Content area
Full Text
Common risk factors are well documented in the returns of publicly traded assets. The Fama–French three-factor model, which includes market (MKT), size (SMB), and value (HML) factors, explains more than 90% of the return variation of diversified public equity portfolios and leaves little room for manager alpha. Evidence for the type and level of factor risk in private assets is less clear. A more complete understanding of the factor betas that explain reported buyout, venture capital, and private real estate returns would provide additional insights for asset allocation and portfolio construction.
Intuitively, the underlying assets of private equity and public equity (operating businesses across industries), or of private core real estate and public real estate investment trusts (REITS) (income properties across property types), are fundamentally the same. However, reported private equity and private real estate returns suggest there are material differences from their publicly traded counterparts. Much of the difference can be attributed to autocorrelation in private asset returns, which is driven by conservative appraisals and delayed adjustments to market prices (stale pricing). In contrast, publicly traded assets are continuously market priced. This presents a challenge when attributing market-priced factor risk to appraisal-based private asset returns.
There are three broad approaches to evaluating the factor risk of private assets: time series, cash flow, and characteristics. A time series of returns represents the periodic returns reported to investors through time. Cambridge Associates, the National Council of Real Estate Investment Fiduciaries (NCREIF), and other organizations publish indexes of quarterly private asset returns, which are constructed from pooled cash flows and intermediate valuations.1 A time series of returns is helpful when using modern portfolio theory (MPT) to identify and select an efficient portfolio that might include private asset classes. MPT requires forecasted return and risk parameters, which are often informed by historical time series, but stale pricing has the effect of muting variance and correlation—and estimates of factor betas. An unsmoothing procedure can reduce these effects before using a private asset return series to estimate MPT inputs or conduct factor attribution.
The time series literature includes work by Fan, Fleming, and Warren (2013), who regressed contemporaneous and lagged market, size, and value factors on buyout and venture capital returns. This approach requires many right-hand-side variables (number of...