Research on private equity (PE) has been challenged by the difficulty of obtaining high quality data. Private equity is called “private” for a reason. There is no requirement for those running private equity funds, the General Partners (or GPs), to make their data available. Since no data set exists on the entire universe of private equity funds, researchers and practitioners are forced to rely on samples, each of which might differ depending upon fund characteristics and collection methods employed by the data provider. This paper combines PE data from multiple providers. Doing so allows us to compare the scope of coverage across sources and conduct a more comprehensive study of the investment performance delivered by PE funds.
We study databases created and maintained by four well-established commercial firms: Burgiss, Cambridge Associates (CA), PitchBook, and Preqin (listed alphabetically) and thank all for supplying data. Each firm has its own business model, methods for gathering data, and approach to dealing with confidentiality issues. If each data provider’s sample were a completely random draw from the same underlying universe, we would expect similar messages to emerge across all databases. This should very likely occur for reasonably large sample sizes that may overlap to a large extent. Therefore, comparing results across databases, each constructed in 3 different ways and with different possible biases, provides insights into the likely effect of any biases on conclusions about PE performance.
Unlike much prior research on venture capital and buyout funds, the samples we use in this study include all funds in each database, not just those focused on investing in North America. Moreover, we examine more recent data (through June 2014 for vintage years 1984-2010) than has prior work comparing data sources. Taking advantage of this more comprehensive data, our study contributes to private equity research on a number of fronts. First, we report sample sizes available for research (both in North America and elsewhere). Our comparisons shed light on the relative coverage and performance information available from different databases. As the industry becomes more global, important research questions may require data from multiple countries. Second, we reassess existing research findings for North America. We use absolute performance data (i.e. investment multiples and internal rates of return) from all four databases. Unlike prior work, we harness two independent sources of cash flow data necessary to compare performance relative to public markets. Third, we provide initial results on performance outside North America and compare it to our findings on North American data. Fourth, we provide detail on data providers’ approaches to obtain information and categorize funds. These details point the way for understanding why results may differ across samples and, hopefully, serve as a guide for both research and practice. Fifth, we investigate a new database provided by PitchBook, not previously studied, in order to help inform conclusions.