Labour market institutions and the industry wage distribution

The paper studies the industry wage structures of Austria, Norway, the union sector of the U. S. as well as the non-union sector of the U. S. We make comparable regressions for each country, and are thus able to compare the sectoral earnings patterns controlling for the usual individual characteristics. Our results confirm the hypothesis that the pattern of the inter-industry pay structure is largely independent of labour market institutions: High paying industries in a non-union environment tend to pay high wages also in regimes where bargaining is very centralised and coordinated.This, however, does not mean that collective bargaining does not matter. The influence is mainly on the amount of wage dispersion: We find considerably lower industry pay gaps in centralised Austria and Norway than in decentralised U. S. Within the U. S., pay differentials within the union sector slightly exceed those of the non-union sector.The results give support to non-competitive explanations of the labour market. If efficiency wage mechanisms are the reason for wage differentials we expect central bargainers to internalise these effects. Competitive explanations, on the other hand, would predict no difference between the non-union outcome and a central agreement aiming at achieving full employment.This work was conducted while we were both affiliated with the University of California at Berkeley, and we thank the Institute of Industrial Relations at the University of California, Berkeley, for its support and hospitality. The research was supported by the Austrian Fonds zur Förderung der wissenschaftlichen Forschung under the project JO548-SOZ (Zweimüller) and the Norwegian NORAS under the LOS program (Barth). A preliminary version of the paper was presented at the Labour Seminar at the University of California, Berkeley. We thank the participants, especially Bill Dickens and Jonathan Leonard for valuable comments. We are indebted to Bill Dickens also for giving us access to the U. S. data set CPS 1983. Thanks also to Herbert Walther for useful comments.