Dallas S. Batten and Daniel L. Thornton point out that, since it was first used in 1968 to investigate the relative importance of monetary and fiscal actions in influencing economic activity, the St. Louis equation has been subject to much criticism. One criticism is that the policy conclusions may be dependent on the equation’s econometric specification and, in particular, on the use of Almon’s polynomial distributed lag estimation technique. To investigate the seriousness of such criticism, Batten and Thornton conduct a general investigation of the sensitivity of the St. Louis equation’s policy conclusions to the choice of lag structure and polynomial degree. Using a variety of model selection criteria, they find that the policy conclusions are virtually insensitive to alternative lag structures or polynomial specifications. In every case examined, they could not reject the hypothesis that a one percentage-point increase in money growth leads ultimately to a one percentage-point increase in the rate of growth of nominal GNP. Moreover, the hypothesis that high-employment government expenditures had no long-run impact on nominal GNP growth was rejected only when contemporaneous government spending growth alone was included in the model; however, this specification was rejected when tested against longer-lag alternatives for government spending.