Are asset managers properly using tracking error estimates?

Abstract

In the investment community, tracking error is often regarded as the single most important risk measure, and many active asset management companies use tracking error as a basis for keeping a portfolio's relative risk under control. Recently, the investment community has become aware of some shortfalls of tracking error. One of the main pitfalls is that ex-ante and ex-post tracking errors differ, since portfolio weights are stochastic and market drivers have changing distributions. This conclusion can be extended to other risk indicators, eg relative value at risk.

This paper shows that, if portfolio managers follow a risk policy and on a regular basis they keep their portfolio relative risk in a given range, ex-ante tracking error does not differ in a statistical sense from ex-post tracking error, and the stochastic nature of portfolio weights becomes almost irrelevant. Therefore, if the goal is to avoid wrong measurement of tracking error and subsequent troubles with clients, the main focus of active asset management companies should be on two points: (i) forecasting risk correctly; (ii) implementing a risk policy on a regular basis.