Researchers often test firm conduct models using pass-through regressions or instrumental variables (IV) methods. The former has limited applicability; the latter relies on potentially irrelevant instruments. We show the falsifiable restriction underlying the IV method generalizes the...