Wage scales and centralized bargaining – a binding constraint on the wage-setting?

Abstract

The wage-setting for Swedish teachers was up and until 1996 characterized by centralized bargaining and wage scales. According to the scales, wages were exclusively determined by teacher type and years in the profession. In 1996, the scale system was replaced by individual wage bargaining. This study uses this change in bargaining structure to examine whether the scale system was a binding constraint on the wage-setting. The results suggest that the scales did impose such a constraint, since the earnings structure changed after the reform. More precisely, entry-wages and earnings dispersion increased.

Acknowledgements

I would like to thank Peter Fredriksson, Per-Anders Edin, Torbjørn Hægeland, Kjell Salvanes and seminar participants at Uppsala University and Göteborg University for helpful comments and suggestions. The article has been presented at the EEA/ESEM-conference in Vienna. Financial support from the Swedish Council for Working Life and Social Research is gratefully acknowledged.

Notes

¹ The impact of institutions and wage bargaining on the wage structure is theoretically and empirically discussed in Katz and Autor (1999). A discussion of the Swedish bargaining system is found in Elvander and Holmlund (1997). Studies on individual wage bargaining on Swedish data include Lundborg (2005) and Granqvist and Regnér (2004).

² Since 1991 private schools are entitled to municipal funding if they fulfil the requirements of the National Agency for Education, and do not charge tuition fees.

³ The age-earning profile is much steeper than in a comparable sample of public employees.

⁴ Data have been divided by gender, and have been restricted to certified teachers, and public school teachers respectively. In all groups there is an upward shift in 2000 for young teachers. This pattern also holds when studying age coefficients estimated form regressions including a set of standard explanatory variables.

⁵ The total variance explained by the variables in Equation 1(1) falls during the time period, from 41% in 1990, to 37% in 200l. This is in line with the results from wage regressions on Swedish data in Le Grand et al. (2001) and Gustavsson (2006).

Fig. 3. Entrant coefficients estimated from log earnings regressions during the 1990s