22
evidence suggesting that the regulation of fixed-term contracts in collective agreements has
become more lax (SOU 1997:27). As the outcome of both collective bargaining and the
subsequent agreement’s day-to-day implementation at the local level is largely a result of relative
bargaining power, one could expect the mass unemployment of the 1990s to have led to a more
liberal regulation than previously was the case. The possibility for the employer to successfully
press for greater flexibility was probably further enhanced by the 1997 law that permitted the
agreement to be struck at the local level. Thus if legislation played any role we would argue that
the shift to the local level of bargaining is the most likely candidate. Again, this cannot explain
the bulk of the rise in temporary work during the 1990s.
5.2 The Supply Side
Open-ended contracts generally impose costs on the employer but not on the employee. It is
difficult to see reasons why an employee would prefer fixed-term relative to open-ended
contracts, all else equal. Moreover, there is to our knowledge no evidence suggesting that fixed-
term workers are offered wage premiums that compensate for the higher separation risks. On the
contrary, the Swedish evidence at hand indicates that there is a wage penalty associated with
temporary work. Standard Mincer-type wage equations estimated on microdata for the mid-
1980s as well as the mid-1990s suggest a wage penalty of around 10 percent.13 There is also
survey evidence suggesting that working conditions among fixed-term workers are generally
inferior to those experienced by workers with open-ended contracts (Aronsson et al, 2000).
The rise in temporary work is a phenomenon pervasive across broad demographic groups.
Over the 1990s, temporary contracts became increasingly common among all age and gender
stated that many collective agreements prohibited their use; see SKr (1999/2000).
13 We grateful to Altin Vejsiu for estimating these equations. The data used were extracted from the “HUS”
database, a representative sample of Swedish households. A host of human capital and personal characteristics as
well as industry affiliation were controlled for in these estimations. Booth el al (2000) report qualitatively similar
results for the United Kingdom.