Comment on: “Estimating the expected marginal rate of substitution”

The methodology proposed in Flood and Rose [2005. Estimating the expected marginal rate of substitution: a systematic exploration of idiosyncratic risk. Journal of Monetary Economics 52 (5) 951-969] fails to distinguish between the single unique marginal rate of substitution (MRS) process and the class of valid pricing kernels, of which the MRS is but a particular member. Thus, at best, this methodology explores the properties of some arbitrary pricing kernel, which may differ radically from the true MRS. Furthermore, the estimates of the expected MRS proposed by Flood and Rose [2005. Estimating the expected marginal rate of substitution: a systematic exploration of idiosyncratic risk. Journal of Monetary Economics 52 (5) 951-969] are highly correlated with ex post shocks, implying that these estimates are not conditional expectations at all. The cure for this misspecification introduces additional econometric problems, suggesting that the model may, in practice, be poorly identified.