Missed the starting gun! Wage compression and the rise of the Swedish model in the labour market

ABSTRACT

A central aspect of the Swedish model was the labour market, distinguished by an egalitarian wage structure and by the particular configuration of two institutions: a centralised wage bargaining that followed upon the Saltsjöbad Agreement in 1938 and the solidaristic wage policy implemented in 1956. The literature argues that these institutions produced an outstanding compression of the wage structure from the late 1960s onwards. In contrast, we argue that this narrow post–World War II focus overlooks the historical dimension of the wage structure. The evidence presented here shows that a compression of the wage structure occurred in the late 1930s and 1940s. Previous research attributes this early episode of compression to market factors. In public investigations and periodicals of the 1940s, however, contemporary observers reckoned that special agreements between SAF and LO during World War II caused wage convergence. These agreements anticipated the solidaristic wage policy of the 1950s. We subject the market-factor view to a statistical test and show its explanatory insufficiency. We thereby corroborate the contemporaries’ view and conclude that the coexistence of the centralised agreements, the solidaristic wage policy, and wage convergence configured the rise of the Swedish model during World War II.

KEYWORDS:

• Wage compression
• labour market institutions
• Swedish model
• solidaristic wage policy
• World War II

JEL CLASSIFICATION:

• J31
• J51
• N34
• N64

Acknowledgements

Earlier versions of this paper were presented at the FRESH meeting in Valencia, June 2013; the Swedish Economic History Meeting in Umeå, October 2015; and at the Higher Seminar in Economic History, University of Gothenburg. We would like to acknowledge the suggestions from the participants. In particular, we have received useful advice from Erik Bengtsson, Christer Lundh, Jakob Molinder, Tobias Karlsson, Kristoffer Collin, Yoshihiro Sato, Evelyn Prado (proof reading), three anonymous referees and Editor Alfred Reckendrees. Svante Prado acknowledges financial support from Riksbankens Jubileumsfond for the research project ‘Swedish Wages in Comparative Perspective, 1860–2008’ and both authors acknowledege financial support from Jan Wallanders och Tom Hedelius stiftelse.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 Another cornerstone in the consolidation of the Swedish model was the implementation of the so-called Rehn-Meidner model in the 1950s. It emphasised low inflation, full employment, high economic growth, and income equality.

2 Not much is known about the development of the skilled-unskilled wage ratio among manual workers, but the record suggests that it was stable (Prado & Theodoridis, 2017).

3 Swenson (1991, 2002, 2009) and Lundh (2010) do not analyse the implementation of a collective bargaining system that was centrally coordinated, and neither do they examine the wage convergence during World War II. They focus instead on the path towards the Basic Agreement and on the post-war development.

4 The basis for branch-specific negotiations was established by the agreement in the engineering industry (Verkstadsavtalet) in 1905 and by the centrally negotiated ‘December compromise’ in 1906.

5 For overviews of the official wage statistics, see Bagge, Lundberg, and Svennilson (1933, pp. 247–277); Prado (2010); SOU 1931:20 (pp. 162–170); and SOU 1931:32.

6 The composite measure that comprises all manual workers regardless of skills does not allow the researcher to control for the effects of, for instance, changing skill ratios and the share of skilled and unskilled workers. The strength of the official statistics for historical research on the labour market resides in its extensive coverage of the number of firms and geographical locations, as well as in the large sample of industries. The sample of separate industries ranges from 52 to 70 between 1921, when the annual statistics began, and 1982, when wages by detailed industry classification ceased to exist.

7 http://es.handels.gu.se/avdelningar/avdelningen-for-ekonomisk-historia/historiska-lonedatabasen-hild

8 To corroborate the steep reduction in the spread of wages between 1935 and 1950, we have computed the coefficient of variation across the eight branches that appear in the official wage statistics. For male workers, the coefficients of variation decline by 48.6%; for females, the coefficients of variation decline by 38.4%. Hence, the weighted measure makes male wage dispersion look wider, whereas it makes female wage dispersion narrower. The weighted measure supports the evidence that inter-industry wage differentials dropped significantly from 1935 to 1950. We have also computed the standard deviations of the log of wages, an alternative measure of sigma σ-convergence that sometimes deviates from the coefficients of variation (Dalgaard & Vastrup, 2001). In our case, the two measures give similar results. For example, if we set the two measures to 100 in 1935, the coefficient of variation arrives at 60 and the standard deviation of the log of wages at 62 for male wages.

9 Rehn (1988) argues that the solidaristic wage policy took on a more radical form in the 1970s than it was originally intended. Instead of equal pay for equal work, it subscribed to equal pay for all work. Molinder (2017) shows that in all likelihood the solidaristic wage policy only affected the wage structure after 1968.

10 The agreement also allowed wages in agriculture, considered a low-wage sector, to grow faster than the average of industrial wages (Collin, 2016). The urban-to-rural wage ratio thereby declined precipitously in the 1940s (Lundh & Prado, 2015).

11 One may, however, encounter instances in which measures of σ-convergence and β-convergence tell different stories (Quah, 1993).

12 Space does not permit us to show that the association between growth rates of wages and changes in export ratios is actually negative.

Additional information

Funding

This work was supported by Jan Wallanders och Tom Hedelius Stiftelse samt Tore Browaldhs Stiftelse; Riksbankens Jubileumsfond: [Grant Number P09-0500:1-E].