Abstract
Having discussed the various steps of an exit process and relating considerations in the previous section, this part elaborates on the empirical study underpinning this book.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
The research focus of this study is limited to buyout firms undertaking midsized and large investments defined with a minimum transaction value of €100 million at the time of exit. While NVCA (2005a) classifies US mid-sized buyouts at a minimum transaction value of US$250 million, the lower €100 million threshold is not uncommon for the smaller European private equity market (EVCA 1998).
Gompers and Lerner (2000) examine a dataset of 4000 venture investments from a database by the consulting firm ‘VentureOne’ with regard to transaction valuation. Cochrane (2005) analyses more than 16,000 financing rounds for venture investments also based upon a dataset provided by ‘VentureOne’.
See Lerner and Hardymon (2002) for a number comprehensive case studies capturing several aspects of the private equity business.
Kraft (2001) conducts a limited number of case studies of private equity transactions in order to explore the investment decision approach from private equity firms in detail and to facilitate the interpretation of his survey results.
The term ‘triangulation’ stems from navigation and military strategy where it means that several reference points are used to determine an object’s exact location. As Scandura and Williams (2000, pp. 1249–1250) summarise, triangulation reduces trade-offs of individual research strategies and can strengthen the validity of a study.
The survey design process lasted three months from April to June 2005 and benefited from input provided by experienced academic researchers, a number of private equity firms, and the ‘European Private Equity and Venture Capital Association’ (EVCA), which has been supporting the survey. Similar surveys have been conducted successfully with private equity firms both in Europe (Bottazzi and Da Rin 2002, Bottazzi, Da Rin and Hellmann 2004) and the United States (Cumming and Macintosh 2001, 2003a, 2003b, KRAFT 2001).
These are questions 1 to 10 and 13 in the questionnaire that is set out in an appendix to this work. It is important to note that Schwienbacher (2002) had a different research target group in mind — early-to mid-stage venture capitalists and not buyout firms. Despite the different type of recipients, the questions included from his survey seem still highly relevant as they primarily serve to categorise private equity investors.
Kraft (2001) received 46 responses at a response rate of 23%, Schwienbacher (2002) obtained 66 European responses at a response rate of 18%, Bottazzi, Da Rin and Hellmann (2004) received 150 responses at a response rate of 15%.
Top 50 and Top 100 ranking of relevant buyout investors is based upon Mergermarket (2005). Ranking according to realised exit volumes by investor.
Gompers and Lerner (2000, p. 285) confirm that in 1998 more than 80% of all private equity funds operating in the United States have been organised in form of limited, closed-end partnerships. Please refer to section 2.1.4.2 for further details on fund characteristics.
Fink (2003, p. 79) sets out considerations for the application of numerical score variables in statistical analysis such as multiple regressions.
The question formulation and scaling relating to variables 2 to 7 is based on Schwienbacher (2002).
The design and coding of this variable as a sum of numerical scores has considered the principles set out in FINK (2003, pp. 78–87) and Fink and Kosecoff (1998, pp. 60–77).
Categorical variables are frequently labelled as ‘dummy’ variables, as they can only take the binary values 0 and 1, expressing whether a qualitative characteristic is applicable or not. Please refer to FINK (2003, pp. 28–29) or Field (2005, pp. 199–201) for an introductory note on dummy variables, or Gujarati (1995, pp. 499–539) for a discussion of the application of dummy variables in regressions.
The formulation and scaling for this question has been based on Schwienbacher (2002).
Nahata (2004) argues that both portfolio firm specific and private equity investor specific aspects require consideration when analysing exit decisions.
Please refer to section 3.2.3 for a detailed review of Gompers’ (1996) ‘grandstanding’ theory and related empirical findings.
Lerner et al. (2003) prove empirically that new issuance activity in the public equity markets tends to be clustered in periods, so-called ‘market windows’.
Cumming and Macintosh (2003b) also indicate that the performance and development of the underlying portfolio firms are key determinants affecting exit decisions.
Historical performance is an important consideration as it credibly helps demonstrating the quality of business to potential buyers (Nahata 2004, p. 22). A private equity investor is likely to consider waiting for a period of successful financial performance, on which basis potential buyers of the company are potentially paying a higher consideration. Neus and Walz (2004) explain this as a reduction in information asymmetry between current and future owner of a company. They argue that a later divestment following positive financial performance can help to overcome high information costs and can lead to higher price achieved.
Please refer to Kraft (2001, pp. 280–289) for an interesting discussion and analysis of value drivers in private equity investments.
Gifford (1997) also developed a model regarding the optimal allocation of a venture capitalist’s time between monitoring existing portfolio companies and acquiring new investments. The timing of exits in her concept is, similarly to Cumming and Macintosh (2001, 2003b), influenced by the relationship of benefits keeping an investment and opportunity costs (time).
Other authors also indicate the relationship between adding value to a business and related forms of cost as a factor in exit decisions (i.e., Lerner 1999, Gompers and Lerner 2001, Tykvova 2003a). Please refer to section 3.2 for a more detailed discussion of this concept.
Please refer to section 2.1.4.2 for an overview of private equity firm types and common compensation structures. Alternatively Gompers and Lerner (1999b) provide a detailed review of private equity compensation arrangements.
Please refer to section 3.2.3 for a discussion of this concept. Alternatively, please refer to Gompers (1996) for a detailed elaboration and empirical support for this theory.
Lee and Wahal (2004) conducted an extensive empirical study of the US venture capital market providing clear support for Gompers’ (1996) theory.
Performing a correlation analysis of the ranks between the capacity of private equity professionals and monitoring requirements leads to a correlation coeffcient r of 0.52 with a r2 of 28.9%, which a Spearman’s rank correlation test confirms as significant. For a discussion and guidelines to conduct rank correlation analysis, please refer to Gujarati (1995, pp. 372–374) or Fink (2003, pp. 56–61). No other pair of factors exhibits a significant correlation exceeding a r2 of 25%, which indicates that a principal component factor analysis is unnecessary (Kirchhoff, Kuhnt, Lipp and Schlawin 2003, pp. 81–93), particularly as other factors have been intentionally selected to cover separate aspects. One further exception could have been fundraising requirements and overall fund performance, which exhibit a r2 of 19.3%.
Examining correlations between ranks and numerical values also the principles of Spearman’s rank correlation test find application (Fink 2003, pp. 60–61).
Findings by Morellec and Zdahnov (2005) also indicate that companies’ appetite for acquisitions varies in industry specific patterns and correlates with the industries’ average stock market valuations.
Nahata (2004) emphasises the relevance of portfolio firms’ financial performance for exit decisions. Cumming and Macintosh (2003b) also indicate that the performance and development of the underlying portfolio firms are key determinants affecting exit decisions.
Please also refer to Leschke (2003, pp. 249–251) for comments on the importance of financial performance in this context.
Examining correlations between ranks and numerical values also the principals of Spearman’s rank correlation test can serve to confirm the significance of correlations (Fink 2003, pp. 60–61).
Prior to being included in the initial regression models, all independent variables were tested for perfect multicollinearity, which has to be avoided in regression models (Field 2005, p. 170).
Multiple regressions are expressed in linear equations such as: Y = β0 + β1X1 + β2X2 +... + βjXj; where Y represents the dependent variable, β0 is the intercept, β1 to βj are coefficients to the explanatory variables X1 to Xj. Like in simple regression models, the equation is iteratively solved through the reduction of the residual sum of squares not explained by the model. Please refer to Backhaus, Erichson, Plinke and Weiber (2003, pp. 60–111) for a comprehensive and user-friendly introduction to multiple regression modelling. Glantz and Slinker (2001) discuss more advanced issues in the application of regression techniques. Fink (2003, pp. 61–80) summarises relevant issues when applying regression models for the analysis of survey data. He flags that regression models that comprise numerical scores primarily should be used to analyse relationships rather than predicting future outcomes. Hansen, Hurwitz and Madow (1993, pp. 56–92) highlight potential pitfalls and biases when testing survey data with regression models and other statistical methods.
Backward elimination means that in an initial regression equation all potential independent variables are included. In iterative steps the variables that contribute least to the overall explanatory power of the model are eliminated, looking at each variable’s impact on the residual sum of squares of a model. Please refer to Glantz and Slinker (2001, pp. 263–273) or to Rencher (2002, pp. 351–358) for further details on sequential variable selection algorithms.
Schwienbacher (2002) used the same question in a survey addressed to venture capitalists.
Although the categorical variables in this context are difficult to interpret as smaller subsets of firms in the sample are differentiated from a majority, their application is still sensible. As Gujarati (2003, pp. 297–303) points out, the application of dummy variables is still a valid approach in these circumstances, provided they contribute to the overall explanatory power of a model and do not have a considerable influence on the overall direction of the results, which has been tested in the analysis.
Please refer to Anson (2004) for a relevant note on trends in the private equity market and statistics about capital commitments to the industry. Also refer to section 2.2.5 for a discussion of industry trends.
Please refer to Lenoir (2003, p. 242–244) for brief comments on the different managerial strategies necessary to facilitate different exit routes. Wall and SMITH (1997, p. 9) list several comments by practitioners indicating that private equity investors are particularly sceptical to buy assets from other investors when already comprehensive restructuring measures have been taken by the previous financial owner.
Please also refer to Sahlman (1990) for an extensive discussion of private equity corporate governance aspects.
See also Bouresli, Davidson and Abdulsalam (2002, p. 71–81) for a detailed discussion of possible measures and actions taken by private equity firms to reduce information asymmetries and agency risks.
Hilb (2005) argues that a successful corporate governance system needs to take into account the interests and perspectives of all major stakeholders and not only shareholders. Anders (1992) already criticised leveraged buyout firms in the early 1990s for governing portfolio firms in an uni-dimensional, shareholder-centred approach and highlighted that the interests and well-being of the employees of firms have often been neglected or downplayed. He stressed that this in a longer term harms private equity firms as they build up bad reputation with certain types of stakeholders which might make future acquisitions and transactions more difficult.
The regression models include a number of categorical or dummy variables. In this context one needs to caution that as these dummy variables are introduced to distinguish smaller subsets of firms from the majority of buyout firms captured in the sample resulting coefficients of determination might be inflated. Nevertheless, as Gujarati (1995, pp. 297–303) points out, the application of dummy variables also if only for a smaller subset of the sample is a valid approach. It is important to state that the dummy variables in this analysis have no considerable influence on the overall direction of the results.
Due to the high overall transaction costs of dual-track divestment processes, such procedures are rarely applied in the context of smaller portfolio firms. Moreover, as in most dual-track exits, an IPO is prepared as one of the divestment channels, portfolio companies typically have to exceed a certain size threshold for a feasible public listing (i.e., Wright and Robbie 1998, p. 551).
Lerner, Shane and Tsai (2003) prove empirically that new issuance activity in the public equity markets tends to be clustered in periods, so-called ‘hot market’ phases when comparatively attractive valuations are achieved. Chiang and Harikumar (2004) also prove the presence of ‘market windows’ in the US stock markets empirically.
Baker and Mckenzie (2005, p. 15) argue that the timing of communication of the dual-track nature of a process to bidders is highly critical. Practitioners in expert interviews admitted that many exit processes start off as secret dual-track procedures, whereby bidders in an M&A process are not informed that another exit channel process is pursued in parallel, in order to minimise the risk of discouraging bidders. Frequently preparations of alternative divestment tracks can be stopped following firm indications of attractive offers by M&A bidders, ideally prior to the need for the disclosure of the dual-track nature of the sale process. However, legal disclosure requirements in connection with a public listing can impose the necessity for an upfront communication to bidders.
Please refer to section 4.2.1 for a discussion of ‘pro-active’ portfolio management for private equity firms. Also refer to Lieber (2004) for a comprehensive note on this topic.
Performance data has been compiled from annual performance review notes published by the ‘European Private Equity and Venture Capital Association’ (EVCA) for the period from 1995 to 2004. The performance review reports have been accessed online under: http://www.evca.com/html/investors/inv_performance. asp on 2 December 2005.
Please refer to Hansen. Hurwitz and Madow (1993. pp. 56–92) for a detailed discussion of potential biases and errors in survey sample results.
Rights and permissions
Copyright information
© 2007 Springer-Verlag Berlin Heidelberg
About this chapter
Cite this chapter
(2007). Empirical analysis: Exit behaviour and efficiency. In: Private Equity Exits. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540-70954-1_5
Download citation
DOI: https://doi.org/10.1007/978-3-540-70954-1_5
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-70953-4
Online ISBN: 978-3-540-70954-1
eBook Packages: Business and EconomicsBusiness and Management (R0)