Reply: “Safety First – An Expected Utility Principle”

Abstract

We are grateful for the opportunity which Professors Gressis and Remaley's Comment [1] has afforded us to clarify our analysis of the relationship between Roy's Safety First principle and the Mean-Variance expected utility rule, which unfortunately they find misleading. In our original presentation we did not explicitly analyze situations in which Roy's criterion leads to an extreme corner solution; and G-R are perfectly correct in noting that if a riskless asset exists, a “Safety-Firster” will not invest in risky securities at all if the risk-free interest rate, r, exceeds the disaster level d. The reason for this is straightforward: such an investment strategy minimizes the risk of disaster. In fact in this particular instance the investor can reduce the probability of disaster to zero.