SEARCH

SEARCH BY CITATION

Keywords:

  • buy-outs;
  • private equity firms;
  • IPO;
  • acquisitions;
  • SMBO;
  • survival analysis

Abstract

  1. Top of page
  2. Abstract
  3. 1. INTRODUCTION
  4. 2. HYPOTHESES FOR DETERMINANTS OF BUY-OUT LONGEVITY
  5. 3. DATA AND SUMMARY STATISTICS
  6. 4. UNIVARIATE ANALYSIS ACROSS PE BACKING, EXIT ROUTES AND VINTAGE DECADES
  7. 5. BUY-OUTS SURVIVAL ANALYSIS
  8. 6. ESTIMATION RESULTS
  9. 7. ROBUSTNESS OF RESULTS AND EXTENSIONS
  10. 8. DISCUSSIONS AND CONCLUSION
  11. REFERENCES

Abstract:  This paper examines 1,089 private equity (PE) backed and non-PE backed (pure) UK buy-outs, determinants of their survival, and their exit behaviour during the period of 1966–2004. Our results suggest that 56% of the pure sample buy-outs remained in a buy-out organisational form for at least seven years after the original buy-out transaction, thus lending support to views that buy-outs present long rather than short term form. PE backed buy-outs exhibit higher exit rates, fewer early (within 12 months) exits and fewer liquidations than their pure counterparts. Buy-outs sponsored by PE syndicates, those harvested during periods with strong market conditions and greater supply of PE funding, tend to have shorter longevity. The most notable difference between the survival experiences of PE backed and pure buy-outs is documented in IPO exits, where a significant number of pure buy-outs exit early to the Alternative Investment Market (AIM).


1. INTRODUCTION

  1. Top of page
  2. Abstract
  3. 1. INTRODUCTION
  4. 2. HYPOTHESES FOR DETERMINANTS OF BUY-OUT LONGEVITY
  5. 3. DATA AND SUMMARY STATISTICS
  6. 4. UNIVARIATE ANALYSIS ACROSS PE BACKING, EXIT ROUTES AND VINTAGE DECADES
  7. 5. BUY-OUTS SURVIVAL ANALYSIS
  8. 6. ESTIMATION RESULTS
  9. 7. ROBUSTNESS OF RESULTS AND EXTENSIONS
  10. 8. DISCUSSIONS AND CONCLUSION
  11. REFERENCES

The presence of active investors, an efficient capital structure and strong managerial incentives are recognised as key features of buy-outs that contribute to improved firm efficiency (Jensen, 1989). Consistent with Jensen's predictions, the buy-out market has grown tremendously since the 1980s in terms of volume, industry scope and variety of exit routes, and has become a global phenomenon (Stromberg, 2008). While the US market is dominated by venture capital (VC) firms that focus on young ventures, the UK market is dominated by private equity (PE) firms that focus on late stage investments in buy-outs.1 The UK PE market is the largest single European market and equals the rest of the European markets put together (EVCA, 2001). The recent credit crunch, however, has highlighted the cyclical nature of PE investments and exposed the UK PE industry to significant criticism. For example, PE firms have been accused of not being involved in the investee companies long enough and of exiting too early to take advantage of favourable market conditions.2

In this paper we study how long buy-outs remain in their original buy-out structure and private ownership. We examine the longevity of PE backed and non-PE backed (i.e., pure) buy-outs separately, thus shedding more light on the alleged myopic behaviour of PE firms. Unlike their PE backed counterparts, pure buy-outs do not face constraints to exit within a certain period and thus provide the ‘ultimate test’ of buy-outs longevity.

Our analysis significantly extends the analyses considered in previous studies on the longevity of buy-outs. Early UK studies examined buy-outs from the 1980s (Wright et. al., 1993; Wright et. al., 1994; and Wright et. al., 1995) while more recent studies have examined samples dominated by PE backed leveraged buy-outs (LBOs) (Stromberg, 2008; and Nikoskelainen and Wright, 2007) and buy-outs exiting via initial public offerings (IPOs) (Jelic et al., 2005). We analyse a unique dataset consisting of 1,089 buy-outs of all types (management buy-out (MBO), management buy-in (MBI), and LBO), sources (divestment, privatisation, receivership, other privately owned) and exit routes (IPO, sales, secondary management buy-outs (SMBO), liquidations), together with data on PE backing of our sample buy-outs, during the period of January 1966 to December 2004. Secondly, we extend the analysis by considering a large number of IPO exits listed on the second board of the London Stock Exchange (LSE). The very successful second board (i.e., Alternative Investment Market (AIM)) is an important institutional feature of the UK market, facilitating quicker and cheaper listings for smaller firms. To the best of our knowledge, there has been no previous separate examination of longevity of buy-outs exiting in the AIM and their comparison with their counterparts in the main market. Finally, we add to the literature by using survival models that enable us to examine determinants of longevity and to show the evolution of survival rates over time, across different exit routes and across PE backing.

The results suggest that 35% of our sample buy-outs remained in a buy-out organisational form, whilst 47% remained in private ownership for at least seven years after the original buy-out transactions. In the subsample of pure buy-outs the relevant percentages are 56% and 49% respectively. PE backed buy-outs exhibit higher exit rates and fewer early exits and liquidations compared to their pure counterparts. The most popular exit routes are IPOs, followed by sales, SMBOs and liquidations. Although the order of preference of exit routes is the same for PE backed and pure buy-outs, IPOs and SMBOs tend to play a less important role for pure buy-outs. Early exits tend to be associated with IPOs of relatively smaller buy-outs in the AIM. Overall, the results lend support to the view that buy-outs present long rather than short term organisational form.

Our survival models identify characteristics of buy-outs (e.g., size) and PE backing (e.g., syndicates and highly leveraged deals), together with market conditions, as significant determinants of buy-out longevity. The survival models reveal important changes in the survival rates across time, and differences in the survival experience of the buy-outs across different exit routes. For example, the exit hazard rates start increasing for IPOs first, followed by sales, SMBOs and liquidations. There is also evidence that PE firms tend to use sales and SMBOs to exit buy-outs that had not exited via IPOs within four years of the original deals. Our results are robust to endogeneity, heterogeneity, clustering, and alternative specification of variables.

The remainder of the paper is organised as follows. In Section 2, we provide a summary of the relevant literature and develop testable hypotheses. Section 3 presents the data and summary statistics. Section 4 deals with the univariate analysis. Survival models for longevity and alternative exit routes are discussed in Section 5. Section 6 presents the estimated results. In Section 7, we test for robustness of our results and perform further analysis. Finally, we discuss the results and conclude in Section 8.

2. HYPOTHESES FOR DETERMINANTS OF BUY-OUT LONGEVITY

  1. Top of page
  2. Abstract
  3. 1. INTRODUCTION
  4. 2. HYPOTHESES FOR DETERMINANTS OF BUY-OUT LONGEVITY
  5. 3. DATA AND SUMMARY STATISTICS
  6. 4. UNIVARIATE ANALYSIS ACROSS PE BACKING, EXIT ROUTES AND VINTAGE DECADES
  7. 5. BUY-OUTS SURVIVAL ANALYSIS
  8. 6. ESTIMATION RESULTS
  9. 7. ROBUSTNESS OF RESULTS AND EXTENSIONS
  10. 8. DISCUSSIONS AND CONCLUSION
  11. REFERENCES

There is extensive literature on the performance, efficiency gains and issues associated with corporate governance mechanisms of buy-outs (see the survey by Thomson and Wright, 1995) and duration of early stage VC investments (Cumming, 2008; Schwienbacher, 2002; Cumming and Johan, 2010; and Abdulkadir, 2009). However, there is a notable paucity of literature on the longevity of PE late stage investments in buy-outs and the different exit routes from buy-out governance structures. A separate examination of the duration of PE investments in buy-outs is important for the following reasons. First, investments in buy-outs require different skills from investments in new companies, confirmed by a clear division between PE and VC specialists. Second, PE investments in buy-outs require significant incremental direct and overhead costs especially in relation to large and more complex buy-outs (e.g., privatisations and receiverships) in comparison to investments made by VC firms. For example, investments in buy-outs could be burdened with complicated ownership and organisational issues less likely to be present in VC investments. Third, the financing model of PE firms is quite different from the VC financing model which could be of importance for investment duration.

We therefore formulate empirical predictions for four groups of determinants of buy-outs’ longevity: determinants associated with PE backing and characteristics of the PE firms (reputation and association with investment banks) and PE deals (syndicated and highly leveraged deals); buy-out specific characteristics such as size, industry, and source of the buy-outs (privatisation, divestment, and receivership); buy-out type (LBO, MBO and MBI); and determinants related to the market conditions (changes in the stock market, hot IPO periods and supply of PE funding). In Table 1, we define all variables used in the empirical section of the paper. A summary of all the hypotheses is presented in Table 2.

Table 1.  Definitions of Variables Used in the Analysis of Longevity and Exits
VariableDefinition
Company characteristics 
 SIZESIZE is the inflation-adjusted value of buy-out transactions (million £) in vintage years.
 LNSIZEThe natural logarithm of SIZE.
 MAINA categorical variable that takes the value of one if the company is listed on the main board of the London Stock Exchange, and zero if listed on the second board of the London Stock Exchange (Alternative Investment Market or its predecessor, Unlisted Securities Market).
 AIMA categorical variable that takes the value of one if the company is listed on the second board of the London Stock Exchange (Alternative Investment Market or its predecessor, Unlisted Securities Market), and zero if listed on the main board of the London Stock Exchange.
 MANUFACTURINGA categorical variable that takes the value of one if the company is from the manufacturing industries, and zero otherwise.
 SERVICESA categorical variable that takes the value of one if the company is from the service industries (Business Services in Gompers et al., 2008), and zero otherwise.
Buy-out type 
 MBIA categorical variable that takes the value of one if the company is a management buy-in (MBI), and zero otherwise. MBI is a transaction where the external managers takeover the company.
 MBOA categorical variable that takes the value of one if the company is a management buy-out (MBO), and zero otherwise. MBO is a transaction where the internal managers takeover the company.
Highly leveraged deal 
 LBOA categorical variable that takes the value of one if the company is identified as a leveraged buy-out, and zero otherwise; the definition of LBOs is consistent with Amess and Wright (2007).
 LBOTA categorical variable that takes the value of one if the company is identified as a leveraged buy-out in the Thomson One Banker database, and zero otherwise. According to the database, a leverage buy-out occurs when an investor or firm offers to acquire a company, taking on an extraordinary amount of debt, with plans to repay it with funds generated from the company or with revenue earned by selling off the newly-acquired company's assets. In addition, the Thomson One Banker database identifies a transaction as an LBO if the investor group includes management or if the transaction is identified as such in the financial press and a majority interest of the target company is acquired.
Buy-out source 
 RECEIVERSHIPA categorical variable that takes the value of one if the source of a buy-out deal is a company in the receivership process, and zero otherwise.
 DIVESTMENTA categorical variable that takes the value of one if the source of a buy-out deal is a divestment transaction, and zero otherwise. Divestment indicates that the deal is a divestiture where the parent company is losing a majority interest in the target or the target company is disposing of assets.
 PRIVATISATIONA categorical variable that takes the value of one if the source of a buy-out deal is a privatisation transaction, and zero otherwise. Privatisation indicates that the government, or a government controlled entity, sells assets to a management team.
 OTHERA categorical variable that takes the value of one if the source of a buy-out deal is a family or other privately-owned company, and zero otherwise.
Market conditions 
 LNFTSEThe natural logarithm of the FTSE All Share Index returns during 12 months prior to exit.
 LNFTSEPERIODThe natural logarithm of the average returns on the FTSE All Share Index over the entire investment period.
 LNPECAPITALThe natural logarithm of the average annual changes in the, inflation adjusted, total amount of the UK private equity investments, during investment period.
 HOTA categorical variable that takes the value of one if the buy-out exited during a hot IPO market. The hot IPO market years satisfy at least two out of the following three criteria: abnormal initial IPO returns, abnormal IPO volume, and non-negative autocorrelation in IPO volume. This classification is consistent with the definition used in studies on the UK hot IPO market (See Coakley et al., 2008).
PE backing 
 PEA categorical variable that takes the value of one if the buy-out received backing from a private equity (PE) firm, and zero otherwise.
 TOP3A categorical variable that takes the value of one if the buy-out received backing from one of the top three PE firms ranked by the reputation score (1), and zero otherwise.
 TOP10A categorical variable that takes the value of one if the buy-out received backing from one of the top ten PE firms ranked by the reputation score (1), and zero otherwise.
 INDEPENDENTA categorical variable that takes the value of one if the buy-out received backing from an independent PE firm, and zero otherwise. An independent PE firm is one that is not owned by (captive) or associated with (semi-captive) banks.
 SYNDICATEThe number of PE firms participating in a PE syndicate.
Exits 
 TIMEXThe number of months between the date of a buy-out deal and the exit day.
 LNTIMEXThe natural logarithm of TIMEX.
 SALEA categorical variable that takes the value of one if the buy-out exited its original buy-out structure via a trade sale, and zero otherwise.
 IPOA categorical variable that takes the value of one if the buy-out exited its original buy-out structure via flotation on the main board or the second board of the London Stock Exchange, and zero otherwise.
 LIQUIDATIONA categorical variable that takes the value of one if the buy-out failed (and therefore was liquidated), and zero otherwise.
 SMBOA categorical variable that takes the value of one if the buy-out exited its original buy-out structure via an SMBO, and zero otherwise
 EXITSA categorical variable that takes the value of one if the buy-out exited its original buy-out structure via IPO, SALE, SMBO or LIQUIDATION, or zero otherwise.
Table 2.  Summary of Testable Hypotheses
VariablesHo:Impact on Longevity
  1. Note: This table presents the testable hypotheses empirically analysed in the paper. All variables are defined in Table 1.

PE backing  
 –PEH1Shorter
 –TOP10 (PE sample)H2Ambiguous
 –SYNDICATE (PE sample)H3Shorter
 –PECAPITAL (PE sample)H4Shorter
 –INDEPENDENT (PE sample)Control 
Buy-out characteristics  
 –LNSIZEH5Shorter
 –MAIN (IPO sample)Control 
 –SERVICESControl 
Buy-out source  
 –DIVESTMENTH6Shorter
 –RECEIVERSHIP (non–LIQUIDATION sample)H7aLonger
 –RECEIVERSHIP (PE backed LIQUIDATION sample)H7bShorter
 –PRIVATISATIONControl 
Leveraged deals and buy–out type  
 –LBOH8Shorter
 –MBI (non–LIQUIDATION sample)H9aLonger
 –MBI (PE backed LIQUIDATION sample)H9bShorter
Market conditions  
 –LNFTSE (non–LIQUIDATION sample)H10aShorter
 –HOT (non–LIQUIDATION sample)H10bShorter

(i) PE Backing and Characteristics of PE Firms

(a) PE Backing

Cumming and Johan (2010) suggest that VCs add value by providing strategic, marketing, financial and human resource advice. VCs costs are related to incremental overheads, direct costs, and opportunity costs associated with alternative use of capital. The optimal exit point is where the VCs’ projected marginal value added exceeds the VCs’ projected marginal cost from maintaining the investment. In other words, the duration of VC investments changes with shifts in the marginal value and marginal cost functions, which are in turn affected by investee, investor and market characteristics. Since incumbent managers have much lower opportunity costs associated with alternative use of capital, they are expected to have a lower exit propensity than PE firms (Ronstadt, 1986). Furthermore, in the absence of PE backing, managers in pure buy-outs are deprived of the advice and skills provided by PE firms. We, therefore, expect shorter longevity of PE backed than non-PE backed buy-outs:

  • H1: 
    PE backed buy-outs exit sooner than their non-PE backed (pure) counterparts.

The comparison of longevity in PE and non-PE backed subsamples, however, should be treated with caution. Since the greater value added provided by the PE firms normally requires a longer investment duration (Cumming and Johan, 2010), quick exits (e.g., within 12 months) are expected to be associated with less skilled PE firms and incidences where PE firms act against the interests of their portfolio buy-outs. Furthermore, in cases of IPO exits, the shorter longevity in subsamples of pure buy-outs could also be consistent with the view that insiders tend to maximise their private benefits and preserve their interest in the company by taking companies public (Berglof, 1994; and Black and Gilson, 1998). The above scenario is particularly plausible in the UK given the very active AIM with less stringent listing rules facilitating exits of smaller buy-outs. For the above reasons we examine early exits in AIM and the main market separately.

(b) PE Firm Reputation and Association with Investment Banks

Because of their greater experience, more reputable firms are expected to be more effective in implementing the changes required to grow buy-out companies and in certifying their quality. These PE firms are, therefore, more likely to take buy-outs to market sooner. Previous evidence for UK and international samples lends support to the reputation hypothesis (Jelic et al., 2005; and Stromberg, 2008, respectively). Kaplan (1991), however, reports that US LBOs backed by more reputable PE firms are no more likely to exit their buy-out structure than others backed by less reputable PE firms. Kaplan's result is in line with the ‘grandstanding hypothesis’ (Gompers, 1996) suggesting that less experienced VC firms may rush deals in order to establish their reputation and are, therefore, expected to have shorter duration of investments. Given the inconclusive evidence, we test the following hypothesis:

  • H2: 
    For PE backed exits, the PE firms’ reputation is associated with investment duration.

The difference between PE firms that are subsidiaries of investment banks (i.e., captive PE firms) and independent PE firms has been examined in the previous literature. Kaplan (1991), for example, considers the captive US PE firms to be the more reputable. The more established UK PE firms are also, typically, subsidiaries of larger financial organisations (Jelic et al., 2005). The above suggests a possible interaction between the reputation and captivity of PE firms that is consistent with two scenarios. In the first scenario the association with investment banks enhances the reputation of PE firms, leading to increased likelihood of exits and therefore shorter longevity. In the second scenario captive PE firms are facing less pressure to exit within a certain period in order to raise funding due to their links with investment banks (Jelic et al., 2005). In the second scenario, captive PE firms have the lower marginal costs of not investing and hence the duration of investments is expected to be longer. Overall, the evidence regarding the association of PE firms’ reputation and their links with investment banks is not conclusive. We, therefore, control for the importance of PE firms’ association with investment banks.

(c) PE Syndicates

The international evidence regarding VC deals suggests that syndicated deals increase the pool of contracts and enhance the skills required for successful exits and higher IPO valuation (Lerner, 1994; Giot and Schwienbacher, 2007; Cumming, 2006; and Xuan, 2007). Stromberg (2008) finds that LBOs acquired by syndicates of PE firms tend to accelerate exits. The above evidence is consistent with conjecture that syndicates reduce the effort made by any individual member of the syndicate and hence shorten the duration of the investment (Cumming and Johan, 2010). There is also some evidence that PE funding provided by several investors could increase the likelihood of agency conflicts among the syndicate members, thus leading to a quicker exit from the original buy-out structure (Wright et al., 1995; and Stromberg, 2008). The above evidence suggests shorter longevity of deals backed by PE syndicates. We, therefore, conjecture the following:

  • H3: 
    For PE backed buy-outs, longevity will be shorter among syndicated investments.
(d) Availability of PE Capital

Growth in investment funds increases the VC opportunity cost of not investing in new ventures and hence shortens investment duration (Cumming and Johan, 2010). Prior work is consistent with the view that more capital for investment shortens duration (Gompers, 1996; and Cumming and Walz, 2010). For all types of exits, therefore, investment duration is shorter when the supply of funding available from institutional investors and other sources is greater. We, therefore, test the following hypothesis:

  • H4: 
    For PE backed buy-outs, time to exit is shorter when the supply of PE funding is greater.

(ii) Characteristics and Sources of Buy-Outs

(a) Buy-Out Size

Previous evidence regarding UK buy-outs suggests that larger buy-outs exit earlier (Wright et al., 1994; and Stromberg, 2008). Cumming and Johan (2010) attribute this to higher marginal costs of investments (due to greater fixed costs) and lower marginal value added (due to less monitoring). We, therefore, expect shorter longevity of larger buy-outs compared to their smaller counterparts:

  • H5: 
    Longevity is negatively associated with size of the buy-outs.

The UK has one of the leading stock markets and one of the most successful second tier markets. These features significantly reduce the costs of going public and make IPO exits more likely compared with other European countries. Because of the significantly lower value threshold for the listings than is the case in the main market and help provided by specially designated advisors (i.e., Nomads), the AIM is particularly important for smaller buy-outs. Although the argument related to high marginal costs in larger investments still applies, the association of size and longevity may be affected by the above mentioned features of the UK IPO markets. We, therefore, control for the listings on the different LSE's boards.

(b) Industry Classification

Industry classification has been examined in the previous literature on longevity of VC investments. Bayer and Chemmanur (2006), for example, find that firms that are less viable in the face of market competition may need more support from a strategic investor. The support in the product market is particularly important within industries that are less concentrated, such as the service industry.

The UK buy-out market was traditionally dominated by manufacturing firms. During the 1990s this changed and service industries have now become the main source of UK buy-outs. For the buy-outs from service industries we would expect higher PE value added created by their facilitating networks and advice regarding potential strategic investors. Marginal costs related to the opportunity costs of not finding a strategic partner, however, will also be higher for the buy-outs from the service sector. The overall relationship between longevity and industry classification, therefore, is ambiguous. Consequently, we control for industry classification in our sample.

(c) Sources of Buy-Outs

Previous studies report that the source of the buy-out affects buy-out exits and longevity. For example, the UK evidence suggests shorter longevity for buy-outs arising from divestments (Wright et al., 1994). The evidence is consistent with the fact that buy-outs arising from divestments often face various constraints on decision making imposed by parent companies (Gailen and Vetsuypens al., 1989). The adoption of a buy-out structure may, therefore, be an attempt to quickly remove the constraints, leading to early exits (Wright et al., 1995). Thus:

  • H6: 
    Buy-outs originating from divestments will have shorter longevity.

Buy-out deals originating from companies in the receivership process (i.e., distressed companies) are riskier and normally require more time for turnaround and are, therefore, expected to remain longer in their original buy-out structure. Stromberg (2008), for example, reports that deals originating from distressed acquisitions are more likely to end up again in financial distress. Given the expertise required to turn such firms around, PE backing in the transactions is expected to create significant value added. The above buy-outs, however, are also expected to have significant marginal costs. For example, distressed companies are normally burdened by significant conflicts of interest between different insider groups as well as between insiders and PE investors. PE firms are, therefore, more likely to write-off those transactions in cases of high carry costs. We, therefore, conjecture the following:

  • H7a: 
    For non-liquidation exits, longevity is longer for buy-outs originating from distressed companies.
  • H7b: 
    For buy-outs originating from distressed companies, we expect shorter longevity of PE backed liquidations compared with their non-PE backed counterparts.

Due to an extensive UK privatisation programme in the 1980s, our sample includes a considerable number of privatised state-owned companies. These companies are clearly different from an average privately-owned company in our sample, both in terms of size and ownership structure. We, therefore, control for cases where our sample buy-outs originate from privatised state-owned companies.

(iii) Types of Buy-Outs

(a) Highly Leveraged Transactions

The control effect of debt has been well documented in the previous literature. Cumming (2008), for example, reports a positive association between leverage and VCs’ strong control rights. Similarly, Nikolaselainen and Wright (2007) identify the amount of debt financing as an important buy-out corporate governance mechanism. Given the association of leverage with control rights, and the fact that highly leveraged investments are profitable only if the exit is within the planned holding period, we conjecture a negative relationship between leverage and longevity. We, therefore, test the following hypothesis:

  • H8: 
    Highly leveraged buy-outs have shorter longevity.

The measures of leverage used in the previous literature on late stage investments are gearing ratio, debt coverage and percentage of debt in initial transactions.3 We use the percentage of debt in initial transactions as a proxy for the disciplinary effect of debt. The transactions with a high percentage of debt in initial transactions were identified as LBOs in at least one of the data sources used in this study. After careful inspection of the deals identified as LBOs in various databases, we found that not all buy-outs classified as LBOs received PE backing. We were, therefore, able to identify LBOs in both the PE backed subsample and the subsample of pure buy-outs.

(b) MBI

Previous UK evidence suggests that managers undertaking MBIs often find substantial unexpected problems (Wright et al., 1995). Consistent with this evidence is the fact that incumbent managers are better informed and aware of problems related to the buy-out deals. Sometimes this makes them hesitant to proceed with a MBO. MBIs are, therefore, more likely to exit via crisis sales and/or liquidation. The MBIs that avoid liquidation, however, are expected to require longer time to turnaround, and to remain longer in their original buy-in structure. Consistent with this view, Wright et al. (1995) report above average longevity for UK MBIs. Due to high marginal costs associated with ‘problematic’ MBIs, PE firms are expected to cull their investments leading to shorter longevity. We, therefore, hypothesise:

  • H9a: 
    For non-liquidations, MBIs will have greater longevity.
  • H9b: 
    For PE backed liquidations, MBIs will have shorter duration compared to their non-PE backed counterparts.

(iv) Market Conditions

Cumming and Johan (2010) report a negative association between market conditions and duration of VC investments. The finding is consistent with the conjecture that the strong market conditions increase the opportunity cost associated with maintaining prior investments. Furthermore, regardless of PE backing, favourable market conditions provide prospects of higher market valuation. Although stock market conditions are important to all exit routes, they are particularly important for IPO exits. Previous studies show that the relative ‘hotness’ of public markets enhances the probability of IPO exits (Brau et al., 2003; and Poulsen and Stegemoller, 2008). Ball et al. (2008) find evidence that IPOs are selected during periods when market-wide demand for growth capital is high, adverse selection costs of equity issues are low and the value of protecting private information is low. The above evidence is consistent with findings reported in the IPO literature that valuation tends to be easier during IPO waves (Lowry and Schwert, 2002) and when information asymmetry is reduced (Stoughton et al., 2001). The favourable market conditions, therefore, accelerate exits, regardless of PE backing:

  • H10a: 
    For non-liquidations, longevity of buy-outs is negatively associated with strong market conditions in terms of returns on the FTSE All Shares Market Index.
  • H10b: 
    For non-liquidations, longevity of buy-outs is negatively associated with strong market conditions in terms of ‘hot’ IPO periods.

3. DATA AND SUMMARY STATISTICS

  1. Top of page
  2. Abstract
  3. 1. INTRODUCTION
  4. 2. HYPOTHESES FOR DETERMINANTS OF BUY-OUT LONGEVITY
  5. 3. DATA AND SUMMARY STATISTICS
  6. 4. UNIVARIATE ANALYSIS ACROSS PE BACKING, EXIT ROUTES AND VINTAGE DECADES
  7. 5. BUY-OUTS SURVIVAL ANALYSIS
  8. 6. ESTIMATION RESULTS
  9. 7. ROBUSTNESS OF RESULTS AND EXTENSIONS
  10. 8. DISCUSSIONS AND CONCLUSION
  11. REFERENCES

(i) Sources and Sample Construction

Limitations of the alternative databases on PE backed buy-outs have been well-documented in the literature (see Kaplan et al., 2002; Jelic et al., 2005; Ball et al., 2008; and Stromberg, 2008). For example, Stromberg (2008) reports that the Capital IQ database lacks data on enterprise values for 58% of the transactions and that pricing information is unavailable for an even higher percentage of the transactions. Furthermore, several databases have a specific threshold for size of transaction. Since buy-outs may have been relatively small at the time of the buy-out but have subsequently grown to a size where an exit (e.g., IPO) is feasible, the data sources that adopt a large minimum size cut-off may have introduced a biased perspective ( Jelic et al., 2005). Finally, the majority of databases (e.g., Thomson One Banker, Capital IQ and SDC M&A) have more observations in more recent years, which may introduce a sample selection bias. As cited in Ball et al. (2008), over 75% of acquisitions listed in the SDC M&A database during 1984–1998 are from the 1995–1998 period. To overcome these limitations, we combined several databases to create a unique dataset covering buy-outs and PE firm activities during the period January 1966 to December 2004.

We started with the data provided in Centre for Management Buy-out Research's (CMBOR) Quarterly Reviews and supplemented it with the following sources: (i) KPMG MBO Commentaries (1981–99); (ii) Thomson One Banker database (1988–2004); (iii) Barclays Private Equity Deal Maker database; and (iv) Growth Business database.4 From the above sources, we collected data on inception and/or vintage years, source of deals, size of deals, PE backing and industry classification for 1,089 buy-outs. We also monitored the exit status and collected further data on size of deals, PE backing and vendors by surveying the same sources plus various issues of Barclays Private Equity – Exits, KPMG New Issue Statistics, LSE Primary Market Factsheets, BVCA reports, and websites of the UK PE firms.

The data on the number of deals, total amount invested by different PE firms, characteristics of PE firms (e.g., association with banks) and characteristics of PE deals (e.g., syndicated vs. sole) for the period from 1981 to 1998 was collected from various issues of KPMG-Corporate Finance and KPMG-MBO publications. More recent data was collected from the Thomson One Banker database, British Venture Capital Association (BVCA) publications, the Growth Business database, and PE firms’ and The Private Equity International websites.5

We were able to compare our dataset with Capital IQ and CMBOR datasets.6 Our coverage of Capital IQ dataset 1970–2002 is around 47%.7 The coverage of CMBOR dataset is around 9%. It is important to mention, however, that this database is the only buy-out database without limit on size of buy-outs. Consequently, around 75% of CMBOR buy-outs have a value of less than £10m. This compares to only 17% of our sample buy-outs with a value below the £10m threshold. If we consider only buy-outs with a market value higher than £10m in respective datasets, our coverage is around 41%.

Table 3 presents a more detailed comparison between our data set and aggregate annual data for all buy-outs from the CMBOR dataset (12,500 buy-outs) during the period of January 1980 to December 2004.8 We compare the datasets across types (MBI), private equity backing (PE), sources (Divestment), and industry classification (Services) of buy-out transactions.9 The reported results of the Mann-Whitney test (for difference in median proportions) and the Kolmogorov-Smirnov test (for equality of distributions) suggest a lack of statistically significant differences between the datasets in different decades and 5-year periods. The only statistically significant differences are present in terms of distribution of annual proportions of buy-outs receiving PE backing (at the 5% level) and distribution of annual proportions of buy-outs classified as service industries (at the 1% level). Overall, our dataset is similar in terms of main buy-out types, sources, industry classification, and PE backing to the CMBOR dataset and thus is a good representation of the UK buy-outs.

Table 3.  Sample Representativeness: Comparison with the CMBOR Dataset
 MBIPEDivestmentServices
  1. Notes: This table presents the comparison of our data set with the CMBOR dataset during the period of January 1980- December 2004. Panel A presents proportions of MBI, PE backed buy-outs, buy-outs originated from Divestments, and buy-outs from Service industries in the CMBOR dataset and our sample respectively. The results in Panel B are based on median proportions across decades (1980s, 1990s, first half of 2000). The results in Panel C are based on median 5-year proportions during 1980–2004. The results in Panel D are based on median annual proportions during 1980–2004. The Mann-Whitney (M-W) test is for difference in medians. The Kolmogorov-Smirnov (K-S) test is for equality of distributions across types (MBI), private equity backing (PE), sources (Divestment), and industry classification (Services) of buy-out transactions. ***, ** and * indicate levels of statistical significance at 1, 5 and 10% respectively.

Panel ACMBORSampleCMBORSampleCMBORSampleCMBORSample
Proportions (%)2019425853465345
Panel BComparison by decades
M-W1.091−1.5280.2180.000
K-S0.667 0.6670.3330.333
Panel CComparison by 5-year periods
M-W0.400 0.4000.4000.600
K-S0.522−0.5220.5220.447
Panel DAnnual comparison
M-W1.108−1.1511.1461.480
K-S0.308 0.423**0.2530.577***

The comparison of our data coverage with the data used in previous studies is presented in Table 4. Our search resulted in a comprehensive new dataset that compares favourably. For example, our sample includes buy-outs from a much longer period compared to previous studies of UK buy-outs. We also examine large and small, PE backed and non-PE backed, and buy-outs of all types, sources and exit strategies. Our sample coverage is between 84% and 100% of the samples used in previous UK studies.

Table 4.  Sample Coverage: Comparison with Previous Studies on Buy-Out Longevity
AuthorsCountryPeriodSampleBuy-Out TypeThis Study –Sample Coverage
  1. Notes: This table presents a comparison of our sample with relevant samples used in previous UK and international studies on longevity of buy-outs. Our study's coverage, during the sample period of comparable studies, is calculated as: coverage = (number of our sample buy-outs/number of sample buy-outs in a comparative study) *100.

Kaplan (1991)US1979–86183 exits and non exitsLarge public to private PE backed LBOs.n.a.
Wright et al. (1993)UK, France (F), Sweden (S), Holland (H)UK(1981–91); S(1986–91); H(1981–91); F (n.a.)UK -(533 exits; sub-sample 158 during 1983–86); H(346); S (33); F(n.a.)Large and small PE backed MBOsOur sample includes 450 UK buy-outs that exited during 1981–91, resulting in a coverage of 84%.
Wright et al. (1994)UK1981–90UK – 434 exitsLarge and small MBOsOur sample includes 410 UK buy-outs that exited during 1981–90, resulting in a coverage of 94%.
Wright et al. (1995)UK1983–86UK – 158 exits and non-exitsPredominantly large PE backed MBOs and MBIsOur sample includes 151 UK buy-outs during 1983–86, resulting in a coverage of 96%
Jelic et al. (2005)UK1964–97UK – 167 IPO exitsPE backed and non PE backed MBOsOur sample includes 402 UK IPO exits during 1964–97, resulting in a coverage of 100%.
Nikoskelainen and Wright (2007)UK1995–2004UK – 321 exits onlyLarge and small PE-backed LBOsOur sample includes 290 UK buy-outs that exited during 1995–2004, resulting in a coverage of 90%.
Stromberg (2008)Worldwide1970–2002(05)UK – 2,229, including buy-outs without data on deal value (1,488 exits).Predominantly PE backed LBOsOur sample includes 1,059 UK buy-outs, 804 of which exited buy-out structure during 1970–2002. Our sample coverage is 47% and 54% respectively.
This studyUK1966–2004UK – 1,089; 866 exits (including liquidations) and 223 non-exits.Exited and non-exited, small and large, PE backed and non-PE backed MBO, MBI, and LBOs 

(ii) Summary Statistics

(a) Buy-Out Characteristics, Exit Status, and Longevity

Table 5 presents the development of buy-outs during the sample period. The median size of our sample buy-outs is £28.5 million, with a median exit value of £57.5 million. The average (median) time to exit original buy-out structures is 36 months. While the size of buy-outs was steadily increasing the longevity of buy-outs was steadily decreasing, reaching only 24 months (median) in the 2000s. Overall, 58% of our sample buy-outs received PE backing. The percentage of PE backed deals increased steadily, reaching 80% in the 2000s.

Table 5.  Sample Buy-Outs in Different Decades
Decade1960s1970s1980s1990s2000sAll
  1. Notes: This table presents the characteristics of the sample buy-outs in different decades during the period of January 1966-December 2004. All variables are defined in Table 1. SIZE, TIMEX, MBO, MBI, DIVESTMENT, PRIVATISATION, RECEIVERSHIP, OTHER, LBO, SERVICES, MANUFACTURING, PE, EXITS, NON-EXITS and different exit routes (IPO, SALE, SMBO and LIQUIDATION) are reported as percentages of sample buy-outs in different decades. Median SIZE is estimated for all sample buy-outs. MAIN and AIM are reported as a percentage of all IPOs in different decades, listed in the respective LSE boards. Non-exits are buy-outs that have not changed their status from the original buy-out structure. n.a. = not available.

Size and time to exit      
 Median SIZE (£m)n.a.7.719.137.95428.5
 TIMEX (mean/median)212/22399/10049/3941/3627/2445/36
Industry classification (%)      
 SERVICES02334517145
 MANUFACTURING1007766492955
Type of buy-out (%)      
 MBO/LBO75/2592/085/877/2680/5181/19
 MBI25815232019
Source of buy-out (%)      
 DIVESTMENT256249463746
 RECEIVERSHIP001403
 PRIVATISATION003303
 OTHER753847436348
PE backing (%)      
 PE251547688058
Exit status and method (%)      
 NON EXITS001926020
 EXITS:100100817410080
 SALE0020242421
 IPO10010054394747
   –MAIN10010058685260
   –AIM0042324840
 SMBO00311349
 LIQUIDATION006103

Overall, the sample is dominated by MBOs (81%). Our sample also reflects the emergence of LBOs as one of the most important developments in the buy-out market. For example, LBOs represent one quarter of all transactions in the 1990s and more than half of transactions in the first half of the 2000s. Divestments are the dominant source of buy-out transactions, followed by family and other privately-owned companies (46% and 41% of the sample buy-outs respectively). Although our overall sample is dominated by buy-outs from manufacturing, the early 1990s saw a major shift towards buy-outs from service industries (particularly those in business services, leisure and retail).10

(b) Full vs. Partial Exits

Given our definitions of variables, presented in Table 1, liquidations are full exits whilst SMBOs are partial exits. When it comes to IPOs and trade sales, the classification is less conclusive and depends on the definition of partial exits. Cumming and MacIntosh (2003) define full exits as IPOs with the sale of all VC shares within one year. In the case of trade sales, full exits are sales of the entire firm to a third party where a cash (rather than share) consideration is received. We were able to cross-check the respective subsamples with data from Thomson One Banker and the LSE Primary Market Factsheets. In total, only 3.7% of our IPO sample had a second exit within 12 months.11 For trade sales, only two of our sample sale exits were not full exits. These characteristics are similar to those of Canadian VCs reported in Cumming and MacIntosh (2003).

(c) Sample Buy-Out Exit Routes

The most popular exit routes are IPOs (47%), followed by trade sales (21%), SMBOs (9%) and liquidations (3%). The IPO exits from the main market and the AIM are 28% and 19%, respectively. SMBO exits exhibited steady growth, reaching 29% of all exits in the 2000s. Our results suggest a higher percentage of IPO exits than in previous studies on buy-outs. For example, the reported percentages for IPO exits from buy-outs were 16% in Nikoskelainen and Wright (2007), 10% in Wright et al. (1995) and 11% in Stromberg (2008). The direct comparison of our results with the results from previous studies, however, is difficult and could be misleading due to differences in the sample coverage, both in terms of sample periods and types of deals, reported in Table 4. For example, Stromberg (2008) reports that the CapitalIQ database, used in his study, tends to under-report deals from the 1970s and 1980s and deals without PE backing. Given that one of the major hot UK IPO market periods was recorded during the eighties, and the fact that the number of IPO exits for LBOs dropped worldwide in the late 1990s and early 2000s, this could have resulted in some under-reporting of IPO exits. Most importantly, we were able to collect additional data on IPO exits in the AIM directly from IPO prospectuses and hence significantly increase our coverage of smaller IPO exits, compared to other studies.12 For example, considering only sample IPOs from the main LSE board the percentage of IPO exits is 28%.13

The reported failure rate of 3% for our sample buy-outs is lower than the 6% reported in Stromberg (2008). It is, however, worth noting that we adopt a narrower definition of failures, considering only the buy-outs that were reported to have ceased trading (i.e., liquidation).14 Our definition is consistent with what is normally described in the PE industry as a write-off (i.e., the write-down of a portfolio company's value to zero). We also compare the failure rate reported for our sample with the failure rate of all UK private companies available in the Fame database.15 The percentage of all UK private companies that resulted in liquidation was 3%.16 Our sample, therefore, exhibits similar failure rates to other UK private companies. The above results are also consistent with Stromberg (2008) who reports relatively modest bankruptcy rates of LBOs, similar to those reported for corporate bond issuers.

(d) PE Firms’ Reputation and Backing

Regarding the reputation of PE firms, there is no consensus as to which proxy is the best. Capital under management and/or total capital available for investment, as used in some of the previous studies (Gompers and Lerner, 1999), is better suited for limited partnerships than for captive firms since in the latter the resources and talent could also be available in parent banks. Age has also been used in some previous studies. For example, Stromberg (2008) reports that more experienced PE partnerships tend to exit earlier, are more likely to take LBOs public and less likely to end up in liquidation or financial distress. One of the difficulties in using age as a criterion in our sample is the presence of many captive PE firms. Until recently, many of these were not legally separated from their parent banks. The question then is whether to use the age of the parent banks or the age of PE firms as the number of years since they became a separate legal entity. Using the former is likely to over-estimate the experience of the captive firms relative to independent firms.17 In the case of the latter, we would ignore all the resources that the banks had accumulated over many years and which are at the disposal of the PE subsidiaries.

In this study, we establish the reputation of the PE firms based on the number of deals where the firms acted as deal leaders and the amount of equity investments made during the sample period. The number of deals is consistent with the measure used for the reputation of PE firms (Jelic et al., 2005) and VC firms (Lee and Wahal, 2004) while total equity funding is consistent with market share often used as a measure of underwriting firm reputation (Megginson and Weiss, 1991). The number of deals criterion (on its own) would discriminate against some well known ‘boutique’ firms specialising in certain types of deals and/or industries. On the other hand, the disadvantage of total funding (on its own) is that it allows a few very large deals to disproportionately influence classification. We, therefore, determine overall ranking as an equally weighted average of the two criteria:

  • image(1)

We collect data on deal leaders and PE investments in more than 4,000 UK buy-outs. First, we establish ranking of the firms according to both number of deal leaderships and total amount of equity invested during our sample period. In cases of syndicated investments, we use reputation of the lead PE firm.18 Based on the reputation score we establish reputation for 60 PE firms involved in the backing of our sample buy-outs. The highest ranked PE firms (i.e., with the lowest reputation score) according to our criteria were 3i, Cinven, and Bridgepoint (formerly known as Nat West Private Equity). Between them they were involved in the backing of 44% of all PE backed sample buy-outs (Table 6– Panel A). Compared to previous UK studies, our classification is similar to the ranking provided in Jelic et al. (2005). For example, two out of our top three (3i and Cinven) and five out of our top ten (3i, Cinven, Bridgepoint, Charter House, Schroeder) are the same as in Jelic et al. (2005). To check the robustness of our classification we compare our list with the most recent ratings of PE firms provided in The Private Equity International.19 The comparison shows that our seven most reputable PE firms also appear among the top 50 world leading PE firms listed by The Private Equity International.20

Table 6.  Characteristics of Sample PE-backing
Decade1960s1970s1980s1990s2000sAll
  1. Notes: This table presents a summary of characteristics of PE backing by exit status and different exit routes for the sample buy-outs in different decades during the period of January 1966 to December 2004. For IPO exits, we also report listings in the main (MAIN) and second (AIM) LSE's boards. The sample buy-outs are stratified by year of original buy-out deals. Panel A presents success rates and characteristics of PE backing and firms. SUCCESSFUL EXITS presents the percentage of sample PE-backed buy-outs exited via IPO, sale, or SMBO. REPUTABLE presents the percentage of all PE backed sample buy-outs that received backing from the more reputable PE firms (TOP3/TOP10). INDEPENDENT presents the percentage of all PE backed sample buy-outs that received backing from independent PE firms. SYNDICATED presents the percentage of all PE backed sample buy-outs that received backing from PE syndicates and the average number of PE firms in the syndicates. PE ROUNDS presents the average number of financing rounds and average investment per PE backed deal. Panel B presents the percentage of PE backed sample buy-outs across their exit status and exit routes. Panel C presents the subsample of buy-outs backed by the more reputable (TOP3) PE firms by exit status and exit routes. Panel D presents the subsample of syndicated deals by exit status and exit routes. Panel E presents the subsample of INDEPENDENT (i.e., PE firms without affiliations with a bank or corporation) firms by exit status and exit routes. n.a. = data not available.

Panel A: PE BACKED = 100%      
 SUCCESSFUL EXITS100100807910081
 REPUTABLE (TOP3/10;%)0/050/5056/7642/5826/3444/60
 INDEPENDENT (%)0/010060605560
 SYNDICATED (%/N)050/2.037/2.037/2.252/3.038/2.3
 PE ROUNDS (N/£m)n.a.2.5/5.31.5/26.81.7/31.52.1/61.61.8/35.6
Panel B: PE BACKED = 100%      
 IPO10010068444652
   –MAIN10010070775872
   –AIM0030234228
 SALE008222518
 SMBO002123511
 LIQUIDATION002202
 NON–EXITS002020019
Panel C: TOP 3 = 100%      
 IPO010068403650
   –MAIN0100518710071
   –AIM004913029
 SALE008263621
 SMBO002132810
 LIQUIDATION000101
 NON–EXITS002122021
 SYNDICATED (%/N)0/00/033/1.986/2.333/2.951/2.2
Panel D: SYNDICATED (%/N) = 100%      
 –IPO0/0100/2.042/2.325/2.441/2.7532/2.4
   –MAIN0/0100/2.072/2.193/2.467/3.281/2.4
   –AIM0/00/028/2.07/3.033/3.019/2.5
 –SALE0/00/012/2.022/1.814/318/2.0
 –SMBO0/00/05/2.024/3.245/3.423/3.1
 –LIQUIDATION0/00/02/2.01/2.00/01/2.0
 –NON–EXITS0/00/040/1.930/1.70/027/1.7
Panel E: INDEPENDENT FIRMS = 100%      
 –IPO010065364646
   –MAIN010067825873
   –AIM0033184227
 –SALE009263120
 –SMBO003123111
 –LIQUIDATION000100
 –NON–EXITS002426023

The changes in PE firm success rates, their reputation, association with investments banks and characteristics of financing (syndication and financing rounds) are presented in Table 6 (Panel A). Four fifths (81%) of sample buy-outs that received PE backing made a successful exit via an IPO, sale or SMBO. The majority (60%) of buy-outs received backing from an independent PE firm.21 The percentage of syndicated deals increased during the sample period and reached 52% in the 2000s. The mean and median number of financing rounds and average investment per deal are £1.8 and £35.4 million respectively. We also document an increasing trend in both the number of financing rounds and the average investment per deal during the 1990s and 2000s.

Concerning exit routes, more than half of PE backed buy-outs exit via IPOs, followed by sales (28%) and SMBOs (11%) (Table 6– Panel B). Although they increased their participation in the AIM during the last decade, PE firms tended to prefer larger buy-outs exiting via IPO in the main market (72% of all PE backed IPO exits). As expected, the percentage of PE backed buy-outs ending in liquidation (2%) is lower than the average for the whole sample.22 The top three PE firms exhibit similar characteristics to an ‘average’ sample PE firm in terms of exit routes (Table 6– Panel C). The notable difference, however, is that they tend to participate in more syndicated deals than the sample average. This was particularly the case in the 1990s when 86% of all deals by Top3 PE firms were syndicated. The most reputable PE firms also exhibit a lower percentage of buy-outs ending in liquidation (1%) than their less reputable counterparts.

Among syndicated deals, SMBOs tend to be an important exit route (23%) coming second after IPOs and before sale exits (Table 6– Panel D). SMBOs also have the highest average number of PE firms in the syndicates (3.1).23 Notably, the percentage of non-exits (27%) in the subsample of syndicated deals is greater than the sample average. In terms of backing and exits, the sample of independent PE firms exhibits similar characteristics to the captive PE firms. They do, however, have a lower percentage of liquidations than their captive counterparts (Table 6– Panel E).

4. UNIVARIATE ANALYSIS ACROSS PE BACKING, EXIT ROUTES AND VINTAGE DECADES

  1. Top of page
  2. Abstract
  3. 1. INTRODUCTION
  4. 2. HYPOTHESES FOR DETERMINANTS OF BUY-OUT LONGEVITY
  5. 3. DATA AND SUMMARY STATISTICS
  6. 4. UNIVARIATE ANALYSIS ACROSS PE BACKING, EXIT ROUTES AND VINTAGE DECADES
  7. 5. BUY-OUTS SURVIVAL ANALYSIS
  8. 6. ESTIMATION RESULTS
  9. 7. ROBUSTNESS OF RESULTS AND EXTENSIONS
  10. 8. DISCUSSIONS AND CONCLUSION
  11. REFERENCES

(i) Size Across PE Backing and Different Exit Routes

Table 7 presents average size of sample buy-outs by exit status, PE backing and exit routes, in decades of original buy-outs. Overall, exits tend to be larger than non-exit buy-outs, but the difference in size is not statistically significant. PE backed buy-outs tend to be significantly larger than their non-PE counterparts. The difference (in median) size is statistically significant at the 1% level. Among different exit routes, SMBOs tend to be the largest and IPOs the smallest transactions.24 However, the difference in average size between PE backed and non-PE backed deals is (highly) statistically significant only in the IPO sample. This was expected as approximately 40% of our sample IPO exits come from the AIM, with a much lower size threshold. The (unreported) results suggest that IPOs from the main market are by far the largest, whilst the AIM IPOs were the smallest sample buy-outs. Furthermore, the IPOs from the main market are larger than sale, SMBO and liquidations.25 The average size of sample buy-outs significantly increased over time. For example, the mean size of PE backed buy-outs was £97m in the 1980s compared to £192m during the 2000s.26

Table 7.  Buy-Out Size Stratified by PE Backing and Exit Routes
 1960s1970s1980s1990s2000sAll
SIZE (mil£)MeanMedianMeanMedianMeanMedianMeanMedianMeanMedianMeanMedian
  1. Notes: This table presents mean and median SIZE of sample buy–outs, by exit status, PE backing, and different exit routes, in decades of original buy–out deals. All variables are defined in Table 1. Reported percentages [%] represent the proportion of buy–outs in PE backed and non–PE backed subsamples exiting via relevant exit routes. P–values for two sample T (difference in means = 0 vs. difference in means ≠ 0) and two sample Mann–Whitney (difference in medians = 0, vs. difference in medians ≠ 0) tests are presented in brackets.

SAMPLE:n/an/a39892187738188549029
 –NON–EXITS1243539226925
 –EXITSn/an/a39882219238188549731
 –EXITS VS. NON–EXITS(0.2792)(0.0000)(0.0000)(0.0135)(0.1475)(0.4755)
 –PE BACKEDn/an/an/an/a97228434192579831
 –NON– PE BACKEDn/an/a39885285921171367623
 –PE VS. NON– PE BACKED(0.7193)(0.7120)(0.0834)(0.0042)(0.8782)(0.0744)(0.2017)(0.0035)
EXIT ROUTES:            
 IPO39865107020167497514
   –PE BACKED [52%]87147927219559221
   –NON–PE BACKED [39%]39810483243114
   –PE VS. NON–PE BACKED(0.1205)(0.0000)(0.0458)(0.0000)(0.2075)(0.0113)(0.0093)(0.0000)
 SALE10546113451844912648
   –PE BACKED [18%]16446115502195614053
   –NON–PE BACKED [28%]894510041242011044
   –PE VS. NON–PE BACKED(0.1836)(0.7106)(0.6428)(0.2314)(0.2075)(0.0113)(0.3621)(0.1266)
 SMBO7322124992095513855
   –PE BACKED [11%]12635117642466716066
   –NON–PE BACKED [6%]38221185944459140
   –PE VS. NON–PE BACKED(0.3235)(0.3938)(0.9828)(0.7767)(0.4147)(0.2332)(0.2388)(0.0563)
 LIQUIDATION13841622212037
   –PE BACKED [2%]40098622216348
   –NON–PE BACKED [5%]97379737
   –PE VS. NON–PE BACKED(0.0567)(0.2317)(0.4687)(0.8905)
 EXIT ROUTES COMPARISON:            
   IPO VS.SALE(0.2963)(0.0000)(0.0278)(0.0000)(0.3556)(0.0209)(0.0190)(0.0000)
   IPO VS. SMBO(0.9364)(0.1975)(0.0326)(0.0000)(0.7243)(0.0448)(0.0297)(0.0000)
   IPO VS. LIQUIDATION(0.2657)(0.0000)(0.8836)(0.3722)(0.3481)(0.0003)
   SALE VS. SMBO(0.6090)(0.0780)(0.6951)(0.5943)(0.5624)(0.5056)(0.6471)(0.7486)
   SALE VS. LIQUIDATION(0.5244)(0.9298)(0.4327)(0.3009)(0.9134)(0.4296)
   SMBO VS. LIQUIDATION(0.4582)(0.0955)(0.3776)(0.2998)(0.7211)(0.4766)

(ii) Exit Routes Across PE Backing

Although the preference order for different exit routes is the same for PE backed and pure buy-outs, IPOs and SMBOs tend to play a less important role for pure buy-out exits (39% and 6% respectively) (Table 7). About 28% of non-PE backed buy-outs exited via trade sales. Notably, about 5% of non-PE backed buy-outs resulted in failure and/or liquidation (Table 7). The popularity of IPO exits for non-PE backed buy-outs is consistent with the earlier mentioned success of the AIM and the lesser involvement of PE firms in AIM listings.27 The result is also consistent with findings that, when control rights are weaker (e.g., absence of PE firms), managers may prefer IPOs in order to preserve their private benefits (Cumming, 2008).

(iii) Time to Exit Across PE Backing and Different Exit Routes

Table 8 presents average time to exit of sample buy-outs by PE backing and exit routes, in decades of original buy-outs. Overall, median time to exit in the sample is 36 months. As hypothesised (H1), PE backed buy-outs tend to exit sooner than their non-PE counterparts (statistically significant at the 1% level) (Table 8– Panel A). The result is consistent over the last three decades and it is highly statistically significant at the 1% level.

Table 8.  Buy-Out Longevity Stratified by PE Backing and Exit Routes
TIMEX (months to exit)1960s1970s1980s1990s2000sAll
MeanMedianMeanMedianMeanMedianMeanMedianMeanMedianMeanMedian
  1. Notes: This table presents mean and median TIMEX of sample buy–outs’ by PE backing (Panel A) and different exit routes (Panels C and D), in decades of original buy–out deals. EARLY EXITS (Panel B) present the number of exits within 12 months following the original buy–outs. All variables are defined in Table 1. P–values for two sample T (difference in means = 0 vs. difference in means ≠ 0) and two sample Mann–Whitney (difference in medians = 0, vs. difference in medians ≠ 0) tests are presented in brackets.

Panel A: EXITS21223991004939413627244536
 –PE BACKED24624613864332403628244035
 –NON– PE BACKED2012211021125346444621225244
 –PE VS. NON– PE BACKED(0.1792)(0.4572)(0.4771)(0.0154)(0.0186)(0.1517)(0.0759)(0.0849)(0.0874)(0.0000)(0.0010)
Panel B: EARLY EXITS    –   –    30    22    8    60 
 –PE BACKED    –   –    17    12    5    34 
 –NON– PE BACKED    –   –    13    10    3    26 
Panel C: EXIT ROUTES            
IPO212223991004237383426244637
 –PE BACKED24624613864231373229253932
 –NON–PE BACKED2012211021125459444621225747
 –PE VS. NON–PE BACKED(0.1797)(0.4572)(0.4771)(0.0136)(0.0192)(0.1062)(0.1427)(0.1123)(0.1202)(0.0004)(0.0000)
SALE5548383625244036
 –PE BACKED4736383628243636
 –NON–PE BACKED5748413624244536
 –PE VS. NON–PE BACKED(0.4615)(0.5628)(0.3854)(0.3729)(0.3725)(0.4233)(0.0284)(0.1505)
SMBO8672576030365348
 –PE BACKED96108596030365248
 –NON–PE BACKED8024514827305236
 –PE VS. NON–PE BACKED(0.7740)(0.6508)(0.3362)(0.4514)(0.6075)(0.6393)(0.5056)(0.9564)
LIQUIDATION343664843836
 –PE BACKED181864845354
 –NON–PE BACKED35363536
 –PE VS. NON–PE BACKED(0.1734)(0.5623)(0.1388)
Panel D: COMPARISON            
 IPO VS. SALE(0.1336)(0.1368)(0.9559)(0.6916)(0.8262)(0.8773)(0.0179)(0.2900)
 IPO VS. SMBO(0.0030)(0.1515)(0.0000)(0.0000)(0.3304)(0.4718)(0.1109)(0.0143)
 IPO VS. LIQUIDATION(0.1395)(0.3107)(0.0689)(0.1762)(0.3895)(0.7038)
 SALE VS. SMBO(0.0571)(0.4265)(0.0000)(0.0000)(0.1779)(0.1405)(0.0025)(0.0014)
 SALE VS. LIQUIDATION(0.0383)(0.0420)(0.0453)(0.1642)(0.7583)(0.9844)
 SMBO VS. LIQUIDATION(0.0055)(0.1923)(0.6856)(0.4354)(0.0930)(0.0839)

Stromberg (2008) reports that only 2.9% of PE backed deals worldwide exited within 12 months of the original transaction. The total number of early exits (within 12 months) in our sample is 5.1% (60 cases) (Table 8– Panel B). In our PE backed sub-sample, we found 34 early exits representing 5.3% of all sample PE backed buy-outs.28 In our non-PE backed sample, we report 26 early exits, representing 5.7% of all non-PE backed buy-outs. Both in our total sample and our non-PE backed sub-sample, the number of early exits decreased over the last three decades. For example, the number of early exits in our PE backed subsample dropped from 17 in the 1980s to 12 in the 1990s and 5 in the 2000s. This decrease in the number of early exits is consistent with the results for the international sample reported in Stromberg (2008).

We further analyse the occurrence of early exits across exit routes. The significant majority (93%) of all early exits were IPOs. For IPO markets, the majority of the early exits were from the AIM.29 For example, 17% of all AIM and 7% of all main board listings were early exits. Overall, the above results suggest that the occurrence of early exits in the UK market tend to be associated with AIM IPO exits of relatively smaller buy-outs. Our finding contradicts Stromberg (2008) who reports that the incidences of the ‘quick flips’ tend to be more likely in larger transactions.

SMBOs take the longest time to exit (median of 48 months), followed by IPOs (median of 37 months) and sales and liquidations (median 36 months) (Table 8– Panel C). The time to exit decreased for all exit routes, except liquidations that took longer to exit in the 1990s than in the 1980s. The difference between SMBOs’ time to exit and time to exit for other exit routes is statistically significant (at 1% level for sales, 5% for IPOs and 10% level for liquidations) (Table 8– Panel D). When we compare time to exit for the subsample of IPOs from the main market, the difference between IPOs and sale exits was not statistically significant. The difference between longevity of IPOs and SMBOs, however, remained statistically significant at the 1% level.

On average, PE backed buy-outs exit their original buy-out structure earlier than their pure counterparts. The difference in longevity between PE backed and non-PE backed buyouts is, however, statistically significant only in subsamples of IPOs (mean and median) and sales (mean) (Table 8– Panel C).30 Interestingly, the overall results for liquidations suggest that PE backed deals take longer to exit than non-PE backed deals (Table 8– Panel C). This is unexpected since, due to higher marginal costs, PE firms are expected to cull their non-performing investments earlier. An alternative explanation would be that PE firms take on more complex cases originating, for example, from large privatisations and/or receiverships. We further examine the origins of our PE backed liquidations and find that the majority do indeed originate from more complex buy-outs in receivership which normally take longer to restructure and harvest.31 Our findings are consistent with Stromberg (2008) who reports 100% higher insolvency rates for LBOs involving acquisitions of distressed companies compared to their counterparts.

5. BUY-OUTS SURVIVAL ANALYSIS

  1. Top of page
  2. Abstract
  3. 1. INTRODUCTION
  4. 2. HYPOTHESES FOR DETERMINANTS OF BUY-OUT LONGEVITY
  5. 3. DATA AND SUMMARY STATISTICS
  6. 4. UNIVARIATE ANALYSIS ACROSS PE BACKING, EXIT ROUTES AND VINTAGE DECADES
  7. 5. BUY-OUTS SURVIVAL ANALYSIS
  8. 6. ESTIMATION RESULTS
  9. 7. ROBUSTNESS OF RESULTS AND EXTENSIONS
  10. 8. DISCUSSIONS AND CONCLUSION
  11. REFERENCES

(i) Choice of a Survival Model

The starting point in our survival analysis is the choice between survival models with constant hazard rates (non-parametric and semi-parametric) and models that allow hazard to change over time (parametric). We therefore test for validity of the assumption of a constant hazard using various graphical methods.32 The unreported results suggest that the parametric models are more appropriate for our sample.

In contrast to semi-parametric models (e.g., Cox), the parametric models require choice of a specific underlying distribution (e.g., Gamma, Weibull, Lognormal, etc.). We start with the Gamma distribution which is known to be the most flexible. We therefore implement the following Generalised Gamma model for all sample exits, IPOs, sales, SMBOs and liquidations:33

  • image(2)
  • image(3)

Where S(t) is the probability of survival past time t; f(t) is the distributional form of the error term; γ= |k|−2; z = sign (k) {ln(t) –μ}/δ; u =γ exp(|k|z); ϕ(z) is the standard normal cumulative distribution function; and I(r,u) is the incomplete Gamma function. The suitability of Gamma distribution will be based on tests for k = 0, and k = 1. The test values are associated with the significance of ancillary parameter Kappa in the survival models. When the tests suggest that the Gamma distribution may not be suitable (i.e., insignificant Kappa parameter), the choice among alternative parametric density distributions will be made based on the Akaike (1974) information criterion (AIC).34

The survival models also differ in the assumptions regarding the association between the covariates and the survival (hazard). Our models are implemented in accelerated failure time (AFT) form by parameterising μj= xjβ, where xi is a vector of covariates and β is a vector of regression coefficients.35 In our AFT model, the natural logarithm of the time to exit (i.e., survival time) is, therefore, presented as a linear function of the following covariates:

  • image(4)

Where, zi is the error term with Generalised Gamma density and the covariates are as defined in Table 1. A positive and statistically significant coefficient implies that an increase in the corresponding variable leads to an increase in the expected waiting time for an exit. A negative sign would, therefore, imply a significantly faster exit from the original buy-out structure. We fit the above Generalised Gamma AFT models using total, non-PE backed and PE backed samples. The models are fitted for the following dependent variables: LNTIMEX − EXITS, LNTIMEX − IPO, LNTIMEX − SALES, LNTIMEX − SMBO, and LNTIMEX − LIQUIDATION. The model specifications are identical except in the model for LNTIMEX−IPO, where we add a control variable for listings at the main LSE board (MAIN).

For the sub-sample of PE backed buy-outs we estimate the following Generalised Gamma AFT model that includes characteristics of PE firms and backing:

  • image(5)

All the above AFT models accommodate the fact that some of the buy-outs exit and some still remain in their original buy-out structure. Buy-outs that have not exited contribute to the likelihood function via their survival, whilst buy-outs that exited contribute via the density function. The exits are defined as either change from the original buy-out structure into a public company (i.e., exit via IPO), new private or public company (i.e., exit via sale), new buy-out structure (i.e., exit via SMBO) or liquidation (i.e., exit via liquidation). Consequently, the censored observations in the model with LNTIMEX – EXITS as a dependent variable are all non-exits; the censored observations in the LNTIMEX − IPO model are all non-exits and non-IPO exits; the censored observations in the LNTIMEX – SALES model are all non-exits and non-sale exits; the censored observations in the LNTIMEX – SMBO model are all non-exits and non-SMBO exits; and finally, the censored observations in the LNTIMEX − LIQUIDATION model are all non-exits and non-liquidation exits.

Table 9 provides a correlation matrix across the variables used in our regression models specified in equations 4 (Panel A) and 5 (Panel B). In Panel A, longevity is longer for sample buy-outs originating from privatisations, receiverships, and MBIs (the correlation coefficients for receivership and MBI are statistically significant at 1% and 5%, respectively). All other variables are negatively correlated with longevity. The correlation coefficient is statistically significant at the 5% level or better for PE backing, services, divestments, LBOs, and hot market periods. The results suggesting shorter longevity for PE backed buy-outs and positive association of buy-out size and PE backing are consistent with the results of our univariate analysis presented in Tables 7 and 8. In Panel B, investment duration is longer for more reputable PE firms and deals originating from privatisations, receiverships, and MBIs (correlation coefficients for both receivership and MBI are statistically significant at the 1% level). All other variables are negatively correlated with duration of investments. Overall, the results presented in Table 9 are consistent with our hypotheses. The statistics also provide guidance for potential collinearity problems in the survival models presented in the next section.

Table 9.  Correlation Matrix
 (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)   
  1. Notes: This table provides correlation coefficients across the variables used in our regression models specified in equations 4 (Panel A): LNTIMEXi=αo +β1 LNSIZEi2SERVICESi3MBIi4LBOi5DIVESTMENTi6PRIVATISATIONi7RECEIVERSHIPi8PEi9 LNFTSEi10 HOTi+ zi; and 5 (Panel B): LNTIMEXi=αo +β1 LNSIZEi2SERVICESi3MBIi4LBOi5DIVESTMENTi6PRIVATISATIONi7RECEIVERSHIPi8TOP10i9INDEPENDENTi10SYNDICATEi11LNPECAPITALi12 LNFTSEi13 HOTi+ zi. All variables are defined in Table 1. In Panel A, correlations greater than 0.0819, 0.0611 and 0.0593 in absolute value are significant at 1, 5 and 10% levels respectively. In Panel B, correlations greater than 0.1059, 0.0848 and 0.0750 in absolute value are significant at 1, 5 and 10% levels respectively.

Panel A:              
 1.LNTIMEX 1.0000             
 2.PE−0.1056 1.0000            
 3.LNSIZE−0.0114 0.1112 1.0000           
 4.SERVICES−0.1017 0.1865 0.0454 1.0000          
 5.PRIVATISATION 0.0206−0.0684 0.0666 0.0577 1.0000         
 6.DIVESTMENT−0.0819 0.1244 0.0268 0.0511 0.0741 1.0000        
 7.RECEIVERSHIP 0.1053−0.0159−0.0192 0.0295 0.0420 0.1067 1.0000       
 8.MBI 0.0634 0.1187 0.0432 0.0528−0.0822−0.0640−0.0185 1.0000      
 9.LBO−0.0940 0.2691 0.1220 0.1707 0.0274 0.1691 0.0374 0.0255 1.0000     
10.LNFTSE−0.0355 0.0635−0.1162−0.1164−0.0355 0.0861−0.0236−0.0535−0.10131.0000    
11.HOT−0.0887 0.0082 0.0125 0.0398 0.0611−0.0593 0.0113 0.0357 0.04380.02531.0000   
 (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)
Panel B:              
 1.LNTIMEX 1.0000             
 2.TOP 10 0.0359 1.0000            
 3.INDEPENDENT−0.0102 0.2572 1.0000           
 4.SYNDICATE−0.0611 0.0801 0.0200 1.0000          
 5.LNPECAPITAL−0.6622 0.0056−0.0013−0.0163 1.0000         
 6.LNSIZE−0.0513−0.0921−0.0179 0.2506−0.0516 1.0000        
 7.SERVICES−0.0348−0.1007 0.1086 0.0748−0.0848 0.1001 1.0000       
 8.PRIVATISATION 0.0402−0.0066 0.0753−0.0337−0.0078 0.0212 0.0860 1.0000      
 9.DIVESTMENT−0.1377−0.0505 0.1272 0.0561−0.0197 0.1089 0.0526 0.0801 1.0000     
10.RECEIVERSHIP 0.1300−0.0012 0.0236−0.0538−0.0150−0.0252 0.0252 0.0238 0.1308 1.0000    
11.MBI 0.1059−0.0663 0.0395 0.0055 0.0174 0.0080 0.0271−0.0750−0.0948 0.0149 1.0000   
12.LBO−0.1308−0.1482 0.0434 0.0424−0.1063 0.1396 0.1412 0.0406 0.1277−0.0291−0.0165 1.0000  
13.LNFTSE−0.1147 0.1407−0.0231−0.0366 0.1357−0.1103−0.1187 0.0013 0.0775−0.0044−0.0661−0.1050 1.0000 
14.HOT−0.0557 0.0144 0.0176−0.0121 0.0540−0.0162 0.0191 0.0632−0.0253 0.0138 0.0646 0.0382 0.1144 1.0000

(ii) Sample Selection Bias and Endogeneity

Potential problems with endogeneity and sample selection are well documented in the literature (Cumming and MacIntosh, 2001; Cochrane, 2005; and Jelic et al., 2005). Cochrane (2005), for example, highlights the importance of addressing potential selection problems in the sample consisting of exited investments only. Cumming and MacIntosh (2001) and Jelic et al. (2005) address the potential endogeneity problem due to the possibility that PE firms do not randomly select buy-outs they are backing. Given that our sample consists of both exited and non-exited buy-outs, our sample is not affected by potential selection problems. We therefore only examine the robustness of our result to the potential endogeneity associated with PE backing by conducting a Hausman test.36 The result of the Hausman test suggests that we cannot reject the hypothesis that the difference in coefficients is not systematic across PE backing. The endogeneity problem described above, therefore, is not statistically significant in our sample.37

6. ESTIMATION RESULTS

  1. Top of page
  2. Abstract
  3. 1. INTRODUCTION
  4. 2. HYPOTHESES FOR DETERMINANTS OF BUY-OUT LONGEVITY
  5. 3. DATA AND SUMMARY STATISTICS
  6. 4. UNIVARIATE ANALYSIS ACROSS PE BACKING, EXIT ROUTES AND VINTAGE DECADES
  7. 5. BUY-OUTS SURVIVAL ANALYSIS
  8. 6. ESTIMATION RESULTS
  9. 7. ROBUSTNESS OF RESULTS AND EXTENSIONS
  10. 8. DISCUSSIONS AND CONCLUSION
  11. REFERENCES

(i) Regression Analysis of Longevity

Table 10 presents our five survival models for LNTIMEX-EXITS (model 1 – Panel A), LNTIMEX-IPO (model 2– Panel B), LNTIMEX-SALE (model 3 - Panel C), LNTIMEX-SMBO (model 4 – Panel D) and LNTIMEX-LIQUIDATION (model 5 – Panel E). For each of the five models, we present results for subsamples of PE backed buy-outs (denoted with subscript ‘a’), and non-PE backed buy-outs (denoted with subscript ‘b’), as defined in equations (4) and (5).

Table 10.  Estimation Results of the Survival Models
 Panel A: LNTIMEX –EXITSPanel B: LNTIMEX – IPO
Ho1: SAMPLE1a:PE Backed1b: Non-PE2: SAMPLE2a:PE Backed2b:Non-PE
PE1(−)−0.0238*  −0.0378  
TOP102(?) 0.0045  0.0095 
SYNDICATE3(−) 0.0092  −0.0199* 
LNPECAPITAL4(−) −0.2265***  −0.2197*** 
INDEPENDENTCONTROL 0.0018  0.0118 
LNSIZE5(−)−0.0211***−0.0169***−0.0286***−0.0251***−0.0189***−0.1192**
MAINCONTROL   0.01260.02480.4436***
SERVICESCONTROL0.0056−0.0011−0.00730.0266−0.0369*0.1009
 
DIVESTMENT6(−)−0.0257**−0.0102−0.0395−0.0124−0.00200.2523**
RECEIVERSHIP7a&b (+;−)0.00980.04380.01190.0280−0.0934−0.0081
PRIVATISATIONCONTROL−0.0381−0.0033−0.0414−0.0988−0.0217-
 
LBO8(−)−0.0486***−0.0486***−0.0694−0.0381*−0.03790.0627
MBI9a&b (+;−)−0.01400.0153−0.0188−0.00950.0329−0.2053
 
LNFTSE10a(−)−0.1845***0.0288−0.1837**−0.1574**−0.13860.2484
HOTP10b(−)−0.0279**−0.0360**−0.0184−0.0054−0.0376*0.2500**
 
CONSTANT 1.63931.42451.68041.60151.60570.5636
Log-likelihood 37.3405158.5510−8.5323−21.671578.7877−54.6381
LR (Prob>chi)2 64.57***221.99***21.40**42.49***132.44***22.83***
Kappa 1.7067***0.5778***1.9597***2.0245***1.3077***-
Sigma -----0.48321***
P       
Pseudo R2 (%) 86.4770.0055.6349.5084.0517.28
Time at risk 2,287.30950.29716.051,272.35428.64266.62
N 1,089 (866)636 (516)453 (350)1,089 (510)636 (333)453 (177)
 Panel C: LNTIMEX-SALEPanel D: LNTIMEX-SMBOPanel E: LNTIMEX – LIQUIDATION
Ho3: SAMPLE3a:PE Backed3b:Non-PE4:SAMPLE4a:PE Backed4b:Non-PE5:SAMPLE5a:PE Backed5b:Non-PE
  1. Notes: LNTIMEX is a dependent variable in all models. Estimates for all models are based on our total sample and the subsamples of PE backed and non-PE backed buy-outs. The column titled Ho presents a number of hypotheses associated with relevant explanatory variables, with expected signs. a & b indicate hypotheses for different subsamples. A failure to survive is specified as follows: (i) EXITS (IPO, SALE, SMBO and LIQUIDATION), in Panel A; (ii) IPOs, in Panel B; (iii) SALEs, in Panel C; (iv) SMBOs, in Panel D; (v) LIQUIDATIONs, in Panel E. Consequently, censored observations in LNTIMEX – EXITS models (Panel A) are all non-exits; in LNTIMEX – IPO models (Panel B), all non-exits and non-IPO exits; in LNTIMEX – SALE models (Panel C), all non-exits and non-SALE exits; in LNTIMEX – SMBO models (Panel D), all non-exits and non-SMBO exits; in LNTIMEX – LIQUIDATION models (Panel E), all non-exits and non-LIQUIDATION exits. N indicates both the total and number of non-censored observations (in brackets) in the respective panels. All dependent and explanatory variables are defined in Table 1. The underlying distributions for the models were selected according to the Akaike (1974) criteria and values of Kappa. A Generalised Gamma density distribution was preferred in all models with reported Kappa values. The Weibull density distribution was selected in models with reported values for ancillary parameter P. Lognormal density distribution was selected in models with reported ancillary parameter Sigma. All models fit the accelerated failure time (AFT) metric with LNTIMEX as analysis time. Pseudo R2 were estimated as R2= 1 –Lu/Lo; where Lu corresponds to the last log-likelihood number before the convergence and Lo corresponds to the first log-likelihood number at the start of the iteration. In Panel B, PRIVATISATION was dropped because of collinearity. ***, ** and * indicate levels of statistical significance at 1, 5 and 10% respectively.

PE1(−)0.1061***  −0.0178  0.3144***  
TOP102(?) −0.0035  0.0374  - 
SYNDICATE3(−) 0.0691***  −0.0183***  - 
LNPECAPITAL4(−) −0.2317***  −0.2030***  −0.1962 
INDEPENDENTCONTROL 0.0202  0.0472  - 
LNSIZE5(−)−0.0750***−0.0459***−0.0862***−0.0470***−0.0292***−0.0649***−0.0462*−0.0284−0.0499*
MAINCONTROL         
SERVICESCONTROL0.0143−0.0083−0.02060.00510.0215−0.0985−0.05500.0328−0.0603
DIVESTMENT6(−)0.0550*0.02910.04100.0331−0.01840.1837*0.0012-−0.0149
RECEIVERSHIP7a&b (+;−)0.08970.11390.13821.73161.56802.1619−0.3993***−0.5823**−0.2319
PRIVATISATIONCONTROL−0.1480**−0.0568−0.1375*0.0305−0.06282.2453−0.2473-2.0635
 
LBO8(−)−0.0851***−0.0282−0.0680−0.0868***−0.0579**2.17870.0081−0.09050.0405
MBI9a&b (+;−)0.04910.05620.1206*−0.0119−0.0098−0.0787−0.2268**-−0.2515**
LN FTSE10a(−)0.17830.3321**−0.17970.09130.2357*−0.1520−0.1849−0.20900.0201
HOTP10b(−)−0.0132−0.0270−0.00830.0909**0.05560.11680.1694-0.2164
CONSTANT 1.72801.35411.91701.69421.49741.92382.14592.31082.0020
Log-likelihood −158.4032−41.6179−34.6688−82.2387−12.5886−30.1726−33.7885−5.1887−23.0376
LR(Prob>chi)2 105.06***106.99***60.67***67.86***46.21***27.56***41.71***21.62***20.10**
Kappa 0.8981***-0.8814***1.0076***-----
Sigma ---------
P  7.8743***  10.8542***6.8469***6.9102***9.5251***6.5647***
Pseudo R2 (%) 24.9056.2446.6729.2164.7331.3538.1667.562.08
Time at risk 2,287.30950.2933716.052,287.30950.29716.052,287.301,569.63716.05
N 1,089 (235)636 (114)453 (121)1,089 (98)636 (69)453 (29)1,089 (28)636 (15)453 (13)

The models with Weibull density distribution were selected for PE backed sales (model 3a – Panel C), PE and non-PE backed SMBOs (models 4a and 4b – Panel D) and all models for liquidation exits (Panel E). The Lognormal density distribution was selected in the model for non-PE backed IPOs (model 2b – Panel B). In all other models, the Gamma density distribution was the best choice for our sample. For PE liquidations, the original model could not converge for any of the selected distributions. We, therefore, present the results of a modified model, having excluded certain variables. In the model for IPO exits of pure buy-outs (model 2b – Panel B), the PRIVATISATION variable was dropped because of collinearity.

Overall, our models exhibit high explanatory power, measured by pseudo R2s and statistically significant likelihood ratios. The ancillary parameters (Kappa, P, Sigma) are positive and highly statistically significant, suggesting increasing exit hazard rates with time and confirming our choice of parametric over semi-parametric models. In all panels, the models for PE backed subsamples exhibit the highest R2s and the highest levels of significance for the relevant ancillary parameters. For example, R2s range from 56% to 84% in the models for PE backed buy-outs and from 2% to 56% in the models for pure sample buy-outs.

The survival experiences of the sample buy-outs across PE backing are presented in Figure 1 (Panel A). The curve representing the survival experience of non-PE backed buy-outs dominates the corresponding curve for PE backed buy-outs throughout the entire investment horizon. The shape of both curves, however, is similar and consistent in suggesting a dramatic increase in exit hazard rates in year 3. The hazard diminishes during year 4 and almost disappears after year 5.

image

Figure 1. Survival Functions

Notes: The x-axis represents time to exit in number of years, while the y-axis represents the estimated probability of survival based on the results of relevant accelerated failure time (AFT) models, reported in Table 10. Panel A presents the survival functions for the total sample. Panel B presents the survival functions for the PE backed subsample by exit methods. Panel C presents the survival functions for the non-PE backed sample by exit methods.

Download figure to PowerPoint

The survival experiences of the sample buy-outs in PE backed and non-PE backed subsamples are presented in Figure 1 (Panels B and C respectively). The diagrams clearly show a different evolution of hazard rates for different exit routes. The hazard rates for IPOs start increasing first. For PE backed IPOs the increase occurs in year two while for non-PE backed IPOs the increase occurs in year one. The exit hazard for sales starts increasing in year 3 and continues increasing during year four. The hazard of SMBO exits is virtually nonexistent in the first three years but from the end of year three we observe an increase. The increase in SMBO hazard is much more evident in the PE backer subsample. Similarly, there is no exit hazard for liquidations until year 4, which is then followed by an increase in the hazard rate. The notable difference between diagrams for PE and non-PE backed SMBOs, liquidations and sales is that for PE backed buy-outs hazard disappears at the beginning of year 5 whilst for the non-PE backed sample it continues throughout year 5. The above evidence suggests the following pecking order for exit routes: IPOs are the preferred exit route, followed by sale, SMBOs, and lastly liquidation. The previously reported early AIM exits clearly contribute to the different survival experiences of PE backed and pure buy-outs exiting via IPOs. For example, in the PE backed sample, IPO hazard starts growing (exponentially) from year 2 to year 4 and then practically disappears. In the pure buy-out sample the hazard starts increasing earlier (in year 1), grows to about 20% in year 2 and reaches 70% in year 4.

(ii) Determinants of Longevity

(a) PE Backing

Overall, the association of PE backing and longevity is negative and weakly statistically significant in model 1 (Panel A). Comparing the results across different exit routes, we see significant differences in the effects of PE backing. While the coefficients for PE backing continue to be negative (but not statistically significant) in models for IPOs (model 2 – Panel B) and SMBOs (model 4 – Panel D), they are positive and highly statistically significant in models for sales (model 3 – Panel C) and liquidations (model 5 – Panel E). Our sample covers buy-outs over five different decades, during which significant events in both the UK economy and PE industry occurred. We therefore examine the robustness of our results and possible differences in longevity of buy-outs by stratifying sample buy-outs across the last five decades and repeating our estimations from Table 10. Unreported results suggest that the PE coefficient is not statistically significant in any of the decades and the sign of the PE variable changes from negative in the 1980s to positive in the 1990s and 2000s. Most importantly, the sign for the PE variable is positive in all decades for IPOs, sales and liquidations. For SMBOs the sign for the PE variable was positive and significant in 1990s and negative (but not significant) in the 2000s.

After controlling for association of PE firms with investment banks, we find no evidence that the reputation of PE firms affect longevity (H2) in any of our survival models. In repeated regressions for different decades, we find evidence that the more reputable firms took longer before taking their buy-outs to IPO exits in the 1980s. Overall, our results do not lend support to our hypothesis 2.

The difference in importance of syndicated deals tends to be divided between IPO and SMBO exits on the one side, and sales and write-offs on the other. While syndicated deals are likely to exit sooner via IPOs and SMBOs (models 2a – Panel B and 4a – Panel D), it takes longer for syndicates to exit via sales (model 3a – Panel C). Simulations run for the median exit time, in model 2a – Panel B, suggest that syndications (with two PE firms) reduce time to exit, relative to non-syndications (i.e., backing by one PE firm), by about 1.5 months. Our results contradict findings for the combined Canadian and US IPO exits reported in Cumming and Johan (2010), but are consistent with the results reported for US IPO exits in Giot and Schweinbacher (2007). Larger syndicate sizes further increase the hazard for IPOs. For example, an increase in the syndicate size from two to four results in a median exit time decrease of about 2.5 months (about 8.5%). The above results support the view presented in earlier literature that syndication adds to overall potential value (Cumming and Johan, 2010) and therefore facilitate exits. Simulations run for the importance of syndicate size for the median sale exit time (model 3a – Panel C) are interesting: increasing the syndicate size from two to four leads to an 80% increase in the median exit time. Obviously, whilst larger syndicates increase the hazard of IPO exits, they tend to drastically reduce the hazard of sale exits. Overall, our results provide support for hypothesis 3 in the IPO and SMBO sub-samples.38 In repeated estimates for different decades, we find that syndication was an important factor of duration in the 1990s and 2000s.

The negative and highly statistically significant (at the 1% level) coefficients for LNPECAPITAL indicate that PE firms exit sample buy-outs faster during periods when the supply of capital is greater (Panels A to D). This is consistent with evidence that higher supply of VC investments shortens duration (Cumming and Johan, 2010). LNPECAPITAL is the variable with the strongest impact on longevity (i.e., exit hazard) across all exit routes (except liquidations). For example, simulations for the median exit time, in model 1a, show that a significant increase in supply of PE funding reduces median investment duration by 12 months.39 The above results provide strong support for our hypothesis 4.

(b) Buy-Out Characteristics

Negative and statistically significant coefficients for LNSIZE in all models provide strong support for our hypothesis 5 (Panels A to E). As expected, larger buy-outs are likely to exit faster. Our simulations for median sale exit time (model 3 – Panel C) show that an increase in size of buy-outs from £10m to £20m, and then from £50 to £100, results in a median exit time decrease of about 22% and 20% respectively. In the models for IPO exits (Panel B) the coefficients for our control variable MAIN are positive but statistically significant only in model 2b – Panel B, confirming that it takes longer for managers in non-PE backed buy-outs to list their companies on the main board. Our control variable for industry classification is negative and only weakly significant in the models for PE backed IPOs (model 2a – Panel B) and PE backed liquidations (model 5a – Panel E). Across the decades for the non-PE backed subsample, services are negative and highly significant in the 1990s and 2000s for sale exits. LNSIZE is negative and significant in the 1990s, but turns positive and statistically significant in the 2000s.

(c) Sources of Buy-Outs

In the total sample (model 1 – Panel A), the coefficient for divestments is negative and statistically significant at the 5% level thus lending support to our hypothesis 6. Across different exit routes, the coefficient for divestments is statistically significant only in model 3 (Panel C) (positive and significant at the 10% level). Across PE backing, the results indicate that management in non-PE buy-outs, originating from divestments, take a longer time to take their companies to IPO and SMBO exits, providing some support for our hypothesis 1 (model 2b – Panel B). We also repeated the estimates for the subsample of buy-outs originating from divestments. Unreported results show a negative and statistically significant PE variable, confirming that PE firms take divestments to exit faster than managers in pure buy-outs.

Overall, receiverships have by far the slowest exits followed by privatisations and divestments. To illustrate this, in model 3a (Panel C), buy-outs originating from receiverships exhibit a relative time ratio of 1.09 relative to divestments (calculated as e0.1139/e0.0291).40 The coefficients for RECEIVERSHIP, however, are negative and statistically significant in models 5 and 5a (Panel E). As expected, PE firms are more likely to write-off more complex deals originating from receiverships in cases of high carry costs (H7b).

Our control variable for PRIVATISATION is not statistically significant, except in models for sales. The negative and statistically significant coefficients for privatisation (models 3 and 3b – Panel C) suggest shorter duration of sales originating from privatisations. Furthermore, we find evidence of the importance of privatisation in repeated estimates for each of the decades covered by our sample. For example, the coefficient for privatisation is negative and statistically significant in the 1980s. The results highlight the importance of buy-outs as a privatisation method in the UK privatisation programme.

(d) Leveraged Deals and Types of Buy-Outs

The results lend very strong support to our hypothesis 8 (models 1 and 1a – Panel A). A negative and highly statistically significant LBO coefficient (at the 1% level) in model 1 indicates that LBOs exit faster than their counterparts. Unreported results for the regression in different decades show that LBOs are consistently negative and statistically significant throughout the 1980s, 1990s and 2000s. This is still the case in the PE backed sample, while in the non-PE backed sample, although negative, the LBO coefficient is not statistically significant. The difference in the results across PE backing highlights the importance of the combined effect of active involvement of PE firms and debt financing on the duration of LBOs.

Our results provide support for our hypotheses 9a and 9b only in samples of pure buy-outs: it takes longer for new management teams to exit via sales and a shorter time to liquidate non-performing MBIs (Model 3b – Panel C and 5b – Panel E). The differences between MBOs and MBIs were also evident when we repeated regressions for the sub-samples of MBOs and MBIs. Whilst the subsample for MBOs exhibits economically and statistically consistent results with the estimates for the total sample, the results for the subsample of MBIs were different.41

(e) Market Conditions

Overall, strong market conditions measured by stock market returns and ‘hot’ exit years are important determinants of longevity and give rise to shorter investment durations (H10a&b). Our results are consistent with the results for the US and Canadian VC market reported in Cumming and Johan (2010). In model 1 (Panel A), the coefficients for the market returns and ‘hot’ IPO years are both negative and statistically significant (at 1% and 5% respectively). LNFTSE is the variable with the strongest impact on the survival of the sample buy-outs. For example, a one-unit increase in this variable increases the exit hazard by approximately one third. Our simulations for the median exit time show that changes from bullish to bearish market conditions contribute to a reduction in the median exit time of six months.42

Regarding exit routes, favourable market conditions, measured by LNFTSE, increase the hazard of IPO exits by more than one-third, but cut the hazard of sale exit to a half (model 2 – Panel B and model 3 – Panel C). For example, simulations indicate that switching from a bullish to a bearish market can increase the median sale exit time by 28%. Sale exits, therefore, tend to be preferred during ‘quiet’ IPO market periods. In the PE backed sample ‘hot’ IPO conditions tend to be more important than the performance of the market index (model 1a – Panel A); the opposite is evident in the non-PE backed sample where the LNFTSE coefficient exhibits statistical significance at the 5% level (model 1b Panel A). Similarly, PE firms tend to take advantage of ‘hot’ market conditions by taking their companies to IPO exits during exit years with exceptional volumes and a degree of underpricing (model 2a – Panel B). On the other hand, managers in pure buy-outs tend not to take their companies to IPO exits during years with ‘hot’ IPO market conditions (model 2b – Panel B). The coefficients for LNFTSE, in models 3a (Panel C) and 5a (Panel E), are positive and statistically significant, indicating that PE firms tend not to exit their portfolio companies via sales, or liquidation, in times of strong market conditions. The combined evidence suggesting a positive association of sale exit time with favourable market conditions and syndications is in line with differences between hazard functions for IPO and sale exits presented in Figure 1 (Panels B and C). PE firms opt for IPOs first (especially during favourable market conditions) and use sale exits as their second choice. Our results are also in line with evidence from the US and Canadian markets suggesting that private exits are not popular in times of strong market conditions (Cumming and Johan, 2010) and that US VC firms consider IPO exits before trade sales (Giot and Schweinbacher, 2007).

7. ROBUSTNESS OF RESULTS AND EXTENSIONS

  1. Top of page
  2. Abstract
  3. 1. INTRODUCTION
  4. 2. HYPOTHESES FOR DETERMINANTS OF BUY-OUT LONGEVITY
  5. 3. DATA AND SUMMARY STATISTICS
  6. 4. UNIVARIATE ANALYSIS ACROSS PE BACKING, EXIT ROUTES AND VINTAGE DECADES
  7. 5. BUY-OUTS SURVIVAL ANALYSIS
  8. 6. ESTIMATION RESULTS
  9. 7. ROBUSTNESS OF RESULTS AND EXTENSIONS
  10. 8. DISCUSSIONS AND CONCLUSION
  11. REFERENCES

In this section we conduct further analysis to examine the robustness of our basic findings. First, we examine the robustness of our results to heterogeneity, clustering, and the assumption of a constant hazard rate. This is followed by an examination of the sensitivity of our results to alternative definitions of explanatory variables. Finally, we compare our results with the results reported in other studies on longevity of buy-outs using different definitions for longevity.

(i) Heterogeneity, Clustering, and a Constant Hazard Rate

Kaplan (1991) highlights the importance of unobserved heterogeneity in relation to modelling longevity. We therefore test for the possibility that some important information may not have been included in our models, resulting in heterogeneity (i.e., frailty). This was achieved by introducing frailty as an unobservable multiplicative effect in our survival models. The (unreported) results of the repeated estimates for all exits were statistically and economically consistent with the results without frailty. We also consider robustness of our results regarding clustering standard errors. We consider two possible clusters: (i) a cluster of observations in different decades and (ii) clusters by PE equity firms. The results are statistically and economically consistent with the previous results. Finally, we estimated the Cox model with the same explanatory variables as in equations (4) and (5). Unreported results suggest that the main inferences are robust to choosing the Cox model with constant hazard rates.43

(ii) Sensitivity of the Results to Alternative Definitions of Explanatory Variables

Given differences in the definition of some variables used in previous studies and various databases, we repeated our estimates using different definitions for LBOs, reputation, time to exit, and market conditions.44 As mentioned earlier, there are notable differences in definitions of LBO deals used in the literature and in various databases. Amess and Wright (2007), for example, define LBOs as transactions with high leverage, highly concentrated equity held by managers, and the PE firm's active monitoring role at board level. In the US, the LBO definition includes all MBIs and highly leveraged PE backed buy-outs (Renneboog et al., 2005). We checked the coherence of our LBO variable with the definition of the variable used elsewhere in relation to concentration of management equity. We were able to collect information on management ownership for LBOs exited via the IPO subsample in our PE backed sample. The mean and median for management ownership of equity (before IPO) in our IPO subsample were 44% and 49% respectively. The high concentration of management ownership in PE backed LBOs is thus consistent with Amess and Wright's definition of LBO transactions. Finally, we check for the robustness of the results using Thomson One Banker's definition of LBOs (LBOT). According to the Thomson One Banker database, a leverage buy-out occurs when an investor or firm offers to acquire a company, taking on an extraordinary amount of debt, with plans to repay it with funds generated from the company or with revenue earned by selling off the newly-acquired company's assets. In addition, the Thomson One Banker database identifies a transaction as an LBO if the investor group includes management or if the transaction is identified as such in the financial press and a majority interest of the target company is acquired. The unreported results of the repeated estimates with LBOT are economically and statistically consistent with the results reported in Table 10.

The results based on a narrower definition of reputation, including only the top three PE firms from our list (TOP3), were not materially different from our original estimates. The same applies for the estimates where market conditions were measured by the natural logarithm of average annual returns on the FTSE All Share Index over the entire investment period (LNFTSEPERIOD). Finally, our results remain robust when the number of months prior to exit (TIMEX) was used as the dependent variable.

(iii) Comparison with the Results from Previous Studies

One of the difficulties in comparing our results on longevity of buy-outs with the results reported in previous studies relates to the fact that some studies report the length of time that buy-outs tend to remain in their original buy-out structure while some report how long buy-outs remain in private ownership. Furthermore, different studies tend to consider exits over different holding periods (e.g., 5, 7 or 10 years). In order to compare our results with earlier buy-out studies, we present results that are directly comparable both in terms of methodology and holding periods (Table 11). We therefore report results by referring to the time to exit from either original buy-out form (i.e., IPO, sale, SMBO and exits via liquidations), private ownership (i.e., IPO, sale to a public company and exits via liquidations) or any buy-out form (i.e., IPO, sale to a non-buy-out company and exits via liquidations).

Table 11.  Comparison of Longevity and Exits from Buy-Outs and Private Ownership
 LongevityExit Status/Routes
  1. Notes: This table presents a comparison of our results on longevity, exits, and exit routes with results reported in previous buy-out research and studies on venture capital (VC) exits. Data and sample coverage for buy-out studies are presented in Table 4. Evidence on PE exits comparable to evidence on VC exits is in bold. The summary presented in Panel C, is adapted from Cumming (2009).

Panel A: Exits of UK Buy-Outs  
Wright et al.,1993; (1981–91)Exits peaked 3–5y26% exits (7y)
Wright et al., 1994; (1981–90)Heterogenous longevity40% exits
Wright et al., 1995; (PE backed-1983–86)Heterogenous longevity; exits peaked 3–5y42% exits; Sales (18.4%); IPO (10%); SMBO (3.8%); failed (9.5%); 29% exits from private ownerhsip (7y)
Jelic et al., 2005; (1964–97)4y (median 3.33y)-IPO 
Nikoskelainen and Wright, 2007; (PE backed-1995–2004)3.5y; 2.6y-IPOIPOs (16%); Sales (28%); SMBO (18%); failed (38%)
This study3.75y (median 3y)-all exits; 3.8y-IPO; 3.3y-sale; 4.4y-SMBO; 3.2y-liquidations; 3.6y-private exits (sale & SMBO); PE backed subsample: 3.3y (median 2.9y)-all exits; 3.25y-IPO; 3y-sale;4.3y-SMBO; 4.4y-liquidations; non-PE backed: 4.3y(median 3.7); 4.8y-IPO; 3.8y-sale; 4.3y-SMBO; 2.9y-liquidations 3.68y in buy-out organisational form– all exits (PE backed subsample: 3.2y; non-PE backed: 4.3y); 3.69y in private ownership– all exits (PE backed subsample: 3.2y; non-PE backed: 4.3y)65% exits from original buy-out structures (70% of PE backed and 58% of non-PE backed) (7y); IPO (47%); sales (21%); SMBOs (9%); liquidation (3%); PE backed subsample: IPO (52%); sales (18%); SMBOs (11%); liquidation (2%); non-PE backed: IPO (39%); sales (28%); SMBOs (6%); liquidation (5%) 50% exits from any buy-out organisational form(55% of PE backed; 44% of non-PE backed) (5y); 53% exits from private ownership(54% of PE backed; 51% of non-PE backed) (7y)
Panel B: Buy-Out Exits Internationally  
World data; Stromberg 2008 (Predominantly PE backed- 1970–2007)World – 6–7y (median) in 1980s and 9y during 1995–99World – 40% exits (IPO-13%; sale -38%; SMBO-24%; failed- 6%); UK- 67% exits (IPOs -11%; sale-42%; SMBO-22%; failed -8%; unknown -10%) World – 55%- exits from LBO organisational form (5y); UK- 59% exits from LBO organisational form (5y)
US data; Kaplan, 1991; (PE backed- 1979–86)Exits peaked in y4; 6.8y longevity in private ownership44% exits (7y); 38% exits from private ownership (7y)
Panel C: VC Exits Internationally  
US data; Cumming and Johan, 2010; (1991–2004)2.95y-IPO; 3.2y-sale & smbo; 2.9y-write-offsIPO (35.7%); sale & smbo (54.6%); write-offs (9.7%)
Canadian data; Cumming and Johan, 2010; (1991–04)2.5y-IPO; 4.1y-sale & smbo; 3.2y-write-offsIPO (5.85%); sale & smbo (74.2%); write-offs (19.9%)
World data;Abdulkadir, 2009; (1990–2006)4.6y-IPO; 5.3y- M&A; 5.9y-write-offsM&A and IPOs are most the popular exit routes.
European data; Cumming, 2008; (1995–2005)3.33y-IPO; 3.38y -sale & SMBO; 3.58y- write-offsIPOs (17%); sale & SMBO (49%); write-offs (34%)
Australasian data; Cumming et.al, 2006; (1989–2001)2.8y-IPO; 3.4y-sale & smbo; 4.6y-write-offsIPO (23%); sale & smbo (60%); write-offs (17%)
European data; Schweinbacher, 2002; (1990–2001)3.7y-all exitsIPO (25%); sale & SMBO (54%); write-offs (21%)

We find that 65% of our sample buy-outs exited their original buy-out form and 53% exited from private ownership within seven years after buy-out (70% and 54% in our subsample of PE backed buy-outs).45 Both percentages are higher than in Wright et al. (1995) and Kaplan (1991). Our results are more in line with the results reported in Stromberg (2008) who reports that while 67% of his UK sub-sample exited the original LBO structure during the sample period, 59% exited LBO organisational form within five years after the original LBO. Our recalculated results for a five year holding period suggest that 50% of our sample buy-outs exited any buy-out form. The relevant percentages are 55% for PE backed and 44% for pure buy-outs. Thus 56% of the pure sample buy-outs remained in a buy-out form for at least seven years. The results for a ten year holding period are even more similar to those reported in Stromberg (2008). For example, 36% of our sample buy-outs remained in any buy-out organisational form at least ten years after the original transactions (30% of PE backed and 40% of non-PE backed) compared to 39% reported for Stromberg's international sample (38% of PE backed and 68% of pure buy-outs). Overall, the duration of various holding periods reported in our study (in original buy-out, private ownership or any buy-out form) lends support to the view that buy-outs present long term rather than short-term form.

Our subsample of PE backed buy-outs exhibited high exit rates, predominantly via IPOs, and an investment duration comparable to the duration reported in other European studies. Overall, the average time to exit (TIMEX) in our PE backed subsample is 3.3 years compared to 3.5 years reported in Nikoskelainen and Wright (2007) and 3.7 years in Schweinbacher (2002). Across exit routes, the duration of our PE backed IPOs (3.25 years) is similar to the average duration of 3.3 years reported for VC backed European deals (Cumming, 2008) but longer than the 2.6 years reported for UK buy-outs (Nikoskelainen and Wright, 2007). The duration of sale and SMBO exits, in our sample, were 3 and 4.3 years respectively. We were not able to compare time to exit for sales and SMBOs separately since previous studies reported their combined results as private exits.46 The only notable difference is the duration of PE backed liquidations (4.4 years) compared to the duration of write-offs reported for European VC deals (3.58 years reported in Cumming, 2008). We attribute this difference to the high percentage of PE backed liquidations originating from receiverships in our sample. Although PE backed deals exhibited shorter longevity than their non-PE backed counterparts, we found no evidence of alleged short-termism of PE firms.

8. DISCUSSIONS AND CONCLUSION

  1. Top of page
  2. Abstract
  3. 1. INTRODUCTION
  4. 2. HYPOTHESES FOR DETERMINANTS OF BUY-OUT LONGEVITY
  5. 3. DATA AND SUMMARY STATISTICS
  6. 4. UNIVARIATE ANALYSIS ACROSS PE BACKING, EXIT ROUTES AND VINTAGE DECADES
  7. 5. BUY-OUTS SURVIVAL ANALYSIS
  8. 6. ESTIMATION RESULTS
  9. 7. ROBUSTNESS OF RESULTS AND EXTENSIONS
  10. 8. DISCUSSIONS AND CONCLUSION
  11. REFERENCES

This paper presents UK evidence on the duration of PE backed and pure buy-outs. Our results lend support to the view that buy-outs present long rather than short term organisational form (Kaplan, 1991; and Stromberg, 2008). Our survival models exhibit high explanatory power, especially in the sample for PE backed transactions. We document that buy-out characteristics (e.g., size), characteristics of PE backing (e.g., syndicates and highly leveraged deals), together with stock market and PE market conditions significantly influence the longevity of buy-outs. The estimated ancillary parameters in our survival models are positive and highly statistically significant, suggesting an increase of exit hazard rates over time. Overall, the exit hazard rates increase dramatically 3–5 years after the original buy-out transaction. Across exit routes, the hazard rates evolve differently: they first start increasing for IPOs, followed by sales and SMBOs. There is also evidence that, in the PE backed subsample, sales and SMBOs tend to be used for buy-outs that have not been brought to IPO exits within four years after original deals. Our results are robust to estimation problems associated with endogeneity, heterogeneity, and alternative specification of variables.

The UK has the most liquid PE and stock markets in Europe, which is likely to affect exit outcomes (Black and Gilson, 1998). It also has one of the most successful second tier markets in the world. This helps to explain the reported popularity of IPO exits in our sample compared to other European countries. Although our PE backed firms exhibit a higher percentage of early exits than their European counterparts, we find no evidence of the alleged short-termism of UK PE firms. Our results suggest that early exits tend to be associated with AIM IPO exits of relatively smaller buy-outs.

The early AIM exits clearly contribute to the different survival experiences of PE backed and pure buy-outs exiting via IPOs. The IPO exits in the sample of pure buy-outs are normally related to the desire of managers to preserve their private benefits associated with the reputational advantage of being the CEO of a public company. In the case of less stringent listing requirements, and in the absence of PE backing, early exits accompanied by a poor post IPO performance are more likely. Given that ‘Nomads’ normally perform some of the roles traditionally performed by PE firms, future work should compare the differences in the post IPO performance of PE backed buy-outs and buy-outs backed by the ‘Nomads’.

Footnotes
  • 1

    In the US literature, PE firms are associated with investments in leveraged buy-outs (LBOs), while VC firms are normally associated with early stage investments. In the European literature, however, PE investments sometimes include both investments in buy-outs and early stage investments. Since we focus on buy-outs, our preferred terms in this paper are ‘PE firm' and ‘PE investment.' When discussing the results of previous studies, however, we refer to the term used by the authors, and clearly state which types of buy-outs are included in their samples.

  • 2

    See Wolf (2007) and The Economist (2009).

  • 3

    Nikolaskenainen and Wright (2007) report coherence of gearing ratio and percentage of debt in initial transactions.

  • 4
  • 5

    The Private Equity International is the leading financial information group dedicated to the alternative asset classes of private equity, real estate, and infrastructure globally.

  • 6

    The comparison is based on information provided in Stromberg (2008) and aggregate data from various CMBOR publications.

  • 7

    The coverage would have been much higher had we excluded Capital IQ observations without data on enterprise value.

  • 8

    The CMBOR dataset is the most comprehensive dataset on UK buy-outs, and the only database covering the entire period of our study. However, we have access only to aggregate annual data during 1980–2004 from various CMBOR publications. Consequently, we were not able to compare the coverage during the 1960s and 1970s.

  • 9

    Our tests are similar to those conducted in Cumming and Zambelli (2010).

  • 10

    Our industry classification follows that of Gompers et al. (2008) which re-classified numerous industry codes into nine industry groups more in line with well documented specialisation within the VC industry.

  • 11

    The percentage is higher in the subsample of pure buy-outs than in the subsample of PE backed buy-outs (7% compared to 1.8%).

  • 12

    For example, our sample includes 205 deals from the AIM while Nikoskelainen and Wright (2007) include only 2 AIM exits out of total of 52 IPO exits included in their sample.

  • 13

    It is worth noting that Schweinbacher (2002) reported 25.3% of IPO exits in his sample of European VC backed deals.

  • 14

    Stromberg (2008) considers bankruptcy filings, financial restructuring and liquidations.

  • 15

    Fame database covers more than three million private, active and inactive, UK companies.

  • 16

    When we adopted a wider definition, similar to one adopted in Stromberg (2008), the percentage was 3.5%.

  • 17

    Especially in the cases of some banks that have only entered the buy-out market recently.

  • 18

    This procedure is consistent with the one used in previous studies on reputation of intermediaries (e.g., Megginson and Weiss, 1991).

  • 19

    This is the first ever official ranking of PE firms. The top 50 accounted for 75% of total PE deal activity globally since 2002 (http://www.privateequityinternational.com/pei50).

  • 20

    3i, Cinven, Bridgepoint, Charter House and Candoven. We included Kohlberg Kravis Roberts & Co (KKR) and Apax in the top 10 most reputable PE firms, outside our criteria, since they had established a reputation internationally, before entering the UK buy-out market in the 1990s. They are also highly ranked in the top 50 list.

  • 21

    This is largely due to the high number of deals backed by 3i and Cinven, in our sample.

  • 22

    Unreported results also suggest that the majority of PE backed buy-outs that resulted in liquidation, received backing from a single, less reputable bank affiliated PE firm.

  • 23

    Unreported results suggest that they also tend to have more funding rounds per deal (2.9) than buy-outs exiting via other routes.

  • 24

    For example, the difference in median IPO size and size of any other exit route is statistically significant at the 1% level.

  • 25

    The Mann-Whitney test for differences in medians is statistically significant at 1% for sale and SMBOs, and 5% for liquidations.

  • 26

    For non-PE backed buy-outs the mean size increased from £85m to £171m.

  • 27

    As reported in Table 6– Panel B.

  • 28

    Unreported results suggest that one fifth of PE backed sample buy-outs exited their original buy-out structure within two years, and about the same proportion remained private more than ten years after the buy-out transaction.

  • 29

    Unreported results suggest that median time to exit for the AIM IPOs was 40 months compared to 34 months for IPOs from the main LSE board. The difference is not statistically significant and both market segments exhibit a decreasing trend in time to exit during the sample period.

  • 30

    The medians for PE backed and non-PE backed in the subsample of IPOs exited in the main market were 31 and 47 months respectively. The respective medians for the AIM IPOs were 33 and 47 months.

  • 31

    Unreported results suggest that 75% of PE backed liquidations originate from buy-outs in receiverships.

  • 32

    Available in Stata (2007).

  • 33

    The model is adopted from, and estimated in, Stata (2007, p.310).

  • 34

    Relevant ancillary parameters for Weibull and Lognormal models are P and Sigma respectively.

  • 35

    The AFT model changes the time scale by a factor of exp (−xiβ).

  • 36

    Specifically, we test whether the categorical variable for PE backing has potentially become a choice variable, correlated to unobservables relegated to the error term.

  • 37

    The results of the Hausman tests are: Chi2 (9) = 5.28, Prob>chi2 = 0.8090.

  • 38

    This is also consistent with the results reported in Table 6–Panel D.

  • 39

    The simulations were based on ±25% average annual changes in the amount of UK private equity investments.

  • 40

    For dummy variables, time ratios can be directly compared with each other, resulting in relative time ratios. The relative time ratio indicates how fast the change of a variable impacts the conditional exit probability. See Giot and Schwienbacher (2007, p. 689).

  • 41

    The unreported results suggest that none of the coefficients remained significant, and the sign of the coefficient for the SERVICES variable changed from positive to negative.

  • 42

    The market conditions were approximated by 25% negative (bearish) and 25% positive (bullish) changes in the FTSE All Share Index returns during 12 months prior to exit.

  • 43

    Above estimates are available upon request, from the author.

  • 44

    All alternative variables are defined in Table 1.

  • 45

    For the 10-year holding period, 41% of our sample buy-outs remained in private ownership (40% of PE backed and 42% of non-PE backed).

  • 46

    The combined average duration of private exits (sale and SMBO) is 3.58 years which is slightly longer than 3.38 years reported for European VC deals in Cumming (2008). The duration, however, is longer than for the US (2.95 years) and Canadian (2.5 years) VC backed IPOs (Cumming and Johan, 2010).

REFERENCES

  1. Top of page
  2. Abstract
  3. 1. INTRODUCTION
  4. 2. HYPOTHESES FOR DETERMINANTS OF BUY-OUT LONGEVITY
  5. 3. DATA AND SUMMARY STATISTICS
  6. 4. UNIVARIATE ANALYSIS ACROSS PE BACKING, EXIT ROUTES AND VINTAGE DECADES
  7. 5. BUY-OUTS SURVIVAL ANALYSIS
  8. 6. ESTIMATION RESULTS
  9. 7. ROBUSTNESS OF RESULTS AND EXTENSIONS
  10. 8. DISCUSSIONS AND CONCLUSION
  11. REFERENCES
  • Abdulkadir, M. (2009), ‘Analytics and Empirical Studies of IPO Survivals and Venture Capital Activities’, PhD Thesis ( University of Manchester ).
  • Akaike, H. (1974), ‘A New Look at the Statistical Model Identification’, IEEE Transactions on Automatic Control, Vol. 19, No. 6, pp. 71623.
  • Amess, K. and M. Wright (2007), ‘The Wage and Employment Effects of LBOs in the UK’, International Journal of Economics and Business, Vol. 14, pp. 17995.
  • Ball, E., H-H., Chiu and R. Smith (2008), ‘Exit Choices of Venture-Backed Firms: IPO v. Acquisition’, Working Paper (SSRN: http://ssrn.com/abstract=1301288).
  • Bayar, O. and T. Chemmanur (2006), ‘IPOs or Acquisitions? A Theory of the Choice of Exit Strategy by Entrepreneurs and Venture Capitalists’, Working Paper ( Boston College ).
  • Barclays Private Equity Exits Publications – various issues.
  • Berglof, E. (1994), ‘A Control Theory of Venture Capital Finance’, Journal of Law, Economics, and Organization, Vol. 10, pp. 24767.
  • Black, S. and J. Gilson (1998), ‘Venture Capital in Europe and Financing Innovative Companies’, Economic Policy, Vol. 34, pp. 23169.
  • Brau, J., B. Francis and N. Kohers (2003), ‘The Choice of IPO versus Takeover: Empirical Evidence’, The Journal of Business, Vol. 76, pp. 583612.
  • British Venture Capital Association (BVCA) reports – various issues.
  • Centre for Management Buy-out Research (CMBOR) Quarterly Reviews – various issues.
  • Coakley, J., L. Hadass and A. Wood (2008), ‘Hot IPOs Can Damage Your Long-run Wealth’, Applied Financial Economics, Vol. 18, No. 14, pp. 111120.
  • Cochrane, J. (2005), ‘The Risk and Return to Venture Capital’, Journal of Financial Economics, Vol. 75, pp. 352.
  • Cumming, D.J. (2006), ‘Adverse Selection and Capital Structure: Evidence from Venture Capital’, Entrepreneurship Theory and Practice, Vol. 30, No. 2, pp. 15583.
  • Cumming, D.J. (2008), ‘Contracts and Exits in Venture Capital Finance’, Review of Financial Studies, Vol. 21, pp. 194782.
  • Cumming, D.J. and S. Johan (2009), Venture Capital and Private Equity Contracting: An International Perspective ( San Diego , CA : Elsevier Science Academic Press).
  • Cumming, D.J. and S. Johan (2010), ‘Venture Capital Investment Duration’, Journal of Small Business Management, Vol. 48, No. 2, pp. 22857.
  • Cumming, D.J. and J.G. MacIntosh (2001), ‘Venture Capital Investment Duration in Canada and the United States’, Journal of Multinational Financial Management, Vol. 11, pp. 44563.
  • Cumming, D.J. and J.G. MacIntosh (2003), ‘A Cross-country Comparison of Full and Partial Venture Capital Exits, Journal of Banking and Finance, Vol. 27, pp. 51148.
  • Cumming, D.J. and U. Walz (2010), ‘Private Equity Returns and Disclosure Around the World’, Journal of International Business Studies, Vol. 41 No. 4, pp. 72754.
  • Cumming, D.J. and S. Zambelli (2010), ‘Illegal Buyouts’, Journal of Banking and Finance, Vol. 34, No. 2, pp. 44156.
  • Cumming, D.J., G. Fleming and A. Schwienbacher (2006), ‘Legality and Venture Capital Exits’, Journal of Corporate Finance, Vol. 12, pp. 21445.
  • European Venture Capital Association (EVCA) (2001), EVCA Yearbook ( Zaventem ).
  • Gailen, H. and M.R. Vetsuypens (1989), ‘Management Buyouts of Divisions and Shareholder Wealth’, Journal of Finance, Vol. 44, pp. 95370.
  • Giot, P. and A. Schwienbacher (2007), ‘IPOs, Trade Sales and Liquidations: Modelling Venture Capital Exits Using Survival Analysis’, Journal of Banking and Finance, Vol. 31, pp. 679702.
  • Gompers, P.A. (1996), ‘Grandstanding in the Venture Capital Industry’, Journal of Financial Economics, Vol. 42, pp. 13356.
  • Gompers, P.A. and J. Lerner (1999), The Venture Capital Cycle (MIT Press, Cambridge ).
  • Gompers, P.A., A. Kovner, J. Lerner and D. Scharfstein (2008), ‘Venture Capital Investment Cycles: The Impact of Public Markets’, Journal of Financial Economics, Vol. 87, pp. 123.
  • Jelic, R., B. Saadouni and M. Wright (2005), ‘Performance of Private to Public MBOs: The Role of Venture Capital’, Journal of Business Finance & Accounting, Vol. 32, Nos. 3&4, pp. 64381.
  • Jensen, M.C. (1989), ‘Eclipse of the Public Corporation’, Harvard Business Review (Sept/Oct), pp. 6175.
  • Kaplan, S. (1991), ‘The Staying Power of Leveraged Buyouts’, Journal of Financial Economics, Vol. 29, pp. 287314.
  • Kaplan, S., B. Sensoy and P. Stromberg (2002), ‘How Well Do Venture Capital Databases Reflect Actual Investments?’, Working Paper ( University of Chicago ).
  • KPMG, MBO Commentaries and New Issue Statistics – various issues.
  • Lee, P.M. and S. Wahal (2004), ‘Grandstanding, Certification and Underpricing of Venture Capital Backed IPOs’, Journal of Financial Economics, Vol. 73, pp. 375407.
  • Lerner, J. (1994), ‘Venture Capitalists and the Decision to Go Public’, Journal of Financial Economics, Vol. 35, pp. 293316.
  • London Stock Exchange, Primary Market Factsheets - various issues.
  • Lowry, M. and G.W. Schwert (2002), ‘IPO Market Cycles: Bubbles or Sequential Learning?’, Journal of Finance, Vol. 57, pp. 117198.
  • Megginson, W. and K. Weiss (1991), ‘Venture Capitalist Certification in Initial Public Offerings’, Journal of Finance, Vol. 46, pp. 879903.
  • Nikoskelainen, E. and M. Wright (2007), ‘The Impact of Corporate Governance Mechanisms on Value Increase in Leveraged Buy-outs’, Journal of Corporate Finance, Vol. 13, pp. 51137.
  • Poulsen, A. and M. Stegemoller (2008), ‘Selling Out to Public Firms vs. Initial Public Offerings’, Financial Management, Vol. 37, pp. 81101.
  • Renneboog, L., T. Simons and M. Wright (2005), ‘Leveraged Public to Private Transactions in the UK’ (Working Paper, European Corporate Governance Institute).
  • Ronstadt, R. (1986), ‘Exit Stale Left: Why Entrepreneurs End their Entrepreneurial Careers Before Retirement’, Journal of Business Venturing, Vol. 1, pp. 32338.
  • Schwienbacher, A. (2002), ‘Venture Capital Exits in Europe and the United States’, Working Paper ( University of Amsterdam ).
  • Stata: Survival Analysis and Epidemiological Table Reference Manuel (2007), Stata Statistical Software: Release 10, [ST], (Stata Press, StataCorp LP, College Station , Texas ).
  • Stoughton, N.M., K.P. Wong and J. Zechner (2001), ‘IPOs and Product Quality’, Journal of Business, Vol. 74, pp. 375408.
  • Stromberg, P. (2008), ‘The New Demography of Private Equity’, in J. Lerner and A. Gurung (eds.), The Global Impact of Private Equity Report 2008, Globalization and Alternative Investments (Working Papers, World Economic Forum), Vol. 1, pp. 326.
  • The Economist, ‘Battle of the Big Beasts’ (25 July, 2009).
  • The Private Equity International (http://www.privateequityinternational.com/pei50).
  • Thompson, S. and M. Wright (1995), ‘Corporate Governance: The Role of Restructuring Transactions’, Economic Journal, Vol. 105, pp. 690703.
  • Wolf, M. (2007), ‘Barbarians at the Gates: The Balance of Pros and Cons’, Financial Times (8 March).
  • Wright, M., K. Robbie, S. Thompson and K. Starkey (1994), ‘Longevity and the Life Cycle of MBOs’, Strategic Management Journal, Vol. 15, pp. 21527.
  • Wright, M., S. Thompson, K. Robbie and P. Wong (1995), ‘Management Buy-outs in the Short and Long Term’, Journal of Business Finance & Accounting, Vol. 22, pp. 46182.
  • Wright, M., K. Robbie, Y. Romanet, S. Thompson, R. Joachimsson, H. Bruining and A. Herst (1993), ‘Harvesting and the Longevity of Management Buy-outs and Buy-ins: A Four Country Study’, Entrepreneurship Theory and Practice, Vol. 18, pp. 90109.
  • Xuan, T. (2007), ‘The Role of Venture Capital Syndication in Value Creation for Entrepreneurial Firms’, Working Paper (SSRN; 954188).