Labour flows in the EU at the beginning of the crisis

One of the most significant aspects of the current crisis is its varying impact on the labour market variables of different EU countries. For instance, in some, during the initial phase of the crisis the unemployment rate scarcely rose and, in fact, in Germany it decreased by approximately 1 pp. Four years after the crisis began, the unemployment rate in Germany is at a record low (of 5.8%), while in other countries – especially Spain and Ireland – it has reached similar or even higher levels than those recorded in the first half of the 1990s (see Chart 1). Understanding the reasons behind the different impact of the crisis on the labour market is crucial for designing policies which encourage job creation and accelerate the recovery. An increase in the unemployment rate may reflect higher job destruction, lower job creation, an increase in the labour force (due, for example, to the inflow into the unemployment pool of new population groups which were previously inactive in the labour market) or various combinations of these three basic driving forces. The type of shock received by an economy, the employment policies implemented in response to it and labour market institutional arrangements determine variations in these three underlying variables of the unemployment rate. Consequently, a mere international comparison of aggregate unemployment rates will generally offer an incomplete picture that will not analyse in enough detail all the different factors behind the change in unemployment in a given country, nor will it analyse the contribution of employment policies and labour market institutional arrangements to said change. Given the divergent performance of the unemployment rate in EU labour markets at the beginning of the crisis, it is reasonable to anticipate that the contribution to rising unemployment of higher job destruction, lower job creation and a fluctuating labour force has also been very mixed. In order to analyse the reasons for these differences, it is necessary to analyse labour flows and their demographic breakdown, i.e. the characteristics of the individuals comprising such flows. This type of analysis may also identify better the role played by employment policies and labour market institutional arrangements. For example, the impact of short-time working schemes, such as those implemented in Germany and Austria, among other countries, should affect transitions in the labour market from employment to unemployment. Similarly, changes in unemployment due to the “discouragement effect” (the unemployed who give up looking for work) or the “additional worker effect” (an individual who becomes active in response to another member of the family unit becoming unemployed) are observed more directly when studying flows between employment/unemployment and inactivity. This article analyses the variation in flows of workers between statuses of employment, unemployment and inactivity at the onset of the crisis, in order to identify some of the factors which may underpin the cross-country differences observed in the increase of unemployment during that period. For this purpose, using the micro data of the European Labour Force Survey published annually by Eurostat, the size of flows of workers between the three possible labour market states (employment, unemployment and inactivity) is quantified as well as its contribution to the rise in the unemployment rate at the onset of the crisis. The advantage of using this database is that a relatively homogeneous comparison can be made both in terms of definition of the variables and data collection between various EU countries. Thus, in this analysis the results obtained for 11 EU countries are compared: Austria, Belgium, Germany, Denmark, Spain, France, Greece, Italy, Portugal, Sweden and the United Kingdom. Furthermore, the micro breakdown of the survey not only makes it possible to quantify changes in the flows, as has been done in other studies, but also to identify the demographic characteristics of their constituent individuals. By contrast, Eurostat makes these data available with a slight lag which limits the time span for performing this analysis. A further caveat worth pointing out is that the results are influenced by the annual frequency of the data used, since with this sample frequency it is not possible to observe labour transitions which occur in a period of less than 12 months. These transitions may be important for certain population groups and, especially, in countries where the labour market institutional arrangements favour high labour turnover. Thus, the analysis presented below generally provides a lower level for the probabilities of moving between the three relevant labour market states: employment, unemployment and activity/inactivity.