The significance of testing empirical non-nested models

In the three decades since statistical procedures were first developed to test non-nested (or separate) families of hypotheses, a burgeoning literature has developed for testing a null model against one or more non-nested alternatives. The purpose of the paper is to evaluate the significance of testing empirical non-nested models, with special emphasis on whether the theoretical literature has had a significant impact on practitioners, and on whether the empirical output has added significantly to our knowledge of various aspects of economics.