General Discussion of Perry and of Cagan-Fellner

Robert J. Gordon noted that the main difference between the pricewage system implicit in his price equation that George Perry reported and the wage equations in the paper by Perry and the one by Phillip Cagan and William Fellner was the elasticity of past inflation on current inflation. He argued that by hypothesizing norm shifts—or in the case of the Cagan-Fellner paper, a succession of cyclical shift dummies-the Perry model estimates a coefficient on past inflation that is too low. That is why the Gordon price equations track the present disinflation well while the Perry equations start to develop large overpredictions of wage inflation. Perry replied that the residuals in the present period did not tell anything about which formulation was superior because the wage norm equations specifically expect residuals of the kind that are being observed. Those equations would be failing if there were not positive residuals in the present period. The choice between fitting long lags on past inflation or allowing for norm shifts to characterize the inflation process had to be settled on other grounds, such as the evidence he had discussed in his previous paper (BPEA 1:1980).