Since the development of Brinson, Hood, and Beebower’s single-period performance attribution model, academics and practitioners have tried to expand this classic performance model to a multi-period set-up; e.g., several “smoothing” techniques have been suggested to solve the disconnection between arithmetic attribution effects and portfolio returns, which grow geometrically. This article claims that the research has missed a simple connection
which makes it possible to derive the exact attribution effects at any portfolio level and at any point of time. A solution is derived from an analytical approach which makes it superior to even a linking algorithm with a perfect residual fit. Starting with the classic one-period model, the article shows how multi-period performance attribution effects can be derived analytically and how this new approach, without problem, can be extended to a multilevel analysis as well.

Exact Multi-Period Performance Attribution Model