We introduce a methodology to estimate the historical time series of returns to investment in private equity. The approach requires only an unbalanced panel of cash contributions and distributions accruing to limited partners, and is robust to sparse data. We decompose private equity returns into a component due to traded factors and a time-varying private equity premium. We find strong cyclicality in the premium component that differs according to fund type. The time-series estimates allow us to directly test theories about privateequity cyclicality, and we find evidence in favor of the Kaplan and Strömberg (2009) hypothesis that capital market segmentation helps to determine the private equity premium.