Content area
Full Text
Maintaining it can be suboptimal.
(ProQuest Information and Learning: ... denotes formulae omitted.)
To manage a portfolio against an explicitly specified benchmark, there are good reasons to aim at benchmark neutrality, in the sense of having a target of unity for the portfolio's beta with respect to the benchmark. For instance, if equities are only one portion of a portfolio, a particular equity benchmark is likely to be chosen according to its perceived volatility and correlations with other asset classes. In this case, deviation of the active equity portfolio from a unit beta exposure to the equity benchmark may change the risk profile of the whole portfolio.
Yet implementation and verification of benchmark neutrality is far from easy. Ex post return-based analysis of benchmark neutrality relies on estimation of the average beta exposure to the benchmark during the sample period, which is subject to the dynamic beta problem identified by Grinblatt and Titman [1989]. Ex ante benchmark neutrality depends on the quality of the managers' covariance matrix. While they may be able to remain benchmark-neutral according to estimated betas, there is no guarantee that the ex post beta of the resulting portfolio with respect to the benchmark will be at one.
The issue becomes more complex if one moves beyond one factor to the multifactor world. For quantitative portfolio managers who tilt portfolios toward factors that are predicted to deliver better returns, factor neutrality is clearly suboptimal. In a sense, factor neutrality unwinds most, if not all, of the information advantage gained by such managers.
Also questionable is the practice of factor neutrality by portfolio managers who claim to be focused on stock fundamentals but who do not rely on a purely quantitative approach. Apparently they define stock-picking as predicting the idiosyncratic return components in the total stock returns that are independent of all factors.
It is true that a natural extension of the one-factor framework into a multifactor framework would be to risk-adjust alpha, so that only the return component of the portfolio that is independent of all factors should be perceived as adding value. In this framework, returns resulting from increased exposure to factors would not be viewed as due to investment skill.1
While we are somewhat sympathetic to this view, we...