RDP 9404: Wage Dispersion and Labour Market Institutions: A Cross Country Study 3. Literature Survey

Much work has been done to evaluate the impact of labour market institutions on economic performance across countries. In a seminal work by Bruno and Sachs (1984) an index of the degree of centralisation of wage bargaining or ‘corporatism’ was constructed which produced a ranking of various countries.[2] Bruno and Sachs found that more corporatist countries (essentially the small non-EC countries of Europe such as Austria and Sweden) performed better in response to the first oil shock. In particular, it was concluded that in countries with near-universal union coverage and highly centralised negotiations, wages were kept closer to market-clearing levels than in more decentralised systems. This conclusion generated controversy and stimulated research by others in the area.

Calmfors and Driffill (1988) defined centralisation as the extent of inter-union and inter-employer cooperation in wage bargaining with the other side. They examined various measures of economic performance and concluded that economic performance was better in those countries which had either low or high levels of centralisation. Countries with moderate degrees of centralisation fared the worst in terms of performance. The reasoning behind this conclusion was that large trade unions inherently recognise their market power and take into account the effect wage increases have on inflation and unemployment. Conversely, unions operating at the individual firm or plant level have limited market power and competitive forces are seen to restrain wages. In intermediate cases where unions exert some market power but ignore the macroeconomic implications of their actions, relatively poor macroeconomic outcomes are observed. The result was a hump-shaped relationship which was related to work done by Olson (1982). Olson's idea was that organised interests may be most harmful when they are strong enough to cause major disruptions but not sufficiently encompassing to bear a significant fraction of the costs to society of their actions taken in their own interest.

Soskice (1990) challenged the conclusions of Calmfors and Driffill on the basis that key countries were wrongly classified. Using an alternative measure of centralisation – the effective degree of coordination – the Calmfors and Driffill result collapsed in a sample of the countries selected. Soskice's approach measured the level at which coordination actually occurred as opposed to the formal location of bargaining. Soskice assigned numerical values to the degrees of economy-wide coordination and the strength of unions at a local (plant) level. Unemployment rates were regressed on these two variables. Soskice concluded that wage bargaining systems with a high degree of coordination are superior to more decentralised systems.

Freeman (1988) did not attempt to construct an index of corporatism given the inherent subjectivity of such an exercise. Instead, he used an index which included dispersion of industry earnings and union density,[3] arguing that high wage dispersion reflects decentralisation. Using this method, he found that both high and low wage dispersion were associated with stronger growth of employment relative to middle ranking levels of dispersion.

A similar pattern was found by Dowrick (1993) using total factor productivity as a measure of performance. An examination of OECD data from 18 different countries over the period 1960 through 1990 revealed a weak inverted hump-shaped relationship between the degree of centralisation and total factor productivity. That is, fully decentralised and fully centralised wage setting systems performed better in terms of productivity growth than partially centralised systems. The relationship, however, was found to be asymmetric, with the performance of fully decentralised systems somewhat better than fully centralised systems.

The major focus of these papers is on macroeconomic performance and the sensitivity of the wage bargaining system thereon. In a 1986 OECD study on flexibility in the labour market, an analysis of microeconomic aspects was presented with reference to different wage setting institutions. Using sectoral data from six OECD countries which encompassed a reasonably wide variety of institutional arrangements concerning labour, a simple model was tested to assess the sensitivity of changes of nominal wages to changes in nominal productivity. A statistically significant relationship was found between the two variables in the United States, Japan and Sweden. As the authors expected, Swedish wages were less responsive than those of the US and Japan to changes in productivity. The elasticity of sectoral wage changes with respect to sectoral value productivity changes was found to be 0.26 in the United States over the period 1958 to 1980, 0.21 in Japan over the period 1970 to 1979 and in Sweden (over the period 1964 to 1983) the elasticity was estimated to be 0.03. The OECD concluded that there was a relationship between changes in wages and productivity in countries with very different collective bargaining arrangements, with elasticities appearing to be higher in countries with more decentralised wage-setting mechanisms and collective bargaining. The consistency of this conclusion is contingent on the interpretation that Japan has a decentralised system.

A recent paper by Calmfors (1993) surveyed the theoretical arguments relating centralisation of wage bargaining systems to macroeconomic performance in a broader manner than his earlier work. The theoretical impact of centralisation on the average wage level, on relative wages, on hiring and investment decisions of firms and on employee effort were among the issues considered. Calmfors concluded that, on the basis of purely theoretical arguments, the diversity of the impact of the various factors renders it difficult to arrive at unambiguous policy conclusions concerning the degree of centralisation of wage bargaining systems. This policy conclusion is consistent with earlier empirical work carried out by Withers, Pitman and Whittingham (1986) who argued that there is not necessarily a relationship between the wage determination system and the degree to which change in relative wages reflect change in labour market conditions.


Following work by Crouch (1985), Bruno and Sachs constructed an index based on the degree of union centralisation, the extent of shop-floor union power, employer coordination, and the presence of works councils (associations which represent non-unionised labourers in a unionised shop). [2]

Union density was defined to be the proportion of the workforce belonging to unions. [3]