Allocating the tracking error for the multi-asset-class fund by reconciling bottom-up model with top-down model

In a portfolio with multiple asset classes, each having its own benchmark, an alpha-seeking manager must decide how much tracking error (TE) to incur in each asset class, subject to given TE constraints. This paper helps practitioners clarify how to formulate this TE-allocation problem from a top-down perspective. By reconciling with a conceptual bottom-up formulation, we discover that the validity of the top-down model relies crucially on how one specifies the correlation structure of the asset classes’ alphas. We propose a method for estimating this correlation structure that helps users of the top-down model avoid misallocation of TEs.