Benchmarking Private Equity Performance When Fund-Level Cash Flows are Missing
针对超50%私募股权基金缺失现金流数据的问题，提出三种模拟方法估算风险调整后业绩，其中基金协变量回归-机器学习方法最准确，发现房地产基金风险调整后表现弱于并购和风投基金。
Abstract Private equity funds without reported cash flows are generally omitted from analysis of risk-adjusted investment performance. This can significantly reduce sample sizes, thus reducing statistical power, as well as possibly introduce bias into the analysis. To address these issues we use Preqin data, which is missing detailed cash flow information at the fund level for over 50% of its funds. To impute risk-adjusted performance for funds without cash flows, we propose three simulation methods. Among the three, the fund covariate regression-machine learning (FCR-ML) method stands out, providing unbiased and least noisy estimates of risk-adjusted performance. Performance of funds without cash flows is well explained by observables, including fund manager characteristics, which reduces concerns over cash flow reporting selection based on unobservables. Imputed performance adds information to the sample, where the combined sample shows Real Estate funds underperform Buyout and Venture Capital on a risk-adjusted basis over our sample period.