Abstract:
It has been an "empirical consensus" that data from developed economies generally do not support the hypothesis of international risk-sharing, either in the form of full risk-pooling via state-contingent assets or in the form of uncovered interest parity enforced by trading non-contingent assets. We reassess these hypotheses in the context of a full DSGE model, as opposed to testing them as single regressions in previous work. We prove that the two model versions behave identically, suggesting that consumers would receive the same scope of protection against risks whether bonds are state-contingent. We further find that the model, when tested appropriately as a whole embracing risk-pooling/UIP, fits the data well and universally through the lens of indirect inference; hence, we provide new evidence of the hypotheses' empirical validity spuriously rejected by single regressions.
