Risk adjustment is a fundamental element of modern portfolio theory. Alpha is defined as the return left over after the impact of all sources of risk, including strategic biases, has been accounted for (see the examples in my series on alpha). But a positive historical alpha alone is not sufficient to conclude that a particular product would have been a desirable holding. Yet I frequently come across managers of multi-asset portfolios who proudly proclaim that they have generated superior risk adjusted returns. What is generally left unsaid is that, on a raw return basis, they actually underperformed their benchmark. This usually happens when risky assets offered a positive risk premium and the manager was underweight those risky assets, but was good at picking securities within the asset classes. Should existing clients consider the superior risk-adjusted performance as a consolation in this case? Should we invest with such a manager? I would argue that “no” is the correct answer to both questions.
As far as existing clients are concerned, they gave the manager a mandate with a benchmark and typically some limits on the acceptable tactical bets. After the fact, clients will always prefer better performance regardless of how it was achieved, as long as the rules of the mandate were respected. To put this in terms of my earlier discussion of alpha, active performance is the sum of strategic biases and tactical bets. If a manager is too timid to exploit the opportunities that a mandate offers and remains strategically short the higher risk assets in his multi-asset portfolio, this fact should be held against him. While I would never consider a positive contribution of a strategic bias to be skill-based alpha, I am quite willing to consider the negative impact of such a bias as negative alpha; in this particular case, if in doubt, against the accused.
One consequence of this view is that we should ideally pay performance fees only on the lesser of skill-based alpha or skill-based alpha minus negative impact of strategic bias.
What role should the historical risk-adjusted performance play in manager selection? Historical performance numbers are relevant only to the extent that they allow us to extract insights about the likely future performance of a manager. But managers who systematically underweight the riskier assets in a multi-asset portfolio are either excessively timid or not very smart. Neither is a very desirable characteristic of an active manager. As discussed in an earlier post, some strategic biases make sense; the most obvious example is an overweight in the riskier class of a two asset class portfolio. The expected long-term return added by this bias is positive. Even if this is not alpha, the client will still benefit. Managers should understand this; it is after all very much in their self-interest.
To conclude, the appeal by a manager to good risk-adjusted return numbers is usually an excuse and a poor one at that. Timidity is a highly undesirable quality in active asset managers. I suspect that more money has been lost with pseudo-active, low risk strategies where even great skill would have been insufficient to overcome the fee drag than with higher risk strategies that could, if successful, add real value for the client. Thus, a manager who focuses on risk-adjusted returns is unlikely to be an attractive candidate for an active multi-asset mandate. Let us not forget that we do not consume risk-adjusted returns – nor Sharpe ratios and the like – but simple raw returns.