtained from 21

23 Dealscan. To measure the tightness of covenant restrictions, I use the covenant intensity index developed by Bradley and Roberts (2004). This index indicates whether each loan package contains the following six specific covenants asset sale sweep, debt issuance sweep, equity issuance sweep, financial covenant, dividend covenant, and secured debt covenant and counts the number these covenants in each package. The first three covenants, also referred to as the sweeps, specify the percentage of net proceeds from an asset sale, debt issuance, or equity issuance that the borrowers must use to pay down any outstanding debt. Financial covenant enforces minimum financial performance measures that the borrowers must maintain. Bradley and Roberts (2004) define a binary variable that is equal to 1 if the loan contract contains more than two restrictions on financial ratios and zero otherwise. Dividend covenant restricts the ability of the borrowers to distribute cash to shareholders and secured debt covenant requires the debt to be secured. The covenant intensity index counts the number of covenants included in each loan package and the index ranges from 0 through 6. Covenants are unique to packages, so that every tranche in a package is covered by all of the covenants. If an LBO uses multiple loan packages, I use the index of the most covenant-heavy package as the covenant index of the deal. Column (7) of Table 4 Panel C presents the median Bradley and Roberts (2004) s covenant intensity index for loans with non-missing covenant information. The number in the bracket shows the proportion of LBOs with non-missing loan covenant information of total LBOs each year. As information on covenants is fairly limited prior to 1994, loans syndicated before 1994 only have secured debt covenant reported. During the period of , LBO loans have a median of 5 covenants based on the measure of Bradley and Roberts (2004) and there is a significant decrease in covenant tightness. In 2002, the median number of covenants was 6; however, the number dropped to 1 for LBO loans syndicated in the late 2000s. The Bradley and Roberts (2004) s index only considers the presence of financial covenants if there are more than two financial covenants. It does not take into account the number or different types of financial covenants used in a debt contract. However, there is large 22

24 variation in the use of financial covenants in my sample and the number of financial covenants included in the LBO loan packages ranges from 0 through 6. Therefore, I modify the index by including the number of financial covenants used in each package. Specifically, the modified covenant intensity index is the sum of (1) number of financial covenants (up to 6), (2) number of sweeps (asset sales sweep, debt issue sweep, equity issue sweep, excess cash flow sweep, insurance proceeds sweep 19 ), (3) dividend covenant (0/1 variable), and (4) secured debt covenant (0/1 variable). 20 The modified index is presented by Column (8). LBO loans constructed between 1998 and 2002 used more covenants, with a median of 8-10 covenants. The number of covenants decreased in the second half of the 2000s with only 1 or 2 covenants for LBO loans in 2010 and Column (9) shows the proportion of LBO loans with no financial maintenance covenants (the covenant-lite loans). 21 There is a general trend towards more covenant-lite loans over time. In summary, results from Columns (7)-(9) show declining tightness of covenant restrictions of LBO loans, suggesting weaker monitoring by lenders that may lead to worse post-buyout performance Private Equity in LBOs Private Equity Involvement The private equity reputation hypothesis on LBO operational improvement states that LBOs sponsored by PE firms of high reputation tend to perform better as these PEs have better 19 The excess cash flow sweep and the insurance proceeds sweep specify the percentage amount of net proceeds a borrower receive from excess cash flows or insurance settlements that must be used to pay down any outstanding loan balance. Including the two sweeps ensure that all sweep covenants are considered in the modified intensity index. 20 Covenant intensity measures used in this paper indicate the presence of certain covenants in the loan contract, not the actual threshold of each covenant. This is because the thresholds for financial covenants and sweeps are related to many factors, such as the credit market conditions and the borrowers specific characteristics. Therefore, it is hard to compare the threshold directly. 21 According to Bavaria and Lai (2007), S&P define covenant-lite loans as those with no maintenance financial covenants that have to be maintained quarterly through the term of the loan. Instead, covenant-lite loans have only incurrence covenants that do not have to be met on an ongoing basis as maintenance covenants do. Incurrence covenants only restrict the borrower s ability to issue new debt, make acquisitions, or take other action that would breach the covenants. 23

25 skills to improve performance and to reduce risks of target firms (Kaplan and Schoar, 2005; Axelson, Stromberg, and Weisbach, 2009). Therefore, I examine the changing characteristics of PE firms that are expected to be instrumental to LBO success. To identify PE firms, I download all private equity funds (PE funds) from Capital IQ. For each LBO transaction that is collected from Capital IQ, I merge the buyer Excel Company ID of the LBO to the PE funds Excel Company ID. For the buyout sample from SDC, I run a text search for the names of PE firms in the transaction synopses and hand match with the PE funds from Capital IQ. As this paper is to look at private equity involvement at firm level, I consolidate PE funds to PE firm level. So if one PE firm has multiple active PE funds, I use the Excel Company ID at the PE firm level for the analysis. For example, both Lehman Brothers Mezzanine Fund and Lehman Brothers Capital Partners IV are identified at the PE firm level as the Lehman Brothers, Private Equity Division. I also track the name changes of the PE firms. Of the 501 LBO deals in the sample, 448 deals have at least one PE firm involved. The remaining transactions are either management buyouts or buyouts by another corporation with no PE firm involved. There are in total 234 PE firms sponsoring the 448 deals. Appendix A2 presents the top 25 PE firms by the number of LBOs and total transaction values of these buyouts they sponsored. The most frequent PE firms are Kohlberg Kravis Roberts & Co (27 deals), TPG Capital (26 deals), The Blackstone Group, (22 deals), Goldman Sachs Capital Partners (21 deals), and Bain Capital Private Equity (20 deals). Table 5 Panel A presents the PE firms involvement in LBOs over the sample period. Column (1) shows the number of buyouts that have PE firms involved and its proportion in all LBOs in the year (in the brackets). There is an overall trend of increased PE involvement. In the late 1980s, the average proportion of LBOs with PE firms involved is around 73%. In the 1990s, almost all LBOs have PE firms. Following Officer, Ozbas, and Sensoy (2010), I categorize a PE firm as a prominent PE if it is listed as the 50 largest PE firms by the Private Equity 24

26 International (PEI) magazine from the year 2007 to Column (2) shows that 52% of the LBOs in my sample have prominent PE firms and there is a trend of more prominent PEs involved in LBOs over time. Column (3) shows the number of bank-affiliated LBOs, defined as buyouts that are sponsored by PE firms that are subsidiaries of banks at the LBO announcement day and that the parent banks provide loans for the deal. 23 There are in total 103 bank-affiliated deals (21% of 501 LBOs) in the sample and the total transaction value of these deals is 35% of the sum of transaction value of all 501 LBOs. This is similar to the findings of Fang et al. (2012) that bank-affiliated groups account for nearly 30% of the overall PE market, and the findings reported by Lopez-de-Silanes, Phalippou, and Gottschalg (2011) that roughly one-third of the investment in the global private equity dataset are done by PE groups that are subsidiaries of banking and finance companies. Column (3) also shows that Bank-affiliated deals exhibits cyclicality corresponding to the LBO market activities. In the years 1989, 1998, and 2007, when the LBO activities were at the peak of the cycle, the proportions of bank-affiliated deals in total LBOs was higher. This is consistent with Fang et al. (2012) that PE firms time the market and bank affiliations allow these PE firms to take advantage of favorable credit market conditions. Columns (4)-(6) present the time trend for club deals. Specifically, Column (4) shows the number of club deals and its percentage in total LBOs and Column (5) presents the total transaction value of club deals and its ratio over the total transaction values of all LBOs in my sample. 30% of the LBOs are club deals and the total transaction values of these deals are 50% of all deals. This is consistent with PE firms pooling their assets to acquire large targets. 22 Starting 2007, the PEI magazine ranks PE firms based on the capital raised over the previous five-year period. I also add PE firms that are listed as the top 25 PEs in my sample from Appendix A2 to this list of prominent PE firms if they are not already included in the prominent list. The PE firms added are mainly those that are subsidiaries of banks (the bank-affiliated PEs), as these firms may not be on the PEI list because they may use internal capital rather than relying on external fundraising. Following Officer et al. (2010) I also add Forstmann Little and HM Capital Partners (formerly Hicks, Muse, Tate, and Furst) because they are historically prominent PE firms that have been less active in recent fundraising. 23 Some PE firms started as subsidiaries of major banks but later became independent. I only consider an LBO as bank-affiliated if it is announced during the time the PE firm is subsidiary of a major bank. 25

27 Column (6) shows the maximum number of PE firms in a club deal consortium by year. Club deals were very rare in the late 1980s and early 1990s and they start to become important in the late 1990s. During , club deals reached their peak almost 50% of LBOs were club deals and the total transaction value of these deals was around 80% of all LBOs during the time. The largest-ever LBO, the buyout of TXU in 2007 with a transaction value of $44.5 billion was conducted by a consortium including KKR, the TPG Capital, and Goldman Sachs Capital Partners. As shown by Column (6), there is a general trend of more PE firms participating in a club deal. During the whole sample period, there are on average 2.6 PE firms in the consortium of the club deals (not tabulated). The deal that has the largest number of PE firms involved is the buyout of SunGard Data Systems sponsored by seven PE firms and completed in 2005 with a transaction value of $11.5 billion. 24 In summary, analyses of this subsection show that PE firms have become more involved in LBO deals, as evidenced by the increasing proportion of LBOs sponsored by PE firms in the more recent years and the increasing importance of club deals. Bank-affiliated LBOs have shown some cyclicality that corresponds to the credit market condition, suggesting some market timing of these deals Private Equity Reputation Having documented the time trend of PE firms involvement in LBOs, I construct measures for private equity reputation. There are two reasons that explain why PE firms reputation based on past experience may be related to the performance of their current deals. First, according to Kaplan and Schoar (2005), performance of PE funds persists over time. The observed performance persistence can be attributed to PE firms experience and skills in selecting, restructuring, and monitoring target firms. Better-performing PE firms gain experience through their experiential learning from previous deals and PEs with lower returns cannot get funds from investors and fail to exist. Second, Axelson et al. (2009) find that highly 24 The seven PE firms were Silver Lake, Bain Capital, the Blackstone Group, Goldman Sachs Capital Partners, KKR, Providence Equity Partners, and the TPG Capital. 26

28 reputable PE firms are less susceptible to risk shifting as they have incentives to pursue relatively more conservative investment strategies in order to maintain their reputation. Previous empirical studies measure PE fund reputation in a number of different ways that include fund size, its market share, the number of recent LBO transactions, and the number of previous fund raisings. 25 As PEs in my sample are at the firm level, I use reputation measures that can be constructed for each PE firm. Following Demiroglu and James (2010), I record the number of LBOs sponsored by a PE firm or the total transaction values of these deals in the past 36 months before an LBO announcement or since Reputation for PE firm j at month t is measured by PE j s market share: Reputation j,t = Number of LBOs sponsored by PE j in prior 36 months or since 1970 total number of LBOs in prior 36 months or since 1970 or Reputation j,t = Total TV of LBOs sponsored by PE j in prior36 months or since 1970 total TVof LBOs in prior 36 months or since 1970 In order to get the LBO transaction history, I use all Capital IQ recorded LBO, MBO, SBO transactions for U.S. target firms since 1970 plus the buyout sample from SDC a total of 19,014 deals. In the case of club deals, I consider the buyout as a full deal for each PE firm. When calculating LBO deal-level PE firm reputation, I use the reputation score of the PE firm with the highest reputation if the deal has multiple PE firms. The reputation score is set to zero if there is no PE firm involved in a buyout deal. Another reputation measure is years of experience, which is calculated as the number of years based on the first ever LBO deals sponsored by the PE firm and the last LBO deals in the sample. Overall, the median PE firm in the sample has 24 years (mean: 21 years) of experience and has invested in 15 LBOs (mean: 22 LBOs) since 1970 (not tabulated). Panel B of Table 5 presents the relation between PE firms reputation, their different forms of participation in LBOs, including club deals and bank-affiliated deals, and LBO deal characteristics and financing structure. Following Demiroglu and James (2010), I run univariate 25 See Demiroglu and James (2010) for discussions on strengths and weaknesses of each reputation measures. 26 The earliest LBO deal documented by Capital IQ is in

29 OLS regressions using various measures of LBO deal and financing characteristics as dependent variable and one reputation or participation variable as explanatory variable with year dummies. Consistent with Demiroglu and James (2010), I find that highly reputable PE firms tend to sponsor larger LBOs with lower per-buyout leverage ratio. LBOs with more reputable PE firms tend to have larger leverage increase, use less bank debt and more institutional debt with lower spread and longer maturity. However, there is no significant relation between market share based reputation measure and covenant intensity. 5. Explanations for Post-Buyout Operating Performance In this section, I examine the relationship between operating performance and LBO deal characteristics that are expected to be related to performance improvement as the following: Industry-adjusted performance change = f(leverage change, debt characteristics, PE characteristics, Management participation, pre-buyout target characteristics) My goal is to test different hypotheses developed in Section 2.2 to determine factors that contribute to operational improvement in LBOs. The analysis will also help to understand whether the documented changing characteristics of LBOs can be used to explain the reduced performance improvement observed in the more recent LBO deals. Table 6 reports the multivariate regression results for post-buyout operating performance and its drivers. The dependent variables are the industry-adjusted percentage changes in EBITDA/sales or NCF/sales from the last full pre-buyout year (year -1) to the second full year after deal completion (year +2). This allows me to include LBO deals completed by the end of 2010 to look at the performance of LBOs during and after the financial crisis. Also, I use EBITDA scaled by sales, instead of total assets, to avoid the buyout accounting problems related to total assets as described in Kaplan (1989b). To control for pre-buyout characteristics of target firms, all regressions include leverage ratios and asset tangibility at the end of year -1, industry-adjusted changes in EBITDA/sales or NCF/sales from year -2 to year -1, and target 28

30 firms cash flow volatiles in the last three years before buyout completion (years -3, -2, and -1). All regressions include year dummies. First, I examine the debt disciplining and lender s monitoring hypotheses in Panel A of Table 6. Columns (1) and (2) show that the effect of leverage change on performance is significantly positive. That is, firms with greater amount of leverage added during the LBOs show more post-buyout performance improvement, supporting the debt disciplining hypothesis. I next examine whether and how performance is related to the monitoring by lenders. I include the proportion of LBO debt that is funded by banks, with the expectations that banks have more incentives and advantages to monitor the borrowing firms and that the percentage of bank debt is proportional to banks monitoring effort. I also include the modified covenant intensity index that measures the presence of different covenants in LBO loans. The maturity structures, calculated as the weighted average maturity (in months) of bank debt and institutional debt for each LBO loan package, are also considered in the regression, with the expectation that shorter maturities that require early principal payment and/or refinancing indicate closer monitoring by lenders therefore leading to better performance. Columns (3) and (4) demonstrate a significantly positive coefficient for covenant intensity, suggesting that controlling for the leverage effect, tighter covenants further improves the post-buyout performance. However, bank debt proportion and maturity are insignificant drivers for performance. I then replace the maturity at package level with bank debt maturity, the coefficient is still insignificant (not tabulated). I next examine the effects of PE firms reputation and their different forms of participation in Panel B. I use different PE reputation measures constructed in the last section as independent variables in the regression. However, none of these reputation measures is significantly related to performance (Columns (1) and (2)) 27. Columns (3) and (4) show result for a regression that includes the club deal dummy as it has been argued that in a club deal, different PE firms bring different expertise to the target firm s management, therefore providing another source of value 27 The reported coefficients and p-values are for the reputation measure calculated as the market share of PE firms based on the total transaction value of the LBOs on the past 36 months. 29

31 creation. 28 However, the regression results do not support this argument. Another view on club deal is that as the size of the consortium of PE firms gets bigger, it is harder to make timely operational and management decisions. Experts of the private equity industry have suggested that the optimal size of the consortium is two or three PE firms. 29 I construct an optimal consortium size dummy variable that takes the value of 1 if there are two or three PE firms in a club, and 0 other wise. However, regression results presented by Columns (5) and (6) show no evidence that the optimal consortium size is related to performance. In columns (7) and (8), I include a dummy variable that indicates whether an LBO is bank-affiliated deal. I find that bank affiliation has no significant effect on performance. In sum, regression results presented in Panel B of Table 6 provide no evidence that PE firms reputation or different form of involvement in LBOs is related to post-buyout operating performance of target firms. In panel C of Table 6, I test the management participation hypothesis that LBOs tend to perform better when managers of target firms contribute equity and participate in the buyout as their incentives are better aligned with other shareholders. I use a dummy variable that indicates management participation. Specifically, the dummy variable equals 1 if Capital IQ labels a transaction as management buyout, management participated, individual investor participated when the individual investor is confirmed to be board member or management of the target firm, or the firm is bought out through an employee stock ownership plan (ESOP). If a transaction is from the SDC database, the dummy is equal to 1 if the SDC synopsis describes the deal as management led or management participated. Results in column (1) support the management participation hypothesis when using industry-adjusted changes in EBITDA/sales as dependent variable. However, the effect of management participation on changes in NCF/sales is insignificant. 28 For example, New York Times commented on KKR, Bain, and Vornado Realty Trust s buyout of Toys "R" Us that it was clear what each firm brought to the table. Kohlberg Kravis has a good reputation in the retail business, Bain has a good record doing turnarounds and Vornado clearly knows real estate. Source: Do Too Many Cooks Spoil the Takeover Deal, the New York Times, April 3, For example, Jeffrey Walker, a managing partner of CCMP Capital, said that I find it very difficult managing a deal that has more than two or three investors. Source: Buyout Veterans Have Questions about Club Deals, Dow Jones Newswires, January 24,

32 I next examine whether management turnover is related to performance. I go through Factiva and manually collect news on CEO and CFO change from the time of buyout announcement until the deal reaches a final outcome (bankruptcy, IPO, or a sale to another buyer). I supplement the Factiva results with the Key Development on Corporate Structure Related news from Capital IQ. From the announcement date to the final deal outcome day, 212 firms (42%) experienced a change in the CEO and 167 firms (33%) experienced CFO change. In the regression, I use a CEO change dummy that takes the value of 1 if there is CEO change from the buyout announcement to two full years after the buyout completion. 30 This is because the independent variable is the change in EBITDA/sale in year +2 compared with year -1. Columns (3) and (4) show insignificant coefficients for CEO change. When all measures of proposed performance drivers are included in regression analyses shown by Columns (5) and (6) in Panel C, leverage change, covenant intensity, and management participation are significant. Test for variance inflation factor does not show any multicollinearity problem. In general, the base-line regression results of Panels A-C of Table 6 support the debt discipline, lenders monitoring, and management participation hypotheses of value creation in LBOs, but do not support the hypotheses of private equity reputation, club deals, or bank affiliations. 6. Robustness 6.1. Subsample Analyses In this subsection, I examine whether the base-line regression results hold in each LBO wave and when I divide the sample by management and PE firms participation in LBOs. I only use the industry-adjusted changes in EBITDA/sales from year -1 to year +2 as dependent variables, as regressions with NCF/sales show similar results. Columns (1)-(3) of Table 7, Panel A present the regression results for each LBO wave. Leverage change and covenant intensity are significant and positive across all columns, 30 I also use a dummy for CFO changes, or a dummy for both CEO and CFO changes. The results are the same. 31

33 suggesting that debt disciplining and lender s monitoring are important performance drivers in all LBO waves. Management participation is significantly related to performance in the first and second wave, however, it becomes insignificant for LBOs in the third wave. PE firm s reputation is significantly and negatively related to performance in the third LBO wave at 10% significance level. One possible explanation is that highly reputable PE firms may focus more on taking advantage of the favorable credit market conditions in the early- and mid-2000s to generate high return to their own investment and less on improving the operating performance of target firms. Column (4) of Panel A reports the regression results that exclude LBOs completed in 2006 and 2007, in order to disentangle the impact of the financial crisis on firm performance. The previous results on leverage change, covenant intensity, and management participation still hold. Table 7 Panel B divides the sample by management and PE firms involvement in LBOs. Column (1) shows the regression results for all management-participated deals whereas LBOs in Column (2) have no management involved. Leverage change and lender s monitoring are still important performance drivers, whether the monitoring is through a larger proportion of bank debt or tighter covenants. When management participates in the deal, changing CEO or management team has negative impact on performance (Column (1)). Column (3) of Panel B shows LBOs sponsored by PE firms and Column (4) with no PE (i.e. management-only buyout). For the subsample of PE-involved LBOs, regression results support the debt disciplining and lender s monitoring hypotheses. Moreover, the coefficient for management-participation dummy is significantly positive, indicating that when PE firms sponsor LBOs, it is important for incumbent management to participate as well. In column (4), only leverage change is significant. This can be explained by the fact that in management-only buyout, managers are the only owners of the firm and their incentives are perfectly aligned. Therefore, no outside lenders are needed to monitor managers besides leverage itself. 32

34 6.2. Credit Market Conditions Studies have shown that LBO buyers, whether they are PE firms, managers of target firms, or other corporations, take advantage of favorable credit market conditions. Kaplan and Stein (1993) present evidence that the 1980s LBO boom was driven by the attractive terms of high-yield bonds. Shivdasani and Wang (2011) find that the LBO boom of 2004 to 2007 was fueled by growth in securitization. Axelson, Jenkinson, Stromberg, and Weisbach (2013) study 1,157 PE deals worldwide from 1980 to 2008 and show that the economy-wide cost of borrowing is the main driver of both the quantity and the composition of debt in LBOs. These findings suggest that more LBOs will be undertaken when the credit market is more favorable and leverage is cheaper to acquire and that LBO buyers may overinvest in unprofitable deals during the time. As a result, LBOs completed during the favorable credit market conditions may perform worse than other deals. In this section, I test whether the key results from my hypotheses hold when controlling for the impact of credit market conditions on performance. Following Barry, Mann, Mihov and Rodriguez (2008), I add to the baseline regression the Baa yield and the difference between the Baa yield in the month of LBO completion and its 60-month historical average. 31 I also include the term structure, calculated as the difference between 10-year T-Bond yield and three-month T-Bill yield. Column (1) of Table 8 presents the regression results. Leverage change, covenant intensity, and management participation are still significantly and positively related to post-buyout performance after controlling for market timing. In addition, there is no evidence of worse performance for deals announced at the time of favorable credit markets. Another way to examine market conditions is to look at the hot versus cold LBO market. Following Colla, Ippolito, and Wagner (2012), I construct a hot market dummy by taking a 12-month centered moving average of the number of LBOs for each month over the sample period. Hot months are defined as above the median in the distribution of the monthly moving average across all months. The hot market dummy takes a value of 1 if a deal is completed in a 31 Regressions with the month of LBO announcement or Dealscan s deal active month generate similar results. 33

35 hot month, and zero otherwise. Column (2) shows that controlling for the LBO market condition, leverage, covenants, and management participation are still important drivers for performance, and that deals announced during hot LBO market do not generate worse performance than other deals. I also examine whether performance is related to the LBO loan spread. The spread at the LBO deal level is calculated as the weighted average of all-in-drawn spread across all tranches. While the Baa yield, its difference from the historical average, term structure, and the hot market dummy are related to the general credit market and LBO market condition, the loan spread measures the actual cost of debt for each LBO deal. If LBO buyers overinvest in unprofitable deals when leverage is cheaper to acquire, I expect to find less performance improvement when the LBO loan spread is lower. Column (3) shows that controlling for loan spread, the effects of leverage, lenders monitoring, and management participation in the LBO transaction are still significant. In the meanwhile, LBO loan spread is not significantly related to performance. I also look at deal price, calculated as the ratio of EBITDA over the total transaction value, adjusted by subtracting the S&P 500 market earnings/price ratios for each month of LBO completion. Column (4) shows that leverage change, covenant intensity, and management participation are still significant and that deal prices do not affect performance. To summarize, robustness analyses of this section show that leverage, covenants, and management participation are still important drivers for post-buyout performance enhancement after controlling for credit market conditions, LBO market conditions, loan spread, and buyout prices. These results again support the debt discipline, lenders monitoring, and management participation hypotheses while rejecting the private equity reputation hypothesis. In addition, I find no evidence that LBOs constructed during favorable market conditions perform worse than other deals, nor is performance related to prices paid for LBOs. 34

36 6.3. LBO Deal Outcome In this section, I conduct additional tests on the effects of LBO deal characteristics on performance, where performance is now measured by the ultimate outcome of these deals. I search Factiva for SEC filings and news to identify deal outcomes that include (1) bankruptcy or distressed exchange, (2) a subsequent IPO, (3) a sale to a strategic buyer (4) a sale to a financial buyer (also known as the secondary LBO), (5) still privately held by the same buyer, or (6) unknown. I supplement Factiva information with the Capital IQ Tearsheet and the company history from its website. Table 9 shows ultimate deal outcomes as of June 30, 2013 by LBO effective year. Over the entire sample period, 83 deals (16.6%) file for bankruptcy, initiate a financial restructuring, or go through distressed exchange. 35.3% of the LBOs exit through an IPO, 17% through a sale to a strategic buyer, and 11.8% through a sale to financial buyer. The majority of the deals in the late 2000s are still privately held by the same buyers. For all deals that have reached outcome, the median months to exit is 42 months (mean 47 months, not tabulated). Exit through an IPO or a sale to a financial or strategic buyer is generally considered as a successful outcome of an LBO, as it is upon an IPO or a sale PE firms returns on LBOs are realized. In addition, Holthausen and Larcker (1996) find that the performance of LBO firms exceeds that of its industry rivals at the time of the IPO, suggesting that the IPO is related to LBO success. Harford and Kolasinski (2011) find that when a sponsor sells a firm to a public strategic buyer, the buyer s stock price reaction is positive. In order to test whether the key results from my hypotheses still hold when using deal outcomes to measure LBO success, I run logit regressions. Specifically, for the dependent variable, I create a success dummy that takes on a value of 1 if an LBO exits through an IPO or a sale to financial or strategic buyer and zero otherwise. 32 As most of the recent deals have not reached outcomes yet, I only consider LBOs that are completed by the end of 2008, taking into account that the last day of information collection on deal outcome is June 30, 2013 and that the median months to bankruptcy is Alternatively, I assign the value of 1 if the deal outcome is an IPO or a sale to financial buyer and zero otherwise, the results are similar. 35

37 months (Table 9, Column (1)). Column (1) of Table 10 presents results of the baseline regression using the success dummy as the dependent variable. Consistent with the lenders monitoring hypothesis, LBOs are more likely to reach successful outcomes if they are financed with higher proportion of bank debt and have tighter loan covenants. CEO changes during the time target firms are privately held by PE firms have negative impact on the deal outcome. Leverage change and the dummy for management participation are not significant. Column (1) also shows that LBOs sponsored by PEs with higher reputation score have better outcomes, supporting the private equity reputation hypothesis. I use different reputation measures that include PE s market share based on the number of deals or the total deal values in the prior 36 months or since 1970, the natural log of the number of deals in the prior 36 month or since 1970, and natural log of PE s years of experience. All measures generate significant and positive estimates on PE reputation. This result provides a clear picture of the roles of PEs in an LBO reputation is not directly related to better operating performance as measured by EBITDA and net cash flows to total assets or sales, but is important in ensuring successful outcomes of LBOs. Regression results also show that bank-affiliated LBOs are less likely to exit through an IPO or a sale, consistent with the bank affiliation hypothesis that deals sponsored by PE firms that are subsidiaries of banks tend to perform worse. This finding is also consistent with Wang (2012) and Fang et al. (2012). Wang (2012) uses accounting measures and finds that bank-affiliated LBOs in the U.K. underperform standalone deals. She argues that this is because bank-affiliated PE firms do not select good targets as other PEs do. Fang et al. (2012) find bank-affiliated deals have worse outcomes if they are consummated during the peaks of the credit market. Controlling for credit market conditions and deal prices in Columns (2)-(4), the results of bank debt proportion, CEO change, PE reputation, and bank affiliation still hold. Moreover, Column (2) shows that the probability of a successful exit strategy is significantly and 36

38 positively related to the Baa spread relative to its historical average over the previous 60 months, after controlling for the absolute level of the spread. This suggests that LBOs are less likely to succeed if they are completed during the time of favorable credit market conditions. This result provides some evidence of market timing of LBO buyers that they tend to overinvest in unprofitable deals that may not exit successfully during the time when the overall credit markets are favorable and leverage is easier to acquire, as suggested by Kaplan and Stein (1993) and Axelson et al. (2009). To summarize, using the exit strategy of IPO or a sale to financial or strategic buyer as an indicator for LBO success, regression analyses show that an LBO is more likely to succeed if it uses more bank debt and tighter loan covenants, experiences no CEO change, and is sponsored by highly reputable PE firms. LBOs are more likely to fail if the buyers are subsidiary of banks that are also financiers of the deal. These results are in general robust to credit market conditions and deal prices. Findings of section provide evidence for the lenders monitoring, the private equity reputation, and the bank affiliation hypotheses. I also find some evidence of the market timing of LBO buyers. Alternatively, I use a failure dummy that takes the value of 1 if an LBO goes bankrupt, enters a distressed exchange, or initiates financial restructuring. Logit regressions using bankruptcy dummy as dependent variables show complementary results that LBOs are more likely to fail if they are sponsored by PE firms of low reputation, are bank-affiliated, and experience CEO change. 7. Conclusion Using a sample of 501 pubic-to-private U.S. LBO transactions completed between 1986 and 2011, I find that better performance is related to larger amount of leverage added during the LBO process, more restrictive covenants of LBO loans, and management contributing equity and participating in the buyout. These results suggest that the main source of value creation in LBOs is the reduced agency costs through the discipline effect of debt, closer monitoring by 37

39 lenders, and the better aligned management incentives. These findings are robust after controlling for the credit market and LBO market conditions, costs of borrowing of target firms, and buyout prices. Findings of this paper deepens our understanding on the observed insignificant performance enhancement in more recent LBOs that use less leverage and relaxed loan covenant, which are important drivers for performance. Using deal outcome as alternative measures of performance, I find that LBO are more likely to exit through a successful strategy (an IPO or a sale to financial or strategic buyers) if they use more bank debt and tighter covenants, experience no CEO change, and are sponsored by PE firms with high reputation. These results are consistent with the lender s monitoring and the private equity reputation hypothesis in value creation in LBOs. Results of this paper suggest that controlling for deal and target characteristics and credit market conditions, private equity reputation is not related to changes in operating performance in the first three years after the buyout but is important in ensuring successful deal outcomes. Future research needs to examine the role of private equity firms in each stage of the buyout process in order to better understand the mechanisms through which reputable PE firms create value. 38

40 References Acharya, Viral V., and Conor Kehoe, 2008, Corporate governance and value creation: Evidence from private equity, Working Paper, London Business School. Axelson, Ulf, Tim Jenkinson, Per Stromberg, and Michael S. Weisbach, 2010, Borrow cheap, buy high? The determinants of leverage and pricing in buyouts, Journal of Finance 68(6), Axelson, Ulf, Per Stromberg, and Michael S. Weisbach, 2009, Why are buyouts levered? The financial structure of private equity funds, Journal of Finance 64, Barry, Christopher B., Steven C. Mann, Vassil T. Mihov, and Mauricio Rodríguez, 2008, Corporate debt issues and the historical level of interest rates, Financial Management 37, Bavaria, Steven M., and Ana Lai, 2007, The leveraging of America Covenant-lite loan structures diminish recovery prospects, Standard & Poor s Ratings Direct, July 18. Bord, Vitaly M., and João A. C. Santos, 2012, The rise of the originate-to-distribute model and the role of banks in financial intermediation, Economic Policy Review 18, Bradley, Michael, and Michael R. Roberts, 2004, The structure and pricing of corporate debt covenants, Working Paper, Duke University. Chava, Sudheer, and Michael R. Roberts, 2008, How does financing impact investment? The role of debt covenants, Journal of Finance 63(5), Cohn, Jonathan B., Lillian F. Mills, and Erin M. Towery, 2011, The evolution of capital structure and operating performance after leveraged buyouts: Evidence from U.S. corporate tax returns, Working Paper, University of Texas at Austin. Colla, Paolo, Filippo Ippolito, and Hannes F. Wagner, 2012, Leverage and pricing of debts in LBOs, Journal of Corporate Finance 18(1), Cumming, Douglas. J., Donald Siegel, and Michael Wright, 2007, Private equity, leveraged buyouts, and governance, Journal of Corporate Finance 13, Demiroglu, Cem, and Christopher M. James, 2010, The role of private equity group reputation in LBO financing, Journal of Financial Economics 96,

41 Diamond, Douglas W., 1984, Financial intermediation and delegated monitoring, Review of Economic Studies 51, Diamond, Douglas W., 1993, Seniority and maturity of bank loan contract, Journal of Financial Economics 33, Fang, Lily, Victoria Ivashina, and Josh Lerner, 2012, Combining banking with private equity investing, Working paper. Gong, James Jianxin, and Steve Yuching Wu, 2011, CEO turnover in private equity sponsored leveraged buyouts, European Financial Management 16, Guo, Shourun, Edith S. Hotchkiss, and Weihong Song, 2011, Do buyouts (still) create value?, Journal of Finance 76, Harford, Jarrad, and Adam Kolasinski, 2011, How do private equity sponsors add value? Evidence from a comprehensive sample of large buyouts and exit outcomes, Working Paper, University of Washington. Higson, Chris, and Rudiger Stucke, 2012, The performance of private equity, Working Paper, London Business School. Holthausen, Robert W., and David F. Larcker, 1996, The financial performance of reverse leveraged buyouts, Journal of Financial Economics 42, Ivashina, Victoria, and Anna Kovner, 2009, The private equity advantage: Leveraged buyout firms and relationship banking, Review of Financial Studies 24, Jensen, Michael C., 1986, Agency costs of free cash flow, corporate finance and takeover, American Economic Review 76, Jensen, Michael C., 1989, Eclipse of the public corporation, Harvard Business Review, sep-oct, Kaplan, Steven N., 1989a, The effect of managements buyouts on operating performance and value, Journal of Financial Economics 24, Kaplan, Steven N., 1989b, Management buyouts: evidence on taxes as a source of value, Journal of Finance 44, Kaplan, Steven N., and Antoinette Schoar, 2005, Private equity performance: returns, persistence, 40

42 and capital flows, Journal of Finance 60, Kaplan, Steven N., and Jeremy C. Stein, 1993, The evolution of buyout pricing and financial structure in the 1980s, Quarterly Journal of Economics 108, Lichtenberg, Frank R., and Donald Siegel, 1990, The effects of leveraged buyouts on productivity and related aspects of firm behavior, Journal of Financial Economics 27, Lopez-de-Silanes, Florencio, Ludovic Phalippou, and Oliver Gottschalg, 2011, Giants at the gate: On the cross-section of private equity investment returns, Working paper. Miller, Steven, 2012, A syndicated loan primer: A guide to the U.S. loan markets, Standard & Poor s Credit Research, September. Officer, Micah S., Oguzhan Ozbas, and Berk A. Sensoy, 2010, Club deals in leveraged buyouts, Journal of Financial Economics 98, Park, Cheol, 2000, Monitoring and structure of debt contracts, Journal of Finance 55, Shivdasani, Anil, and Yihui Wang, 2011, Did structured credit fuel the LBO boom?, Journal of Finance 66, Smith, Abbie J., 1990, Corporate ownership structure and performance: The case of management buyouts, Journal of Financial Economics 27, Stromberg, Per, 2008, The new demography of private equity, Working paper. Wang, Yingdi, 2012, Bank affiliation in private equity firms: distortions in investment selection, Working paper. 41

43 Appendix Table A1: Variable Definitions and Data Sources. Variables Descriptions Sources Panel A: Target Firm Characteristics Assets Book assets measured in 2005 dollars. Capital IQ, Compustat, and SEC Filings Sales Sales measured in 2005 dollars. Same as above EBITDA Earnings before income, tax, depreciation, and amortization, in Same as above 2005 dollars. NCF Net cash flow, calculated as EBITDA minus capital expenditure, Same as above in 2005 dollars. Leverage The ratio of total debt to book assets. Pre-buyout leverage is the Same as above leverage in the last fiscal year prior to the buyout completion and post-buyout leverage is the leverage in the first full fiscal year following the buyout completion. Leverage change Leverage in one year after LBO completion (year +1) minus Same as above leverage one year prior to the LBO completion (year -1) Asset tangibility The ratio of net PP&E to book assets. Same as above Cash flow volatility Standard deviation of EBITDA/Sales in the past 3 years before the LBO completion. Same as above Panel B: LBO Deal Characteristics Transaction Value Total LBO transaction value in 2005 dollars. Capital IQ and SDC Size Ln(Transaction Value) in 2005 dollars. Capital IQ and SDC Club dummy Optimal club dummy Bank affiliation Mgmt participation An indicator variable that takes a value of one if an LBO is sponsored by more than one PE firm. An indicator variable that takes a value of one if an LBO is sponsored by 2 or 3 PE firms. An indicator variable that takes a value of one if an LBO is sponsored by a PE firm that is a subsidiary of a bank and the bank provides loans to the buyout transaction. An indicator variable that takes a value of one if an LBO is labeled by Capital IQ as management buyout, management participated, individual investor participated when the individual investor is confirmed to be board member or management of the target firm, or the firm is bought out through an employee stock ownership plan (ESOP); or if the SDC synopsis labels it as management led or management participated. Capital IQ and SDC Capital IQ and SDC Capital IQ, SDC, and Dealscan Capital IQ and SDC 42

44 Variables Descriptions Sources CEO Change Success An indicator variable that takes a value of one if there is CEO change from the buyout announcement to two full years after the buyout completion. An indicator variable that takes a value of one if an LBO exit through an IPO or a sale to strategic or financial buyer(s), and zero if an LBO goes bankrupt. Capital IQ and Factiva Factiva, SEC filings, Capital IQ Tearsheet, and company website Panel C: LBO Financing Details Bank debt The sum of the revolving credit facility and the Term Loan A Dealscan facility in a debt package. Bank debt % The ratio of bank debt amount to total LBO debt amount Dealscan Institutional Debt Junior Debt Spread Bank debt spread Inst. Debt spread Maturity Bank debt maturity Inst. Debt maturity BR Covenant Intensity The sum of Term Loan B, C, D facilities and notes sold to institutioanl investors. The sum of the bridge loan facility from Dealscan and the mezzanine debt from Capital IQ Package-level spread, calculated as the average spread over reference rate across all debt facilities for a package (weighted by facility size). For each facility, teh spread is the difference between the all-in-drawn interest and the corresponding reference rate defined in the debt contract (e.g., 6-month LIBOR). Same as package-level spread, but using only facilities classified as bank debt. Same as package-level spread, but using only facilities classified as institutional debt. The package-level maturity (in months), calculated as the average maturity across all debt tranches of a deal (weighted by tranche size). Same as package-level maturity, but using only facilities classified as bank debt. Same as package-level maturity, but using only facilities classified as institutional debt. Brandley and Robert (2004)'s covenant intensity index. Sum of: (1) financial covenants (0/1) ; (2) number of sweep covenants (asset sales sweep, debt issue sweep, equity issue sweep); (3) dividend covenant (0/1); (4) secured debt covenant (0/1). Ranges from 0 to 6. From Bradley and Roberts (2004). Dealscan Dealscan and Capital IQ Dealscan Dealscan Dealscan Dealscan Dealscan Dealscan Dealscan 43

45 Variables Descriptions Sources Covenant Intensity Cov-lite loans Deal pricing Modified Covenant Intensity. Sum of: (1) number of financial covenants (up to 6); (2) number of sweep covenants (asset sales sweep, debt issue sweep, equity issue sweep; excess CF sweep, insurance proceeds sweeps); (3) dividend covenant (0/1); (4) secured debt covenant (0/1). Ranges from 0 to 13. Covenant-lite loans, defined as syndicated loans that do not have any financial maintenance covenants. The ratio of EBITDA (at year -1) over total capital of the LBO, adjusted by subtracting the ratio of earnings to price for the S&P 500 in the month the deal is priced. Total capital equals the sum of (1) the market value paid for the firms equity; (2) the value of firm s outstanding debt; and (3) the fees paid in the transaction; less (4) any cash removed from the firm to finance the buyout. Dealscan Dealscan Capital IQ, SDC, Compustat, SEC Filings, S&P Index Data Platforms. Panel D: PE Reputation PE Reputation PE Reputation (N) PE Reputation (N, 70) PE Reputation (TV,70) PE Experience (year) The market share of a PE firm in the past 36 months, based on the total transaction value of all LBOs it sponsor. The market share of a PE firm in the past 36 months, based on the number of LBOs it sponsor The market share of a PE firm since 1970, based on the number of LBOs it sponsor The market share of a PE firm since 1970, based on the total transaction value of LBOs it sponsor PE firm's years of experience, calculated as number of years between the first ever LBO sponsored by a PE to its last LBO. Capital IQ and SDC Capital IQ and SDC Capital IQ and SDC Capital IQ and SDC Capital IQ and SDC Panel D: Market Measures Baa Moody's Seasoned monthly Baa Corporate Bond Yield. FRB's H.15 Release Baa-HBaa Term Hot The difference between the Baa yield in the month of LBO completion and its 60-month historical average. The difference between 10-year T-Bond yield and 3-month T-Bill yield. An indicator variable that takes a value of one if an LBO becomes effective in a month during which a 12-month centred moving average of buyout deals exceeds the long term average of buyout activity on that continent. FRB's H.15 Release FRB's H.15 Release Capital IQ and SDC 44

46 Table A2: Top Private Equity Firms This table ranks private equity (PE) firms by the total dollar amount of transactions and the number of transactions they sponsored in my sample. Total transaction value is inflation-adjusted with the base year of Percentage in the brackets of Column (1) shows the sum of transaction values of all deals sponsored by each PE firm as a proportion of total transaction value of all 501 LBOs. Percentage in brackets of Column (2) presents the ratio of the number of deals sponsored by each PE firm over the total number of LBOs in the sample. Column (3) shows the number of club deals each PE firm participates in where club deals are LBOs sponsored by two or more PE firms. If a PE firm is a subsidiary of a bank at the time of deal announcement and the bank also provides loans for the buyout, the parent bank is listed in Column (4). PEI 2013, 2010, and 2007 indicate the ranking of each PE firm by the Private Equity International (PEI) magazine in the years 2013, 2010, and 2007 based on the capital raised by each PE firm over the previous five-year period. (1) (2) (3) (4) (5) (6) (7) Total Transaction Value Number of Transactions Club Parent Bank PEI PEI PEI Name of Private Equity Firm Rank Value % Rank Number % deals Kohlberg Kravis Roberts & Co. L.P. 1 $ 228,989 (25.45%) 1 27 (5.4%) TPG Capital, L.P. 2 $ 183,501 (20.40%) 2 26 (5.2%) Goldman Sachs Capital Partners 3 $ 142,897 (15.88%) 5 21 (4.2%) 15 Goldman Sachs Bain Capital Private Equity 4 $ 100,316 (11.15%) 6 20 (4.0%) The Blackstone Group 5 $ 83,372 (9.27%) 3 22 (4.4%) The Carlyle Group LP 6 $ 78,450 (8.72%) (3.0%) Thomas H. Lee Partners, L.P. 7 $ 69,421 (7.72%) 7 19 (3.8%) Lehman Brothers Private Equity 8 $ 66,788 (7.42%) 24 6 (1.2%) 1 Lehman Brothers 25 Merrill Lynch Global PE 9 $ 64,841 (7.21%) 15 8 (1.6%) 5 Merrill Lynch Citigroup Private Equity LP 10 $ 55,982 (6.22%) 31 5 (1.0%) Apollo Global Management, LLC 11 $ 49,629 (5.52%) 8 17 (3.4%) Morgan Stanley Private Equity 12 $ 44,419 (4.94%) 88 2 (0.4%) 1 Morgan Stanley Providence Equity Partners LLC 13 $ 42,050 (4.67%) 20 7 (1.4%) Madison Dearborn Partners, LLC 14 $ 37,172 (4.13%) (2.8%) Riverstone Holdings LLC 15 $ 30,017 (3.34%) 52 3 (0.6%) Silver Lake 16 $ 28,807 (3.20%) 33 5 (1.0%) Clayton, Dubilier & Rice, Inc. 17 $ 25,267 (2.81%) 29 6 (1.2%)

47 DLJ Merchant Banking 18 $ 25,024 (2.78%) 4 22 (4.4%) 11 Credit Suisse J.P. Morgan Partners, LLC 19 $ 18,848 (2.10%) (3.0%) 10 JPMorgan Chase 13 Deutsche Bank AG, Investment Arm 20 $ 13,271 (1.48%) 30 5 (1.0%) 4 Deutsche Bank Warburg Pincus LLC 21 $ 13,071 (1.45%) 17 8 (1.6%) Credit Suisse Private Equity, LLC 22 $ 12,770 (1.42%) 18 8 (1.6%) 7 Credit Suisse 33 Canada Pension Plan Investment Board 23 $ 11,545 (1.28%) 87 2 (0.4%) 2 20 Court Square Capital Partners 24 $ 9,704 (1.08%) 9 16 (3.2%) 4 Citigroup Leonard Green & Partners, L.P. 25 $ 9,679 (1.08%) (2.4%) Both DLJ Merchange Banking and the Credit Suisse Private Equity, LLC are PE firms listed as subsidiary of Credit Suisse. But I confirmed that they are different PE firmks. 46

48 Table 1: Comparing the 501 LBOs in the Final Sample with 1, 586 Original LBO Sample This table compares the median and average transaction value between the 1,586 original LBOs collected from Capital IQ and SDC and the 501 LBOs with post-buyout data available from Compustat, Capital IQ, and SEC filings. The percentage in the last column shows the proportion of the deals that have post-lbo financials available and therefore staying in my sample. Transaction values are inflation-adjusted in 2005 dolars. Original 1, LBOs with post-lbo financials Transaction Value ($Mil) Count Transaction Value ($Mil) Count Year median mean median mean % % % % % % % % % % % % % % % % % % % % % % % % % % % 47

49 Table 2: LBO Year and Industry The table classifies transactions by LBO effective year and target firm industry. Eight broad industry classifications are defined according to SIC codes: (1) Agriculture/Fishing/Forestry (SIC 0-999), (2) Mining (SIC ), (3) Construction (SIC ), (4) Manufacturing (SIC ), (5) Transportation/Communication/Electric/Gas (SIC ), (6) Wholesale/Retail (SIC ), (7) Finance/Insurance/Real Estate (SIC ), and (8) Services (SIC ). The percentage in the brackets of Column (9) shows the number of deals in each year as a proportion of total number of deals. The percentage in the brackets of the last row shows the number of deals in each industry as a proportion of total number of deals. Year (1) Agr (2) Mining (3) Constr (4) Mftr (5) Trans (6) Wholesale /Retail (7) Fin (8) Services (9) Total (percentage) (2.4%) (2.4%) (7.2%) (5.4%) (1.6%) (0.2%) (0.4%) (0.8%) (0.6%) (1.2%) (3.0%) (7.2%) (8.2%) (6.0%) (5.4%) (2.2%) (3.0%) (5.6%) (7.8%) (6.2%) (4.6%) (10.2%) (1.6%) (0.6%) (3.8%) (2.6%) Total (1.2%) (2.0%) (0.8%) (44.5%) (8.2%) (19.4%) (3.2%) (20.0%) 100% 48

50 Table 3: Median Changes in Operating Performance This table presents median changes in operating performance. All variables are defined in Appendix Table A1. Year -1 is the last fiscal year prior to the buyout completion. Years +1, +2, and +3 are the first, second, and third full fiscal year following the buyout completion. Industry-adjusted changes subtract the medians for firms in the same industry based on four-digit SIC code. Significance levels of medians are based on a two-tailed Wilcoxon rank test. ***, **, and * next to the percentage change denote levels that are significantly different from zero at 1%, 5%, and 10% level, respectively. Significance levels of time trend are based on the nonparametric trend test, and ***, **, and * next to the time trend bracket denote the nonparametric trend test statistics is significant at 1%, 5%, and 10% level, respectively. Panel A: Median Changes in Operating Performance between 1986 and to -1-1 to +1-1 to +2-1 to +3-2 to -1-1 to +1-1 to +2-1 to +3 Unadjusted Changes Industry-adjusted Changes NO. Obs EBITDA/sales 1.0%*** 0.8% -0.1% -0.8% 3.7%*** 7.0%** 6.9%** 8.9%** NCF/sales 0.1% 0.5% -3.8% -2.6% 9.6%*** 18.5%** 13.9%** 19.6%*** EBITDA/assets 9.6%*** -26.3%*** -26.6%*** -24.1%*** 13.0%*** -12.2%** -9.8%** -6.1% NCF/assets 7.2%*** -29.7%*** -31.0%*** -26.5%*** 12.7%*** -5.7%* 5.3% 3.4% NO. Obs Panel B: Median Changes in Operating Performance in Sub-Periods -2 to -1-1 to +1-1 to +2-1 to +3-2 to -1-1 to +1-1 to +2-1 to +3 Unadjusted Changes Industry-adjusted Changes EBITDA/sales % 7.6%*** 4.7% 2.5% 3.3% 10.6%** 8.1%* 8.4%* %** -0.3% -2.8% -3.6%* 2.7%** 7.7%** 6.1%* 10.7%** %** 0.0% 0.1% -0.7% 5.1%*** 4.8% 6.6% 8.5% time trend (-) (-)** (+) (-) (+) (-)* (-)* (-) NCF/sales % 15.3%** 13.4%** 11.4%** 5.7% 32.7%*** 28.2%** 31.5%** % -2.3% -7.0% -3.0% 6.0%** 18.5%** 13.7%** 29.8%*** % -2.8% -4.0% -7.4% 12.9%*** 13.3%** 8.4% 5.7% time trend (+)** (-)** (-) (-) (+) (-)*** (-)*** (-)** 49

51 EBITDA/assets -2 to -1-1 to +1-1 to +2-1 to +3-2 to -1-1 to +1-1 to +2-1 to +3 Unadjusted Changes Industry-adjusted Changes %* -18.2%*** -22.4%*** -14.2%*** 7.6%*** -8.9%* 0.0% 0.0% %*** -17.7%*** -17.8%*** -21.2%*** 18.6%*** 0.9% -4.7% -0.2% %*** -34.0%*** -34.4%*** -32.1%*** 13.2%*** -27.5%*** -22.5%*** -11.5%** time trend (+) (-)*** (-)*** (-)* (+) (-)*** (-)*** (-)** NCF/assets % -12.7% -16.3%* -11.4% 7.1% 10.1% 14.9% 13.4% %** -20.7%** -20.2%* -16.1%* 11.7%*** 7.7%** 20.5%*** 19.2%*** %*** -40.3%*** -40.7%*** -44.9%*** 15.3%*** -24.2%*** -6.7% -8.4% time trend (+)** (-)*** (-)* (-)** (+) (-)*** (-)** (-)*** 50

52 Table 4: Leverage, Debt Structure, and Debt Contractual Features Panel A: Leverage This table presents the annual medians of leverage. All variables are defined in Appendix Table A1. ***, **, and * denote that the nonparametric trend test statistics is statistically significant at 1%, 5%, and 10% level, respectively. Year (1) Pre-buyout leverage (%) (2) Post-buyout leverage (%) (3) Leverage change (%) (4) Compustat Leverage change (%) Time Trend (+) (-)*** (-)** (-) 51

53 Panel B: LBO Debt Structure This table presents the structure of LBO debt. All variables are defined in Appendix Table A1. ***, **, and * denote that the nonparametric trend test statistics is statistically significant at 1%, 5%, and 10% level, respectively. (1) (2) (3) (4) (5) Revolvers (%) Term Loan A (%) Bank Debt=(1)+(2)(%) Institutional Debt (%) Junior Debt (%) Year Revolver to LBO debt %with Revolver Term A to LBO debt %with Term A Bank debt to LBO debt %with Bank debt Inst. debt to LBO debt %with Inst. debt Junior debt to LBO debt %with Junior debt Time Trend (-)* (-) (-)** (-)** (-)*** (-)* (+)*** (+) (-) (-) 52

54 Year Panel C: Spread, Maturity, and Loan Covenant This table compares the median spread and maturity for bank debt versus institutional debt. It also presents the median covenant intensity measure using the Bradley and Roberts (2004) s measure, a modified measure proposed by the author, and the proportion of Cov-lite loans. All variables are defined in Appendix Table A1. ***, **, and * denote that the nonparametric trend test statistics is statistically significant at 1%, 5%, and 10% level, respectively. (1) Bank debt Spread (2) Inst. debt Spread (3) Spread Diff Inst. vs Bank (4) Bank debt Maturity (months) (5) Inst. debt Maturity (months) (6) Maturity Diff Inst. vs Bank(months) (7) BR Covenant Intensity (8) (Modified) (52%) 1 Covenant Intensity (34%) (56%) (73%) (64%) (60%) (10%) (57%) (94%) (77%) 5 77% (86%) 8 26% (74%) % (82%) 9 29% (78%) 10 31% (76%) 8 29% (82%) 8 26% (88%) 9 19% (78%) % (95%) 9 22% (92%) 9 22% (80%) 7 40% (89%) 6 39% (64%) 4 39% (9) Cov-lite (0%) 0 100% (81%) 1 64% (77%) 2 54% Time Trend ( ) (+)*** (+) (-)** (+) (-)*** (-)** (-)** ( ) 7 ( ) (-)*** ( ) Loans (%) 34% (+)* 53

55 Table 5: Private Equity Involvement and Reputation Panel A: Private Equity Involvement This table presents the involvement of private equity (PE) firms in LBO transactions over time. All variables are defined in Appendix Table A1. ***, **, and * denote that the nonparametric trend test statistics is statistically significant at 1%, 5%, and 10% level, respectively. Year (1) # LBOs with PE (2) # LBOs with Prominent PE (3) # LBOs with Bank-Affiliated (83%) 8 (67%) 6 (50%) (75%) 4 (33%) 3 (25%) PE (4) Number of Club Deals (5) Total Transaction Value ($million) of Club Deals (58%) 8 (22%) 7 (19%) 1 (3%) 233 (1%) (74%) 10 (37%) 10 (37%) 4 (15%) 11,126 (14%) (63%) 3 (38%) 1 (13%) (100%) (100%) 1 (50%) 1 (50%) 1 (50%) 723 (87%) (25%) (100%) 2 (67%) 1 (33%) 1 (33%) 450 (50%) (100%) 5 (83%) 1 (17%) (100%) 6 (40%) 3 (20%) 4 (27%) 3,880 (39%) (100%) 16 (44%) 5 (14%) 10 (28%) 5,601 (30%) (98%) 21 (51%) 12 (29%) 10 (24%) 7,425 (24%) (100%) 16 (53%) 7 (23%) 10 (33%) 5,978 (34%) (93%) 9 (33%) 7 (26%) 12 (44%) 8,069 (31%) (73%) 3 (27%) 2 (18%) 5 (45%) 2,950 (54%) (87%) 6 (40%) 2 (13%) 6 (40%) 2,487 (52%) (86%) 18 (64%) 4 (14%) 6 (21%) 5,452 (17%) (92%) 24 (62%) 4 (10%) 16 (41%) 17,011 (44%) (100%) 20 (65%) 5 (16%) 14 (45%) 47,465 (72%) (91%) 15 (65%) 4 (17%) 12 (52%) 72,038 (88%) (94%) 35 (69%) 15 (29%) 22 (43%) 179,252 (66%) (100%) 7 (88%) 1 (13%) 3 (38%) 49,625 (85%) (100%) 1 (33%) 1 (33%) 1 (33%) 6,107 (91%) (100%) 12 (63%) 1 (5%) 5 (26%) 5,135 (23%) (100%) 11 (85%) 0 (0%) 6 (46%) 18,554 (64%) (89%) 261 (52%) 103 (21%) 149 (30%) 449,562 (50%) Time trend (6) Max # of PEs in the (+)*** (+)*** (-) (+)*** (+)*** (+)** club 54

56 Panel B: Relation between PE Reputation, Participation, and LBO Deal Characteristics This table presents the relation between PE firms reputation, their different forms of participation in LBOs, including club deals and bank-affiliated deals, and LBO deal characteristics and financing structure. I run univariate OLS regressions using various measures of LBO deal and financing characteristics as dependent variable and reputation and participation as explanatory variable with year dummies. I only report the coefficient estimates and the p-value. All variables are defined in Appendix Table A1. All regressions are OLS with heteroskedasticity adjusted standard errors. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. (1) (2) (3) (4) (5) (7) (6) (8) (9) (10) Size Pre-buyout leverage Leverage change Revolvers to LBO Debt Term A to LBO Debt Bank debt% Inst. debt% Covenant Intensity Spread Maturity PE Reputation *** ** 1.639*** ** *** 0.800** *** 6.591** PE Reputation (N) ** ** 5.502* * * * PE Reputation (N, 70) *** * 5.278** ** ** 4.033** ** 2.699** Ln(PE Experience (year) 0.119** *** *** 0.039*** 0.287*** *** 3.827*** Bank affiliation 0.668*** ** Club dummy 0.386*** # PE in club 0.266*** ** **

57 Table 6: Regression for Post-buyout Performance: Baseline Regression This table reports the multivariate regression results for post-buyout performance. Panel A tests the debt discipline and lenders monitoring hypotheses. Panel B examines the impact of PE firms reputation and their different form of participation. Panel C first tests the management participation hypothesis and then include all potential performance drivers. Dependent variables are the industry-adjusted percentage changes in EBITDA/sales or NCF/sales from the last full pre-buyout year (year -1) to the second full year after deal completion (year +2). Pre-buyout target characteristics include leverage ratios and asset tangibility at the end of year -1, industry-adjusted changes in EBITDA/sales or NCF/sales from year -2 to year -1, and target firms cash flow volatiles in the last three years before buyout completion (years -3, -2, and -1). All variables are defined in Appendix Table A1. P-values are in brackets. All regressions are OLS with heteroskedasticity adjusted standard errors. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. VARIABLES Panel A: Debt Disciplining and Lenders Monitoring Hypotheses (1) EBITDA/sales (2) NCF/sales (3) EBITDA/sales (4) NCF/sales Leverage change 0.348** 0.780** 0.265** 0.515** [0.013] [0.012] [0.033] [0.011] Bank debt % [0.666] [0.952] Covenant intensity 0.025** 0.081** [0.019] [0.046] Maturity [0.468] [0.540] Constant [0.595] [0.893] [0.273] [0.332] Year dummies Yes Yes Yes Yes Pre-LBO target characteristics Yes Yes Yes Yes Observations R-squared

58 Panel B: Private Equity Reputation, Club Deal, and Bank-affiliation Hypotheses VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) EBITDA /Sales NCF /Sales EBITDA /Sales NCF /Sales EBITDA /Sales NCF /Sales EBITDA /Sales NCF /Sales Leverage change 0.351** 0.780** 0.279** 0.535* 0.281** 0.556* 0.279** 0.555* [0.013] [0.012] [0.042] [0.089] [0.041] [0.078] [0.043] [0.078] Bank debt % [0.732] [0.961] [0.736] [0.941] [0.752] [0.929] Covenant intensity 0.028*** 0.099** 0.028*** 0.096** 0.028*** 0.095** [0.003] [0.035] [0.002] [0.050] [0.002] [0.039] PE Reputation [0.543] [0.627] Club dummy [0.657] [0.254] Optimal club dummy [0.636] [0.455] Bank affiliation [0.721] [0.712] Constant [0.570] [0.774] [0.269] [0.435] [0.328] [0.391] [0.262] [0.313] Year dummies Yes Yes Yes Yes Yes Yes Yes Yes Pre-LBO target characteristics Yes Yes Yes Yes Yes Yes Yes Yes Observations R-squared

59 Panel C: Management Participation Hypothesis and All Variables VARIABLES (1) (2) (3) (4) (5) (6) EBITDA /Sales NCF /Sales EBITDA /Sales NCF /Sales EBITDA /Sales NCF /Sales Leverage change 0.297** 0.563* 0.278** 0.552* 0.293** 0.527* [0.0258] [0.0772] [0.0403] [0.0810] [0.0279] [0.0904] Bank debt % [0.683] [0.943] [0.854] [0.961] [0.797] [0.996] Covenant intensity 0.025*** 0.093* 0.028*** 0.094* 0.025** 0.098* [0.009] [0.061] [0.002] [0.078] [0.010] [0.064] PE Reputation [0.843] [0.247] Club dummy [0.607] [0.451] Bank affiliation [0.485] [0.933] Mgmt participation 0.190** ** [0.024] [0.839] [0.046] [0.726] CEO change [0.243] [0.667] [0.222] [0.663] Constant [0.498] [0.333] [0.401] [0.302] [0.521] [0.456] Year dummies Yes Yes Yes Yes Yes Yes Pre-LBO target characteristics Yes Yes Yes Yes Yes Yes Observations R-squared

60 Table 7: Regression for Post-buyout Performance: Subsample Analyses This table reports the multivariate regression results for post-buyout performance by each LBO wave. Panel A shows regression results by each LBO wave and the sample period that excludes deals completed in 2006 and Panel B presents results by management and PE firms involvement. All dependent variables are the industry-adjusted percentage changes in EBITDA/sales from the last full pre-buyout year (year -1) to the second full year after deal completion (year +2). Pre-buyout target characteristics include leverage ratios and asset tangibility at the end of year -1, industry-adjusted changes in EBITDA/sales from year -2 to year -1, and target firms cash flow volatiles in the last three years before buyout completion (years -3, -2, and -1). All variables are defined in Appendix Table A1. P-values are in brackets. All regressions are OLS with heteroskedasticity adjusted standard errors. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. Panel A: By LBO Waves VARIABLES (1) (2) (3) (4) EBITDA/sales Wave1: EBITDA/sales Wave 2: EBITDA/sales Wave 3: EBITDA/sales Exclude 06&07 Leverage change 0.187* 0.288* 0.580*** 0.272* [0.527] [0.065] [0.007] [0.067] Bank debt % [0.707] [0.777] [0.633] [0.863] Covenant intensity 0.059** 0.025** 0.032** [0.022] [0.012] [0.023] Maturity * [0.189] [0.066] [0.636] [0.150] PE reputation * [0.480] [0.480] [0.060] [0.443] Club dummy [0.256] [0.587] [0.275] [0.340] Bank affiliation [0.271] [0.299] [0.714] [0.322] Mgmt participation 0.169** 0.497** ** [0.047] [0.015] [0.339] [0.037] CEO change [0.223] [0.397] [0.544] [0.400] Constant [0.425] [0.284] [0.473] [0.432] Year dummies Yes Yes Yes Yes Pre-LBO target characteristics Yes Yes Yes Yes Observations R-squared

61 Panel B: By Participants (1) EBITDA/sales (2) EBITDA/sales (3) EBITDA/sales (4) EBITDA/sales VARIABLES Mgmt-participated No Mgmt PE-participated No PE Leverage change 0.452* 0.180** 0.292** 0.261* [0.054] [0.017] [0.037] [0.090] Bank debt % 0.547** [0.034] [0.307] [0.690] [0.586] Covenant intensity ** 0.020* [0.319] [0.042] [0.095] [0.576] Maturity [0.712] [0.392] [0.547] [0.793] PE reputation [0.428] [0.148] [0.637] Club dummy [0.906] [0.766] [0.617] Bank affiliation [0.730] [0.170] [0.268] Mgmt participation 0.192** [0.023] [0.617] CEO change ** [0.033] [0.930] [0.518] [0.213] Constant [0.355] [0.153] [0.211] [0.945] Year dummies Yes Yes Yes Yes Pre-LBO target characteristics Yes Yes Yes Yes Observations R-squared

62 Table 8: Market Timing This table reports the multivariate regression results for post-buyout performance. Dependent variables are the industry-adjusted percentage changes in EBITDA/sales from the last full pre-buyout year (year -1) to the second full year after deal completion (year +2). All variables are defined in Appendix Table A1. P-values are in brackets. All regressions are OLS with heteroskedasticity adjusted standard errors. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. VARIABLES (1) EBITDA/sales (2) EBITDA/sales (3) EBITDA/sales (4) EBITDA/sales Leverage change 0.316** 0.290** 0.314** 0.301** [0.029] [0.031] [0.021] [0.025] Bank debt % [0.574] [0.660] [0.564] [0.832] Covenant intensity 0.030** 0.025** 0.026*** 0.024** [0.020] [0.011] [0.008] [0.012] PE reputation [0.925] [0.859] [0.862] [0.959] Mgmt participation 0.188** 0.190** 0.181** 0.196** [0.031] [0.026] [0.035] [0.023] Baa [0.373] Baa-HBaa [0.600] Term [0.801] Hot [0.524] Spread [0.140] Deal Pricing [0.205] Constant [0.479] [0.475] [0.453] [0.558] Year dummies No No Yes Yes Pre-LBO target Yes Yes Yes Yes characteristics Observations R-squared

63 Table 9: LBO Year and Exit Strategy This table presents the post-buyout outcomes as of June 30, The number of observations for each exit strategy is reported, followed in parentheses by the proportion of the outcome in all LBOs each year. Year (1) Bankruptcy (2) IPO (3) Acquired by Corp (4) Secondary LBOs (5) still private (6) Unknown (41.7%) 3 (25.0%) 2 (16.7%) 2 (16.7%) 0 (0.0%) 0 (0.0%) (50.0%) 4 (33.3%) 1 (8.3%) 1 (8.3%) 0 (0.0%) 0 (0.0%) (22.2%) 19 (52.8%) 6 (16.7%) 2 (5.6%) 0 (0.0%) 1 (2.8%) (22.2%) 6 (22.2%) 11 (40.7%) 2 (7.4%) 1 (3.7%) 1 (3.7%) (0.0%) 3 (37.5%) 3 (37.5%) 0 (0.0%) 0 (0.0%) 2 (25.0%) (100.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) (0.0%) 0 (0.0%) 1 (50.0%) 1 (50.0%) 0 (0.0%) 0 (0.0%) (0.0%) 1 (25.0%) 1 (25.0%) 1 (25.0%) 0 (0.0%) 1 (25.0%) (0.0%) 2 (66.7%) 1 (33.3%) 0 (0.0%) 0 (0.0%) 0 (0.0%) (0.0%) 3 (50.0%) 2 (33.3%) 1 (16.7%) 0 (0.0%) 0 (0.0%) (40.0%) 6 (40.0%) 1 (6.7%) 2 (13.3%) 0 (0.0%) 0 (0.0%) (19.4%) 15 (41.7%) 7 (19.4%) 5 (13.9%) 1 (2.8%) 1 (2.8%) (24.4%) 11 (26.8%) 11 (26.8%) 7 (17.1%) 2 (4.9%) 0 (0.0%) (20.0%) 15 (50.0%) 5 (16.7%) 3 (10.0%) 1 (3.3%) 0 (0.0%) (7.4%) 8 (29.6%) 11 (40.7%) 3 (11.1%) 1 (3.7%) 2 (7.4%) (9.1%) 1 (9.1%) 1 (9.1%) 6 (54.5%) 1 (9.1%) 1 (9.1%) (13.3%) 6 (40.0%) 1 (6.7%) 5 (33.3%) 1 (6.7%) 0 (0.0%) (7.1%) 11 (39.3%) 4 (14.3%) 8 (28.6%) 3 (10.7%) 0 (0.0%) (15.4%) 23 (59.0%) 5 (12.8%) 2 (5.1%) 3 (7.7%) 0 (0.0%) (9.7%) 17 (54.8%) 4 (12.9%) 2 (6.5%) 5 (16.1%) 0 (0.0%) (21.7%) 5 (21.7%) 1 (4.3%) 3 (13.0%) 8 (34.8%) 1 (4.4%) (11.8%) 13 (25.5%) 4 (7.8%) 2 (3.9%) 26 (51%) 0 (0.0%) (0.0%) 2 (25.0%) 0 (0.0%) 0 (0.0%) 6 (75.0%) 0 (0.0%) (0.0%) 1 (33.3%) 1 (33.3%) 0 (0.0%) 1 (33.3%) 0 (0.0%) (0.0%) 2 (10.5%) 1 (5.3%) 1 (5.3%) 15 (78.9%) 0 (0.0%) (7.7%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 12 (92.3%) 0 (0.0%) Total 83 (16.6%) 177 (35.3%) 85 (17.0%) 59 (11.8%) 87 (17.4%) 10 (2.0%) Months to Outcome Median[mean] 43 [49.5] 36 [39.5] 50 [57.9] 44 [48.6] 62

64 Table 10: Deal Outcome The table presents the results from a logit regression where the dependent variable is equal to 1 if an LBO exits through an IPO or a sale to financial or strategic buyers and 0 otherwise. All variables are defined in Appendix Table A1. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. (1) Success (2) Success (3) Success (4) Success (5) Success Leverage change [0.563] [0.992] [0.697] [0.558] [0.552] Bank debt % 0.834** 0.667* 0.912** 0.844** 0.862** [0.029] [0.085] [0.020] [0.029] [0.025] Covenant intensity 0.079** ** 0.079** 0.079** [0.028] [0.613] [0.048] [0.027] [0.028] Maturity [0.118] [0.116] [0.107] [0.117] [0.127] PE Reputation 0.288*** 0.250** 0.306*** 0.288*** 0.288*** [0.009] [0.019] [0.005] [0.009] [0.009] Bank affiliation *** ** *** ** *** [0.009] [0.013] [0.007] [0.010] [0.009] Club dummy [0.658] [0.702] [0.597] [0.650] [0.640] Mgmt participation [0.406] [0.518] [0.515] [0.412] [0.412] CEO change *** *** *** *** *** [0.000] [0.000] [0.000] [0.000] [0.000] Baa ** [0.029] Baa-Hbaa 0.437** [0.029] Term [0.181] Hot 0.731*** [0.005] Spread [0.872] Deal pricing [0.448] Constant 152.5*** 565.4*** 154.1*** 153.2*** 153.5*** [0.000] [0.000] [0.000] [0.000] [0.000] Year dummies Yes No No Yes Yes Observations Pseudo R-squared

65 Figure 1: A typical LBO Transaction and Hypotheses in LBO Value Creation Lenders Banks Institutional investors Public debt holders H1.2: Lenders Monitoring (bank loan proportion, covenant, maturity) H1.1: Debt Disciplining H2.3: Bank Affiliation Debt Exit IPO Equity Investors Sale Private Equity firms Management of Targets Equity Operational Improvement H2.1: Private Equity Reputation H2.2: Club Deal PE s Return on Equity H3: Management Participation 64

66 Figure 2: LBO Transactions Each Year The figure shows the number of LBO deals and total transaction value by LBO effective year. The solid line that corresponds to the left y-axis plots the number of LBO deals each year. The bar that corresponds to the right y-axis shows the inflation-adjusted total transaction value, based on the 2005 dollar. LBO transaction sample is constructed from the Standard and Poor s Capital IQ and the Securities Data Company s (SDC) U.S. Mergers and Acquisitions Database. 65