I have great pleasure in seconding the vote of thanks to the authors for this extremely interesting paper. One of the important policy problems of our time is that of maintaining a high level of economic activity and employment while avoiding a continual rise in prices. The results obtained in this paper provide valuable information on the possibilities of reconciling these objectives.

There are formidable difficulties in statistical analyses of this sort. A wide range of alternative hypotheses is possible and the series available for testing them are short. The hypotheses used in this paper seem very reasonable, though non-linear forms of the relationship might also have been considered.

The authors have chosen to work mainly with four-period moving averages of quarterly data. If there is anything in the notion of an ‘annual wage round’ a straightforward use of annual figures may have been more appropriate. It is of interest to see from table 5 that the estimates obtained using annual figures for the first quarter of each year are similar to those obtained using quarterly data, and the fit appears to be extremely good.

If we start with an error term which is a four-period moving average of independent errors of autoregressive transformation of the form 1 −¾ E−1 cannot reduce the autocorrelations for all lags to zero. The Durbin-Watson test applies only to the first autocorrelation coefficient whereas the validity of the significance test may therefore not be completely reliable.