Alternative assets are gaining increasing importance in investors' portfolios. One of their defining characteristic is their poor liquidity, which often translates into an inherent smoothing process of the returns. For asset allocation purposes, this feature has to be seriously addressed as it leads to a severe underestimation of the variance of returns and their correlation with other (standard) assets. In this paper, in order to deal with practical issues, we extend previous researches that model the smoothing process as a moving-average one in several directions: (i) we propose a correction for the case of numerous illiquid assets; (ii) we investigate the implications of the standard practice of fitting autoregressive models in place of moving-average models for the correction of the returns variance; and (iii) we provide a generalisation to the case where the returns process is jointly governed by smoothing and true (economically) time-dependent behaviour. All the theoretical results are illustrated empirically with applications to US real estate and venture capital indexes.
