11
Available survey evidence indicates substantial heterogeneity among fixed-term workers
with regard to self-reported job satisfaction (Aronsson et al, 2000). On-call contracts are
associated with “bad” working conditions along a number of dimensions whereas project work,
and, in particular, probationary contracts are perceived as more attractive. It is noteworthy that
women account for some 65 percent of all on-call contracts whereas they represent less than 40
percent of the project workers.
3.3 Temporary Work and Labour Market Transitions
By using data on gross labour flows we can shed light on the dynamics of temporary work.
Respondents in the monthly labour force surveys stay in the survey for two years and are asked
every third month about their current labour market status. The design of the survey is such that
7/8 of the sample in month t will remain in the sample at month t+3. The data are reported as
quarterly flows, i.e., averages of the flows computed from three pairs of panels. The tabulations
are available since 1987. The panel records transitions between several labour force states but the
data at our disposal include no information on job-to-job transitions, unless the transitions
involved a change of contract (from temporary to permanent or vice versa).8
Let U denote the number (stock) of unemployed and O the number of nonparticipants (out-
of-the labour force). Employment can be disaggregated into three states: permanent contracts
(P), temporary contracts (T), and self-employment (S). For the flows the notation is ZX for flows
from state Z to state X. For example, TU is the flow from T to U. A transition rate is defined as
the ratio between a particular flow and the stock that represents the relevant population at risk. In
general, the transition rate from Z to X is denoted ZXR and is defined as ZXR=ZX/Z; for example,
the transition rate from T to U is given by TUR=TU/T.
8 We have access to flow data for the period up to the second quarter of 2000. There are well-known problems with
flow data that have been discussed by, for example, Abowd and Zellner (1985) and Blanchard and Diamond (1990).
One problem is misclassification; another is lack of information on multiple transitions between the interviews.