Box 5 provides some evidence suggesting that a larger share of temporary employment does tend to be associated with higher cyclical volatility of total paid employment in a number of EU Member States.
Bentolila (2010) considers that in Europe there are three major modes for the adjustment of labour markets to cyclical fluctuations, predominantly via: i) temporary employment (e.g. Spain); ii) wages (e.g. UK); and iii) working hours (e.g. Germany).
This box discusses some aspects related to the role of temporary contracts in adjusting labour demand. The scope for using temporary contracts during a downturn/recession obviously depends on their initial use (Table 12), and therefore varies considerably across Member States.
On average, across Europe during the recent economic recession, numbers of temporary jobs declined by more than numbers of permanent ones. Charts for Germany, Spain, and the UK show the estimated cyclical fluctuations of temporary and permanent employment, and GDP, using the Hodrick-Prescott (HP) filter.
In Spain, the brunt of the adjustment of employment levels to labour demand (both during expansions and contractions) is borne by temporary workers (Chart 36). Conversely, Charts 35 and 37 suggest that the role of temporary contracts is much less prominent in Germany and the UK, respectively.
The rest of this box provides a tentative answer to the following question: Is a higher incidence of temporary jobs (over the cycle) associated with stronger cyclical fluctuations for total paid employment?
Okun’s law(1) is used to estimate a pooled regression for the cyclical component of total paid employment. Eurostat’s annual data are used for the estimate. Cyclical values are calculated using the HP filter (with λ=6.5).
The estimated regression is
where
i: Country
t: Year
yit: Cyclical component of total paid employment
X1: Year dummy
X2: Country dummy
X3: Cyclical component of GDP
D: Sign of the GDP gap dummy {0: negative output gap i.e. X3<0; 1: positive output gap i.e. X3>0}
yit and X3 are expressed in logarithms
In equation 1, the cyclical component of total paid employment is regressed on the current and lagged one period cyclical component of GDP, allowing for different slopes according to the economic cycle, and controlling for country and time-fixed effects. When the output gap is positive (negative) (i.e. X3>0 or X3<0), current year effects are given by, respectively, b3*(1+b4) and b3.
(1) | Okun’s law ( reflects the idea that over the business cycle additional production of goods and services requires more employed workers (EiE, 2002, chapter 2). |
Sala and Silva (2009) developed a matching model in order to reproduce the cyclical behaviour of the Spanish labour market and found that the gap in firing costs and productivity between permanent and temporary employees plays a large role in explaining the high rate of employment volatility in Spain, where temporary contracts are the main tool for adjusting labour demand.
Box 5 provides some evidence suggesting that a larger share of temporary employment does tend to be associated with higher cyclical volatility of total paid employment in a number of EU Member States.
Table 13 presents some results, namely the estimated elasticities of the cyclical variation in paid employment to the GDP gap, and countryfixed effects. The latter can be interpreted as country dummies. Results suggest significant lags in the adjustment and a marked asymmetry depending on the sign of the output gap, with a negative gap having a much stronger impact on employment destruction (in absolute terms) than a positive output gap has on employment creation (i.e. a cyclical contraction of 1pp in GDP changes paid employment in absolute terms by more than an expansion of the same magnitude).
This regression tracks reasonably well actual developments (Chart 38).
In order to assess whether a higher incidence rate of temporary jobs is likely to increase the cyclical volatility of total paid employment, country dummies are correlated with the average incidence of temporary jobs in the period covered by the regression (Chart 39).
Chart 39 suggests that the incidence of temporary employment does increase the cyclical volatility of total paid employment i.e. the correlation is significant at 5%.(2)
(2) | Six countries were considered as outliers and therefore excluded from the calculation of this correlation, namely Estonia, Latvia, Lithuania, Luxembourg, Malta and Poland. Apart from Poland, all these Member States can be characterised as small open economies. Poland’s position as an outlier might reflect the exceptional labour market conditions at the turn of the century, when unemployment rates stayed close to 20% for a number of years. |