28
probability of lining up a permanent job is substantially lower than it was a decade ago. This
suggests that temporary jobs have become increasingly important as stepping-stones to
19
permanent jobs.
A rise in the share of temporary work has consequences for various labour market outcomes,
including the level of unemployment.20 The effects on flows into unemployment are obvious, at
least in an accounting sense: the higher the share of fixed-term contracts, the larger the inflow to
unemployment. There has been a marked rise in unemployment inflow over the decade.21 In fact,
most of the rise in Swedish unemployment can be attributed to a rise in the inflow rather than a
rise in duration. Roughly 50 percent of the rise in inflow can be accounted for by higher inflow
from temporary jobs. To the extent that there has been a trend rise in temporary work that will
prevail also in favourable macroeconomic conditions, it is conceivable that it has contributed to
some increase in equilibrium unemployment through higher worker separation rates. Of course, a
complete analysis of this issue needs to consider a host of other factors, including effects on
wage bargaining and firms’ recruitment practices.
Appendix
The Firms’ Choices Between Contracts22
Consider a large number of firms that operate in a labour market with frictions and costly recruiting. A
firm can open two types of vacancies corresponding to two types of contracts that are referred to as
temporary (vT) and permanent (vP). Time is continuous, the time horizon infinite and vacancies and
unemployed job seekers (u) meet according to type-specific constant-returns matching functions of the
form H j = (vj)1-η(u)η, j=T, P, where η∈ (0,1) . A firm with a vacancy of type j meets a job searcher at
the Poisson rate qj = H j /vj = (θ j)-η, where θj ≡ vj /u is a measure of labour market tightness.
18 The figures for Denmark are taken from European Commission (2000) and the Finnish data from Kauhanen
(2000). Erling Barth has kindly supplied data for Norway based on the labour force surveys.
19 A recent paper by Hâkansson (2001) looks at this issue in some detail by using longitudinal micro data.
20 See Dolado et al (2001) for a comprehensive discussion of the effects.
21 In 1989-1990, the annual inflow (from employment as well as nonparticipation) measured relative to the labour
force amounted to 5 percent; in 1990-2000, the corresponding figure was 11 percent. These figures are based on
labour force survey data on the number of unemployed with one week of elapsed unemployment and generally
differ from flow figures obtained from the labour force panels.
22 The model of this appendix draws on the analysis in Wasmer (1999).