New Approaches to Institutional Portfolio Performance Attribution and Private Equity Risk

Abstract

This work is comprised of three separate works which are summarized as follows: 1.) We develop a new method to estimate private equity funds' market beta from cash flows. Our methodology extends the widely known public market equivalent calculation to a cross-sectional regression. By simply regressing funds' internal rates of return on their paired market internal rates of return, we are able to estimate private equity market betas. For venture funds, we find a high market beta. For buyout funds, we find a low beta. Though we have a small sample, our results fall in line with those recently reported in the literature. 2.) We extend the Brinson et al. (1986) performance attribution framework to support institutional-specific requirements, including a hierarchical structure and multiple benchmark styles. By attributing performance to four statistics (e.g. manager alpha, portfolio construction, tactical and strategic), we are able to remove the interaction term, which is a commonly referred to shortcoming of Brinson attribution. We subsequently modify Frongello (2002) linking to produce pro-rated multiperiod attributes which sum to meaningful statistics using notional portfolios of multitier excess returns. 3.) A number of methods have been developed to link single-period arithmetic attribution results. We present the first institutional portfolio empirical study comparing the most referenced methods for producing additive multiperiod attributes from their single-period counterparts. While our findings suggest the methods typically produce similar results, we find a pattern in the way the methods' results relate to one another. We find the Modified Frongello Method and Cariño Method to produce nearly identical results, the Frongello and Cariño methods to cluster and the Naïve and Menchero methods to be outliers.