مدت زمان بیکاری و تعاملات میان بیمه بیکاری و کمک های اجتماعی

Existing studies suggest that reforms that reduce the generosity of the unemployment benefits should lower unemployment. Despite the large number of such reforms implemented in Europe in the past decades, evidence from various data sources shows very little correlation with the evolution of unemployment. This paper suggests that the scant success of these labour market reforms can be explained by the interactions between unemployment insurance and other social assistance programmes. Evidence from the European Community Household Panel shows that recipients of unemployment insurance who are also eligible for other welfare schemes are indeed less sensitive to changes in the level and the duration of their benefits.

Many studies have already explored the effects of unemployment insurance (UI) on the duration of unemployment (Atkinson and Micklewright, 1991). Two empirical findings are now widely accepted. First, as initially showed by Nickell (1979) and Lancaster (1979), higher benefits are associated with longer unemployment spells. Later, Moffitt (1985) and Meyer (1990), having access to information about both the level and the duration of benefit entitlement at the individual level, showed that the probability of exiting unemployment increases around the time of benefit exhaustion. The literature also provides a comprehensive theoretical framework for interpreting these results. Mortensen (1977) develops a simple search model that easily delivers negative correlation between exit rates and unemployment benefits via search effort and reservation wages being respectively negatively and positively affected by income out of work. When UI entitlement expires, income out of work drops, inducing an increase in search effort and a decrease in the reservation wage, thus providing an explanation for the observed higher exits rates around the time of exhaustion. Cahuc and Lehmann (2000) recently extended the model by endogenising wages and allowing the unemployment benefit to gradually decline over time. On the basis of these findings, one would expect reforms that reduce either the level or the duration of unemployment benefits to have a positive impact on the unemployment rate. However, this prediction does not seem to conform with the recent experience of many European countries. Fig. 1 plots the time series of the unemployment rate for selected European countries. The vertical lines indicate the implementation years of reforms that have modified either the level or the duration of the unemployment benefits. The solid bars refer to changes in the amount of the benefits and the dashed lines to changes in their duration. The thickness of the lines indicates the direction of the reform: thick for reductions (either in the amount or duration of benefits) and thin for increases.A first simple fact is evident from Fig. 1: the past 20 years have been constellated by labour market reforms in virtually all European countries (also the ones not shown in the figure). Moreover, despite the coexistence in many countries of reforms of opposite sign, often implemented close to each other (in Finland and France, for example), most of the changes (19 out of 29 in all the 15 pre-enlargement EU countries) modified the system towards less generous benefits paid for a shorter time. However, already a simple visual inspection of Fig. 1 suggests that the correlation between these reforms and the evolution of unemployment is rather weak and it does not improve even when controlling for the business cycle. More convincing evidence can be produced for those countries where reforms took place during the years covered by the European Community Household Panel (ECHP)1, namely between 1994 and 2001. This is possible for six countries: Austria, Belgium, Finland, Germany, Ireland and the United Kingdom. Table 1 shows the conditional difference in the probability of finding employment between individuals who entered unemployment before and after the reform. The estimates are produced with a standard hazard model controlling for gender, age, health status, education, marital status, family size, presence of children, household income, regional unemployment and year dummies.Results confirm the visual impression from Fig. 1: reforms that reduced the generosity of the UI system did not have a significant effect on the probability of finding a job. There are marginally significant effects only in Austria (where the estimate is actually negative) and Ireland. The scant success of labour market reforms in Europe has been acknowledged by other authors as well, for example Coe and Snower (1997), Elmeskov et al. (1998), Orszag and Snower (1999), Nickell and van Ours (2000). This paper suggests an explanation for the failure of so many reforms of the unemployment compensation system. All European countries have complex welfare states and unemployment compensation is only one element of the system that necessarily interacts in various ways with all the other programmes. In particular, many unemployed persons receive other social assistance benefits together with their unemployment insurance. For instance, in the sample used in this paper, more than 40% of UI recipients receive other welfare benefits at some point during their unemployment spell. Most of these other benefits are means-tested, therefore a reduction or an anticipated withdrawal of unemployment insurance is often compensated, at least partly, by higher transfers from other programmes. Moreover, even those unemployed who only receive unemployment insurance may still expect to become eligible for some social assistance programmes when their benefits expire. For these workers, too, reductions or exhaustion of unemployment insurance are less of a concern and might not affect much their search effort nor their reservation wages. These arguments will be tested empirically using data from the European Community Household Panel, which allow to reconstruct monthly labour market histories for samples of individuals from all the 15 pre-enlargement EU countries. Contrary to most studies in this literature that use data from administrative sources, this paper exploits survey data which have two main advantages. First, by relying on self-reported information about one's labour market status, it is possible to separate unemployment spells that end into employment from those that end into inactivity. Hence, I will be able to estimate the probability of finding a job rather than that of leaving unemployment or the UI registry. This is an important innovation since a non negligible fraction of unemployment spells usually terminate into other types of non employment. For example, in the largest sample used in this study exits into inactivity represent about 15% of total exits from unemployment. A second advantage of the ECHP is the high level of cross-country comparability, which allows to use variation in the administrative rules of the welfare systems to identify causal effects. In particular, the institutional features of family benefits offer an interesting source of variation. In 10 out of the 15 pre-enlargement EU countries eligibility for family benefits is determined only by the number and the age of the children in the household, regardless of income. Only in the remaining 5 countries – Germany (but only up to 1997), Greece, Italy, Portugal and Spain – family benefits are subject to means testing. This paper exploits this variation to identify the differential effect of unemployment benefits on the job finding probabilities of potential recipients of family benefits. Intuitively, in countries where family benefits are means tested UI recipients who are potentially eligible for child assistance will be less worried about reductions or withdrawal of their benefits, as these could be compensated by higher social assistance payments. In countries were family benefits are universal, i.e. not means tested, there should be no such interaction effect. Results indicate significant differences in the elasticity of the job finding probability to the unemployment benefit and to the time of benefit exhaustion between male heads of households with children who are potentially eligible for means-tested family benefits as opposed to those who have access to universal benefits. The importance of interactions between welfare programmes has lately been recognised by both academicians and policy makers. Belot and van Ours (2000) provide evidence from macro data showing that countries where unemployment has fallen often owe their success to comprehensive rather than piecewise reforms of labour market policies. A theoretical justification for the importance of these interactions is discussed in Coe and Snower (1997) within a search and matching framework. Despite the acknowledged importance of potential overlappings between welfare programmes, specific evidence from micro data is still lacking. This paper aims at filling this gap by providing detailed evidence on one specific interaction. The paper is organized as follows. Section 2 briefly describes the institutional details of unemployment benefits and other social assistance programmes in Europe. Section 3 shows how interactions between welfare programmes can be analysed in a simple search model. Section 4 describes the data while Section 5 discusses the empirical strategy. The results of the empirical analysis are presented in Section 6 and Section 7 concludes.

This paper investigates how interactions between unemployment insurance and social assistance affect the search behaviour of the unemployed. The theoretical framework presented in Section 3 formalizes the idea that people are eventually interested in total payments (i.e. unemployment benefit and social assistance) and their time profile. Unemployed workers will react differently to changes in the rules of the UI system depending on what alternative or complementary welfare programmes are available. These theoretical predictions are tested using monthly labour market histories of European unemployed persons and exploiting variation in a particular feature of family benefits across Europe. While family benefits are universal in most countries, i.e. allocated according only to the number and the age of children but regardless of income, in 5 countries they are actually means tested. The comparison of similar unemployed persons who are subject to different family assistance schemes indicate significant differences in the elasticity of the probability of finding employment to both the generosity and the duration of the unemployment benefit. These results suggests that interactions between welfare programmes are, indeed, important. From a more general policy perspective, this paper highlights the need to design welfare reforms with very careful consideration for the overlappings between different programmes in the system: reducing the level or the duration of unemployment benefit may not be very effective in inducing the unemployed to search harder if they can easily shift into other social assistance programmes. This result is consistent with some recent studies that show how comprehensive reforms of labour market policies, even if politically harder to implement, are often more effective than reforming one programme at a time.