The evolution of the Swedish wage structure: new evidence for 1992–2001

Abstract

Changes in the Swedish wage structure from 1992 to 2001 are investigated and placed in a longer-term perspective. The results show that the increase in overall wage dispersion that began in the early 1980s have continued during the 1990s, and that the increase during the 1990s is driven by increased wage differentials in both the lower and upper half of the wage distribution. The results also show that the university wage premium in Sweden has exhibited a clearly positive trend since the early 1980s whilst the returns to lower levels of education have been constant since the mid 1970s. Unlike the 1980s, relative supply changes do not appear to have been responsible for the rise in the university wage premium over the most recent decade.

Acknowledgements

I am grateful to Per-Anders Edin, Meredith Beechey, Patrik Hesselius, Bertil Holmlund, Stephen Jenkins, and Markus Jäntti. Financial support from the Jan Wallander and Tom Hedelius Foundation is gratefully acknowledged.

Notes

¹ Only one previous study has, as far as the author knows, touched on changes in the Swedish wage structure during the 1990s. Based on the Swedish Level of Living Surveys (LNU), le Grand et al. (2001) estimate wage equations for 1991 and 2000. However, they rely upon a small sample and do not report standard errors, making it difficult to judge the precision of their estimates.

² The month for which wages are collected varies between the years, but it has often been in April, May, September or October, and never during June, July, or August.

³ The 1992–1997 data for the private sector is not fully representative as individuals working in firms with less than ten employees are underrepresented. This could potentially affect the magnitude of the estimated wage differentials, but there is no reason to expect it to affect changes over time, which is the focus of this paper. Furthermore, Gustavsson (2004) shows that in terms of changes in annual earnings differentials between 1992 and 1997, firms with less than ten employees do not differ to other firms.

⁴ The original numbers for 1992–2001 are found in Table 1.

⁵ See also Palme and Wright (1998) for estimates of the returns to education in Sweden between 1968 and 1991.

⁶ Because of this, wage equations based on samples where all immigrants are excluded have also been estimated. The obtained point estimates are very similar to those in Table 2.

⁷ Technically, there is no single standard error of the residual. Because weights are used, there is heteroscedasticity in the residuals, and hence, the standard error of the residual differs between individuals.

⁸  Table 1 shows that 41% of the individuals in the 2001 sample are employed in the public sector. Of these, 69% are women.

⁹ The same conclusion is also reached if the wage differential between some university and 3-year gymnasium is analysed in the same way.