Money Wage Inflation in Industrial Countries: An Alternative Explanation

Abstract

Eatwell, Llewellyn and Tarling [1] present what they regard as empirical support for the key industry hypothesis of the determination of money wage inflation. A 15 country cross-section for the period 1958-67 is used to test the theory that the rate of wage change in a country's manufacturing sector is dependent on productivity growth in a key sector of the three industries with the highest rate of productivity growth. However, it is not at all clear that the key industries are the pace setters as far as earnings growth is concerned. For Sweden, for example, the key industries include Petroleum and Chemicals, yet the rate of earnings growth in these industries is only 7l57-almost the lowest of the growth rates for Swedish industries. For Japan, earnings growth in the key industries is less than in manufacturing industry as a whole and for one key industry-Basic Metals-is only 7l86 compared with the overall manufacturing average of 9l88. Overall in the 15 countries considered by Eatwell et al., eight have key industries with earnings growth below the manufacturing average. This being so, it is difficult to regard these industries as being pace setters for the rest of manufacturing industry. These findings suggest that the high correlation found by Eatwell et al. between W and Z3 is misleading and that alternative explanations of the data should be considered. In this note it is shown that variations in cross-country rates of wage inflation can be satisfactorily explained by variations in such traditional Phillips curve variables as the level of unemployment and the extent of unionization. A serious problem in any cross-country study of unemployment rates is that of finding comparable data. European figures are usually based on unemployment registrations while US data, for example, are based on sample surveys which generally lead to higher recorded unemployment series. Maddison ([2, p. 220]), however, provides a consistent set of data for the period 1950-60 which are adjusted to a labour force sample survey basis. The data cover nine of the 15 countries considered by Eatwell et al.-Belgium, Canada, Denmark, France, Germany, Netherlands, Sweden, UK and US. Figures for the period 1958-67 were obtained by applying to the official 1958-67 series the ratio between the 1950-60 Maddison series and 1950-60 official series. Three other countries-Finland, Norway and Japan-publish official series which are based on labour force sample surveys. Thus comparable data is available for 12 of the 15 countries in Eatwell et al.'s cross-section. For these 12 countries average earnings growth was regressed on the reciprocal of U, the percentage of the labour force unemployed, with the following results: