It is commonplace criticism that econometricians never refute a model that they set out to “test” in the sense that they never reject the underlying theoretical model they have used. It has been argued that refutations cannot be expected from econometrics for various reasons of a logical or philosophical nature. The less kindly, more casual observer might note that econometricians sometimes blame the paucity of data or problems with the estimation method or computing power, rather than reject a theory; in other words that, as bad workers, they blame their tools.

Without wishing to detract from the sound philosophical arguments about the inability to test economic theories, and ignoring the unkind aspersions on econometricians, I want to argue from a different viewpoint–an historical viewpoint. This viewpoint suggests that although econometricians have described their activity as that of testing economic theories, this testing should not be understood in quite the same terms as methodological discussions about falsification and verification. In my view, econometricians have been primarily concerned with finding satisfactory empirical models, not with trying to prove fundamental theories true or untrue.

Historically, econometricians sought applied counterparts to theory that “worked” with reference to observed data. This involved not only the translation of theory into empirical models but also the parallel development of criteria for labeling empirical models “satisfactory.”