This paper develops an empirical strategy for testing competing hypotheses of expectation regimes when direct measures of expectations are unavailable. The procedure takes as given an assumed structural relationship between expected values of exogenous variables and a given decision variable. By imposing different expectation regimes on this model, we obtain an artificial nesting of the hypothesized regimes which allows us to test whether any specification dominates. This methodology is extended to multiple equation applications with any number of hypothesized expectation regimes. The tests are illustrated using a model of the response of county-level farm acreage allocation to expected commodity prices.