Is expected utility theory applicable? A revealed preference test

We proffer a method to assess the adequacy of expected utility theory (EUT) in empirical studies involving discrete and continuous choices. The method calibrates a utility function to revealed choices and rejects EUT for absurd degrees of implied concavity over the wealth at risk. We find EUT rejections in cases where risk-averse choices are made between gambles with similar stakes, regardless of their absolute sizes. EUT is shown to be applicable only when expected payoffs of gambles are similar, or when more than half of wealth is at risk. We illustrate our method with an application to agricultural production.