The Ambiguity Brake

Portfolio managers with performance-based incentives face a nonconcave payoff: below a benchmark, their fee is zero; above it, their fee is a convex function of outperformance. Standard portfolio theory assumes concave utility — diminishing returns to wealth — which produces moderate risk-taking. Nonconcave payoffs break this: the convex region creates incentives for aggressive bets, because the upside of exceeding the benchmark by a lot is worth more than proportionally.

Under smooth ambiguity — where the manager is uncertain not just about returns but about which probability model is correct — the aggressive risk-taking is dampened. Ambiguity aversion shifts beliefs toward adverse states, limits the range of outcomes that would trigger more aggressive positioning, and reduces volatility through lower risky exposure.

The framework recasts the ambiguity-averse problem as an ambiguity-neutral one with adjusted probability distributions. The mathematical trick: instead of solving a complex decision problem under uncertainty about models, solve a simpler problem under a pessimistically weighted mixture of models. The ambiguity aversion is absorbed into the probability measure.

The through-claim: ambiguity aversion is a natural brake on the risk-amplifying effects of convex incentives. Convex fees make managers take risks that concave preferences would prevent. Model uncertainty makes managers take fewer risks than their preferences alone would dictate. The two forces oppose each other, and the net effect depends on which is stronger. In delegated portfolio management, the principal’s problem is not just designing the fee structure — it’s understanding how the manager’s epistemic uncertainty about the market interacts with the fee’s risk incentives.