Labour Productivity in the South Wales Coal Industry: Reply

Abstract

In our earlier article in this Journal, we estimated several empirical models attempting to explain the determination of labour productivity in the British and, separately, South Wales coalfields between 1874 and 1914. A major finding for South Wales was a significant negative relationship between average labour productivity and both a nominal and real-wage rate index, thus providing support for the contention of economic historians (and some contemporary observers) that a significant reduction of work effort occurred in response to rising wage rates. Professor Greasley is sceptical of these findings and offers several criticisms of our data and interpretation.' Indeed, he claims that our principal result is without statistical foundation and that we were led to reach conclusions that are implausible. We disagree strongly with these contentions. While some of his comments may warrant an airing, we find little in his paper that alters our original conclusions. Greasley makes three potentially relevant points: (1) there exist measurement problems with the labour productivity and wage variables used in our paper (which were taken from Walters, 1975); (2) our empirical results rely, Greasley contends, on the assumption of money illusion by miners; and (3) because wages are closely tied to coal prices, causation may run from productivity to prices to wages and not, as we modelled, from prices to wages to productivity. We address each of these points in turn. Following Walters, we used data on overall labour productivity even though much of our discussion concerned the response of hewers to rising wages. It is clear that hewers comprised the primary class of mine workers who could vary their intensity of effort in response to changing wage rates.2 Thus, Greasley, in his Figure 1, constructs a separate productivity series for hewers for 18741914. Rather remarkably, this series is based on interpolation and extrapolation from three data points (1889, 1905 and 1914).3 Yet even using his new series as a measure of productivity, Greasely finds a negative relationship between productivity and wages equal in magnitude and statistical significance to that found in our paper, thus reinforcing our original findings. Greasley also criticizes us for using Walters's wage rate index, although he offers no substitute measure for this time period. He is furthermore concerned by the high correlation of the wage index with the price of coal. Yet we modelled the relationship between coal price, wages and labour productivity explicitly in our paper (as both a recursive system and as a system of equations estimated by FIML). Hence, this criticism is not valid unless one has a strong prior that miners responded to coal price independent of their compensation (a view with which we do not sympathize), or that causation runs in the opposite direction (a point of potential interest on which we comment below). While Greasley provides no substitute for Walters's wage index in his comment, he has elsewhere (Greasley, 1982) provided an alternative wage