What Will Take the Con Out of Econometrics?

The paper begins with the question of whether Leamer's Extreme Bounds Analysis (EBA) really does "Take the Con Out of Econometrics" By analytically demonstrating that the extreme bounds are simply functions of the F-statistic for the deletion of variables from a regression, we conclude that the information provided by EBA represents no advance over that available from traditional methods. Furthermore, there is a degree of arbitrariness in EBA which exactly parallels the selective reporting of regressions it was designed to supplant. The last part of the paper attempts a positive response to its title. By following a well defined series of modelling steps, we maintain that Cooley and Le Roy's EBA-derived conclusions concerning the interest elasticity of money demand owe more to a faulty methodology than to the data.