12
Table 3.2 shows transition rates pertaining to the five labour market states mentioned above.
We show averages for the period from 1987 to 2000 as well as two sub-periods: boom (1989-
1990) and slump (1993-1996). Note that the total outflow rate from fixed-term contracts is more
than ten times larger than the outflow rate from permanent employment. Temporary workers
typically end up in non-participation or in permanent jobs. On average for the period, 10 percent
of workers on temporary contracts had made a transition to permanent jobs after one quarter.
By using the data on flows and stocks we can obtain crude estimates of the average duration
of various spells. The exercise reveals that permanent employment is the most persistent state
with an average duration of over 40 quarters (10 years). Self-employment is also quite persistent
with an average spell length of 30 quarters. Temporary employment, by contrast, lasts for only
three quarters on average. Unemployment is the most fluid state with average durations between
1.7 to 2.3 quarters.9
The flow data also reveal that the remarkable rise in temporary employment over the 1990s is
entirely driven by a rise in the inflow to the stock. In fact, the duration of fixed-term employment
spells has shown a trend decline over the period. As was seen in Figure 3.3, a major shift in the
composition of fixed-term contracts over the period was the increase of “on-call” jobs that
typically are of very short duration.
We proceed by looking at the cyclical patterns of transition rates, focusing on permanent and
temporary employment. To emphasize proportional changes we show transition rate diagrams in
natural logarithms with the data seasonally adjusted and smoothed by a 3-quarter centred moving
average. Figure 3.4 displays transition rates between permanent and temporary employment. The
TP transition rates are about 15 times higher than the PT rates. However, the two series exhibit
similar cyclical patterns. From the trough in 1993 and onwards, both transition rates roughly
doubled. By the end of 2000, both rates are close to their peaks of late 1989.
9 We made use of the steady-state relationship, i.e., stock=(flow)×(duration), with the flow measured as the average
of inflows and outflows.