As regards labour market flows, a high incidence of temporary contracts has a large positive impact on the magnitude of labour market dynamics. Using a sample of large Spanish firms in the period 1993-1994, Garcia-Serrano (1998) found that the size of both job(41) and labour turnover rates varied strongly by type of employment contract, being much higher for temporary work. The analysis suggests that a rise of 1% in the share of temporary employment increases flows from employment to unemployment by 0.26, flows from unemployment to employment by 0.16, and flows between jobs by 0.34. Another interesting difference emerges, namely that temporary contracts are much more likely than permanent ones to be associated with job-to-job flows (employment rotation) than with the creation or destruction of jobs.
As regards the impact of temporary contracts on the incidence of longterm unemployment (LTU) and the duration of unemployment, Bentolila et al. (2008) found that during an economic expansion phase, LTU tended to decrease in Spain, while in slowdown/recession periods there were episodes of LTU increases, despite a persistently high share of temporary employment. Following the labour market reforms introduced in Spain in the late 1990s-early 2000s aimed at reducing the EPL gap, LTU decreased (Dolado et al., 2002).
Guell (2006) found that the 1984 liberalisation of temporary contracts in Spain increased the duration dependence of unemployment – in other words, the probability of exiting unemployment declines with the duration of unemployment. The intuitive explanation is that temporary workers go through recurrent short unemployment spells given the low conversion rates of temporary contracts into permanent ones and so systematically push the long-term unemployed further back along the hiring ‘queue’, decreasing their probability of exiting unemployment.
Chart 28 suggests the existence of a negative correlation between the incidence of temporary work and the share of long-term unemployed in EU Member States. However, such evidence is quite weak as the correlation is statistically significant only for workers in the age bracket 15 to 39.
(41) | See footnote 5. |