Measuring the Effects of Expectations Shocks

We show that expectation shocks -- revisions in expectations unrelated to changes in current economic fundamentals -- have positive significant effects on US economic activity. To measure the expectation shocks, we estimate a mixed-frequency VAR model that allows economic conditions in the current quarter to affect current-quarter GDP expectations. The expectations shock is estimated with real-time data so such shocks do not suffer a “look-forward” bias by incorporating future data revisions. Dynamic responses are estimated with the aid of a quarterly VAR and using older vintages as instruments to account for measurement errors in the observed values. Expectations shocks explain 10% of the two-year variation of output, investment, consumption and hours. We find that expectations shocks are correlated with alternative belief-based shocks, but nevertheless have significant additional short-run effects.