Comment

This paper is motivated by Backus and Smith (1993). They show under fairly general conditions that in a world with state by state complete risk sharing, realized real exchange rate changes are perfectly correlated with a linear combination of realized aggregate consumption changes in the home and foreign countries.!rert = sigma ⋅!ct − sigma * ⋅!ct*. This holds regardless of nontraded goods, violations of law of one price, nominal rigidities, and so forth. A key point is that faster growth in home consumption should be perfectly correlated with real deprecia-tion of home exchange rate. Here the sigmas are intertemporal elastici-ties of substitution home and abroad. This is a very powerful way to test one implication of complete risk sharing across countries. I wish I had thought of it. But, as is well known, when this equation is tested in the data, it is almost always rejected (Backus and Smith 1993; Kollmann 1995; Kocherlakota and Pistaferri 2007; and Hess and Shin 2010, among oth