Labour Markets
Measuring the union mark-up

Assessing union power is difficult. One approach is to measure the percentage differential of union over non-union wage rates, the union mark-up (UMU). Layard, Metcalf and Nickell have estimated this mark-up using cross-section data for UK manufacturing industry over the period 1951-83. Their UMU measure is derived from a regression of the hourly earnings of male manual workers in each sector on variables such as the proportion of skilled workers, the age structure of workers, and the proportion of workers covered by a negotiated collective agreement. The coefficient of the last explanatory variable, 'coverage', measures the UMU. Excluding short-term fluctuations, this measure rose between 1950 and 1980, with particularly large increases during 1967-72 and 1979-80. The mark-up remained at a high level between 1980 and 1983, the last year for which data are available.

In Discussion Paper No. 158, Research Fellow Michael Beenstock and Chris Whitbread use Layard, Metcalf and Nickell's data to model the behaviour of the mark-up over time and to test three hypotheses concerning the determinants of union power. It has been argued that if unions set wages and firms determine employment, then higher unemployment benefit levels increase the union's desired wage. In this model, union members know that higher wages are achieved only at the cost of lower employment, which entails a greater risk of becoming unemployed. Higher benefit levels enable them to accept a higher risk of unemployment. A second hypothesis is that the relative bargaining power of union and non-union workers varies with the level of unemployment. Higher unemployment is associated with a higher UMU because it reduces the strength of non-union workers' threats to quit by more than it reduces the strength of unions' threats to strike. Finally, more widespread union membership (union density) may reduce scabbing and provide unions with extra bargaining power.

Beenstock and Whitbread estimate an equation in which the union mark-up depends on the real benefit rate, the unemployment rate, the inflation rate, dummy variables representing incomes policies and union density, and a variable representing the party in power. Their estimates support the first and second hypotheses described above: between 1953 and 1983 the doubling of the real Supplementary Benefit rate added about 17 percentage points (or 200%) to the mark-up, while the rise in the unemployment rate from 1% to 6% added about 9 percentage points (100%). The model containing union density, however, is unable to explain the increase in the mark-up after 1980. The authors suggest that the model should overpredict the mark-up during the Thatcher government; instead, it underestimates it. The measure of union density may not be appropriate, however, since it is for the entire economy, whereas the mark-up variable covers only manufacturing. Curiously, the variable representing incomes policy adds little explanatory power. Unions are apparently no better at fending for themselves during periods of incomes policy than are non-union workers. Finally, the mark-up is 5 percentage points higher under Conservative governments than under Labour.

Beenstock and Whitbread note the imperfections of the mark-up measure, but the influence of benefits on the mark-up is, they argue, the most robust of their results. The ability of their equation to predict the UMU between 1980 and 1983 suggests that there is no evidence yet of a change in union power, at least as manifested by the mark-up.

Explaining Changes in the Union Mark-up for Male Manual Workers in Great Britain, 1953-83
Michael Beenstock and Chris Whitbread

Discussion Paper No. 158, March 1987 (IM/ATE)