Unemployment insurance schemes, liquidity constraints and re-employment: a three country comparison.INTRODUCTIONUnemployment insurance (UI) schemes represent a well-established institution that aims to support individuals during unemployment spells. These schemes exist in all OECD countries, and although the exact mechanisms and regulations differ from country to country, the backbone of the schemes is almost the same everywhere: the UI schemes provide periodic payments to unemployed individuals for a defined period of time, subject to some eligibility criteria that workers have to meet. While helpful in mitigating the adverse conditions that unemployed workers may face, job search theory stresses that unemployment benefits may reduce the probability of re-employment and increase unemployment duration. However, things are more complicated than this simple depiction and, as Atkinson and Micklewright (1991) stresses in their survey on this subject, the actual mechanism that governs the implementation of benefits is certainly relevant in establishing the exact link between benefits and unemployment duration. As a matter of fact, actual UI schemes have some eligibility criteria and some of these criteria require workers to prove they are actively searching for a job and to devise a plan, together with employment centers (ECs) or similar institutions, on appropriate steps to take for an effective search. The existence of these incentives to search actively and the support and counseling obtained from ECs may outweigh the incentives to reduce search efforts, and, thus, UI schemes could, in principle, reduce unemployment duration. Even from an empirical point of view the actual effect of benefits on duration is not so clean cut. (1) Atkinson and Micklewright (1991) review a large number of empirical studies and conclude that the evidence is mixed and, all things considered, benefits seem to affect positively duration but their effect is, at most, weak. Another relevant aspect that should be considered when investigating the determinants of unemployment duration is related to liquidity constraints. The liquidity constraints of individuals largely affect both reservation wages and search effort and, through them, unemployment duration. In particular, we can imagine that the wealth and degree of financial stress faced by the unemployed affect duration. Unemployed workers in wealthier households facing less financial stress are likely to feel less pressure to search for a job or to accept any offer they might receive. Thus, all things being equal, wealthier unemployed should experience longer spells of unemployment. This latter aspect has received some attention but has not been fully assessed nor has it been analyzed in a comparative perspective. An interesting analysis is Bloemen (2002) where a proxy variable for wealth is computed and its effect on the probability of obtaining a job is tested. However, the analysis does not focus on unemployment duration. The role of wealth on benefits has been tackled more in detail by Chetty (2005 and 2008, with the latter being an extend version of the former). The author uses data from the United States to disentangle the moral hazard and liquidity constraint effects that benefits have on unemployed workers. The results indicate that the liquidity constraint seems to be more relevant than the moral hazard effect. Similar conclusions were reached by Card et al. (2007) using a regression discontinuity approach on Austrian data. Our paper contributes to these analyses on the determinants of unemployment duration, and it focuses both on the effect of UI schemes and on the role of liquidity constraints. In particular, we take a comparative approach, analyzing three different countries. The comparative dimension is particularly useful because, on one hand, it allows us to exploit institutional differences in the mechanisms regulating the UI schemes to better assess the exact relationship between UI and duration, and, on the other, it allows us to assess which schemes are more effective in terms of unemployment duration. We analyze three countries: Finland, Italy and Poland. This allows us to analyze a wide spectrum of economic systems and UI schemes characteristics. Finland is an advanced country with a strong welfare system. Its UI scheme is generous and long in duration, and it provides support in the job search process and imposes and effectively supervises active search by individuals. On the other hand, Poland is a former transition economy that is still facing some economic problems, unemployment being probably the most relevant. Its UI scheme is not particularly generous, it offers only minor employment services, and it does not have any job search requirements. Italy lies between these two extremes, being an advanced country, but with problems of growth and regional income disparities. Its UI scheme is among the least generous, and while, in principle, the law establishes both employment services and active search requirements, their implementation is left to the local employment services so that the effectiveness of these measures vary considerably across the country. Our investigation uses data on employment status, income and wealth of individuals for the year 2007 from the EU-SILC survey. The data allows us to determine the exact unemployment duration and the payments received from UI schemes. We focus only on workers who have just become unemployed so that the duration of unemployment before the period of observation is equal to zero for all individuals. This yields more homogenous observations and eliminates any possible left censoring problem. We develop a survival analysis where the non-survival condition consists in finding a job and we use Cox hazard models to estimate the determinants of the duration of unemployment, paying particular attention to the role of unemployment benefits, both at the start of the scheme and during the duration of the unemployment spell, and of liquidity constraints. We use three main variables to assess the stringency of liquidity constraints: the payment of interest for mortgage, the amount of taxes paid on wealth, which can be seen as a proxy for actual wealth, and the self-assessed degree of economic stress. We allow for the effect of benefits and of the financial variables to be different across the three countries, so that we are able to assess how the different UI schemes affect the unemployment duration as well as to measure the differences in the role of liquidity constraints. The results we obtain are interesting both with regards to the effect of unemployment benefits and to the role of liquidity constraints. In particular, while the more developed and search supportive schemes of Finland and, partially, of Italy seem to initially have a positive effect on re-employment probabilities, they turn out to affect negatively those probabilities as time passes. Apparently, the initial boost to the effort and to the quality of the job search wears off and the standard effect of benefits on reservation wage and search effort takes over. This is true even for Finland where strict control of employment services and search activity is maintained through the duration of the benefits. The Polish scheme, on the contrary, does not affect duration, either positively or negatively. The results on the effect of liquidity constraints are also interesting. The financial conditions and the degree of financial stress of individuals are particularly relevant in Italy but not in Finland, suggesting that individual financial conditions are less relevant in countries whose economy is doing well and that has an extensive welfare system. Results for Poland are mixed as only some dimensions of financial stress reduce unemployment duration. The paper is organized as follows: in the next section we present the UI schemes for the three countries, in the subsequent section we describe the data and we highlight some descriptive differences between the countries, in the penultimate section we perform our empirical analysis and discuss possible interpretations of our results and in the final section we conclude. THE UI SCHEMES IN FINLAND, ITALY AND POLAND We describe here the UI schemes of the three countries as they existed in 2007. Table 1 contains a summary of the main characteristics of the schemes as well as an overall assessment of their generosity in terms of the OECD ranking of the generosity of the scheme (in 2007) as reported in the OECD Employment Outlook (2009). Finland has a two-tier scheme. The first tier is a voluntary UI to which workers can subscribe. The benefits are payable to any registered unemployed person between 17 and 64 years old, that is, available for and actively seeking full-time work. The eligibility criteria require 43 weeks of work in the last 28 months and the payment of at least 10 months of voluntary contributions before the claim. Workers receive a basic benefit of 17% of the average national wage (115 [euro] per week in 2007) plus up to 45% of past earnings exceeding the basic benefit. Maximum duration is 100 weeks but older workers can qualify for extensions. The second tier is defined as a labour market support. It is payable to any registered unemployed person who is between 17 and 67 years old and is available for, and actively seeking, full-time work. There are no pre-employment conditions to receive this assistance. The actual amount is given by the basic benefit (17% of the average wage) reduced according to the household income. The benefit for younger workers living with their parents is reduced even further. The duration of this support is unlimited. The overall ranking in generosity of the Finnish scheme among the OECD countries is 9th out of 29, making it a quite generous scheme and, in particular, it displays a much longer duration than average. It should also be noted that Finland's welfare system allows for another social assistance benefit that acts as a safety net and that is given solely on the base of income and independently of employment and job-search status. The Italian UI scheme entitles unemployed individuals to receive ordinary unemployment benefits under the following conditions: to not have voluntarily left the last job; to have hold a job during the last 2 years, to have paid compulsory contributions to social security for at least 52 weeks during the last 2 years and to have declared to the local EC a willingness to work and to have agreed with the EC to a specific program to search for jobs. The benefit amounts to 50% of the average wage computed during the 3 months before losing the last job. Standard maximum duration is 6 months, but workers receive lower benefits (40% of wage) during the 7th month. Workers whose age is above 50 also receive benefits after the 7th month for a total maximum of 10 months although during the 10th month the benefits are 30% of the wage. The overall ranking in generosity of the Italian scheme is 27th, placing it almost at the bottom of OECD countries and being particularly lacking in terms of duration. Apart from basic UI, two other Italian institutions are worth mentioning: Cassa Integrazione Guadagni (CIG) and the mobility unemployment benefit. The CIG is given to temporarily laid-off workers or those working reduced hours, and it is given in the case of unfavorable economic conditions following an agreement between firms and the government. Individuals on this scheme receive 80% of their gross wage for the work time lost. It is important to stress that workers under the CIG scheme retain their job contract and are not classified as unemployed. Thus, even if they receive this form of income support, they do not enter into our analysis. Mobility unemployment benefits are given to workers previously on CIG benefits whose firms have undertaken collective dismissals or have gone bankrupt. In the former case, whenever the firm that has laid-off a worker hires a new employee, it is forced to offer the job first to workers currently on mobility unemployment benefits. It follows that these workers are slightly more likely to find a new job. The duration of these benefits is particularly long, from 12 to 48 months, depending on the sector and the geographical area, and they receive 80% of their gross wage. There are actually very few individuals receiving this kind of benefit, only 3% of all unemployed workers, according to the Bank of Italy survey on household income and wealth. The Polish UI scheme grants benefits to registered unemployed individuals who are able and ready to take up employment. Contribution to the scheme is compulsory and workers are entitled to receive the benefits if, during the period of 18 months preceding the day of registration as unemployed, they have been employed for at least 365 days and if the termination of the contract was not voluntary. The benefit amounts to 24% of the average national wage, about 151 [euro] per month in 2007, but it is adjusted according to the length of past employment spells so that more experienced workers receive higher benefits. Maximum duration is 6 months, but it is increased to 12 or 18 months for individuals from areas where the unemployment rate is higher than the national average. The overall ranking in generosity of the Polish scheme among the OECD countries is 21st out of 29, making it lower than average in terms of generosity. It should also be noted that the Polish welfare system allows for another social assistance benefit that acts as a safety net and is given solely on the basis of income and independently of employment status. Thus far we have described the UI schemes in terms of eligibility criteria, benefit amount and maximum duration. However, together with actual income support, the UI schemes may also provide some form of employment counseling and require active job search activities. In particular, the schemes usually provide some employment services and counseling through the local ECs and might require some proof of active search. We summarize in Table 2 the main forms of services and search requirement in the three countries. A detailed description of these can be found in the OECD Employment Outlook (2007). The Finnish scheme requires that, at the moment of registration, the ECs have to check for suitable vacancies and offer them to unemployed workers. On the other side, workers have to apply for the offered vacancies. In addition, the ECs and the unemployed have to agree on an individual action plan (IAP) describing specific action to search more effectively for jobs. After this initial stage, workers have to report monthly to a counselor to provide proof of actual search efforts. The Italian laws on the UI scheme prescribe that local ECs have to devise, together with the worker, an IAP describing specific actions that workers have to follow and how to check search progress. In this sense, Italian law does not prescribe compulsory placement efforts from ECs nor describe exactly if and how workers' search activity should be checked. However, these aspects may become compulsory depending on the local EC programs. The Polish scheme requires that, at the moment of registration, ECs have to check for suitable vacancies and offer them to the unemployed workers. Accordingly, workers have to apply for the offered vacancies. No IAP is expected to be carried out nor is there a search requirement to be fulfilled during the unemployment spell, although individuals have to report monthly by post whether they are still unemployed. Summing up, we observe that the three countries offer different degrees of counseling and requirements in terms of search. Finland offers extensive employment services and requires strict controls of search effort; in Poland on the contrary employment services are barely present and search activity is not required. Somewhere between falls Italy, whose legislation in principle prescribes that employment services should be offered and that workers should search actively to be eligible for benefits. However, the actual implementation of these principles is left to the local ECs so that large differences may exist from area to area. DATA DESCRIPTION We use data from the EU-SILC 2008 survey, which contains detailed data on individuals and households in 2007. The survey allows us to identify newly unemployed individuals and we perform our analysis on them. We define as newly unemployed an individual who is currently unemployed and who, in the previous months, was in paid employment or self-employed. The survey contains the working status for each calendar month and therefore we are able to identify the newly unemployed and to compute unemployment duration in months for those individuals who end up finding a job. According to the EU-SILC survey classification we consider unemployed an individual who has specifically declared unemployment to be his status and who has declared not to be currently in paid work, in self-employment nor to fall in the following categories: retired, student, military activity or other inactivity. According to the EU-SILC classification, individuals on temporary lay-off are considered employed if they receive at least 50% of their gross wage, and so we do not consider workers on the Italian CIG scheme as unemployed. In the computation of unemployment duration we also include workers who were still unemployed during December 2007, but their condition results censored as we do not know when and if they eventually find a job. We end up with a sample of 195 newly unemployed workers for Finland, 536 for Italy and 471 for Poland. The survey contains information about the demographic characteristics such as age, gender, marital status, education, region of dwelling and so on, and the economic characteristics both of the individual and of the household (basically, income from unemployment benefits and some measures of the household wealth). The information on income from unemployment benefits takes the form of the total income from ordinary unemployment benefits, mobility benefits and severance payments. We divide this amount by the total months of unemployment in 2007 to obtain the average monthly benefits. The EU-SILC survey also contains some data on household wealth and financial conditions that we use in our analysis. In particular, we use the amount of taxes on wealth per household equivalised components a proxy for household wealth. (2) We also use the amount of interests on the mortgage (if any) paid yearly and, finally, we use some qualitative information on the household economic situation that in the survey takes the form of a question on whether the household was able to make ends meet to which the individuals could give six different answers: 'with great difficulty', 'with difficulty', 'with some difficulty', 'fairly easily', 'easily' and 'very easily'. (3) It is useful to give a description of how some key variables vary across the three countries. In Figure la, we represent the share of newly unemployed who found a job within the year 2007. As is clear from the figure, the share is fairly large in Finland and in Italy, but is significantly smaller in Poland. Similarly, the share of newly unemployed who received benefits (see Figure l a) is large in Finland and quite large in Italy, whereas benefits are much less frequent in Poland. A depiction of the wealth and financial condition of the unemployed is given in Figure 2. Mortgages (Figure 2a) are relatively frequent in Finland whereas they are not so common in Italy and hardly present in Poland. (4) What is really interesting is the distribution of unemployed workers in terms of wealth (Figure 2b). In Finland about 23% belong to the top quartile of the wealth distribution, considering all the households in the population, implying that unemployment is spread almost evenly among households of all wealth classes. On the contrary, and probably more in line with what one should expect, unemployed individuals in Italy and Poland are less likely to belong to a wealthier household. On similar lines, unemployment in Finland does not appear to affect too much the self-perception of the degree of economic problems (Figure 2c), as only 10% of unemployed individuals come from households with major economic problems. This is not true for Italy and in particularly for Poland, where unemployed workers come from households that often report having major problems. The overall impression from this data is that, in Finland, being unemployed is more common and less traumatic than it is in the other two countries. Finnish unemployed workers are spread quite evenly among all wealth classes and they do not often report major economic problems. EMPIRICAL ANALYSIS In this part, we perform an econometric analysis of unemployment duration. We focus on the effect that unemployment benefits, wealth and financial pressure have on duration. Our total sample is made of 1,202 individuals who, during the year 2007, became unemployed. We perform a survival analysis, that is, we aim at estimating the probability that an unemployed finds a job and how this probability is affected by the passing of time and by some selected covariates. Basically, we assume the existence of a function h(t) that determines the probability that individuals move from unemployment to employment at time t, conditional on the fact that the individual is still unemployed at time t. This is called the hazard function. If we define as F(t) the probability of not being unemployed after t periods, with S(t) = 1-F(t), that is, S(t) is the probability of still being unemployed after t periods, also known as survival function, and with f(t) = F'(t) so that f(t) is the probability of switching from employed to unemployed at exactly time t, we have: h(t) = f(t)/S(t), (1) To carry out our estimation we assume that f(t) takes the form of a specific distribution, and that it thus depends on a set of parameters [theta] describing the distribution and on a set of covariates x that influence the probability of leaving unemployment. Given a certain f(t) it is possible to determine h(t), and we can write the hazard function as h(t, [theta], x) where [theta] represents the actual parameters to be estimated. We also assume that the effect of the covariate is the same in each period, an assumption that gives the Proportional Hazard Model, which can be written as: h(t) = [h.sub.0](t, [[theta].sub.0]) x [rho](x, [[theta].sub.x] (2) where [h.sub.0](t, [[theta].sub.0]) is known as the baseline hazard function, which is the same for all individuals and only depends on time and parameters and where [rho](x, [[theta].sub.x]) determines the effects of the covariates that are independent of time t. In our econometric analysis we estimate Equation 2 through Cox estimation, and we obtain estimates of the parameters [[theta].sub.x], which allows us to determine which variables are relevant in explaining duration. At first, we perform a parametric Cox regression assuming a Weibull distributed hazard function, and we then perform a semi-parametric analysis without making any assumptions on the exact function form [h.sub.0](t,[[theta].sub.0]) to check for robustness. Given that we are dealing with observations from different countries, we adopt a stratified approach and assume the baseline hazard functions to be country specific so that we estimate an ancillary parameter, which defines the shape of the hazard function that is different for each country. In addition, since we are particularly interested in the role of the UI schemes and of liquidity constraints, we allow for the variables measuring benefits and the financial conditions to have country-specific coefficients. Therefore, we estimate the following: h(t,j) = [h.sub.0], j (t, [[theta].sub.0]) x [rho] (x, [[theta].sub.x], y, [[theta].sub.y]j) (3) where j determines the country of origin and y are the variables whose coefficients are country specific. We start our analysis by presenting the hazard estimates (Figure 3) for the three countries: basically, these are descriptive measures of the probability of finding a job conditional on having spent a given amount of time in unemployment. The patterns represented in Figure 3 show that the conditional probability has similar patterns but different scales in the three countries. In particular, the conditional probability at first slightly increases, in particular in Poland, then is stable for a while and finally declines. This homogeneity of behaviors changes if we examine the conditional probability of finding a job for individuals with and without unemployment benefits (Figure 4). [FIGURE 3 OMITTED] [FIGURE 4 OMITTED] The patterns in Figure 4 are quite striking. In Finland and Italy, workers on unemployment benefits initially have higher re-employment probabilities but this relationship reverts through time. The contrary is true for Poland, where individuals on benefits have lower probability at the start but they end up with higher probability of reemployment at the end. These findings, although extremely interesting, create a problem in the actual estimation of the Cox proportional hazard model because one of the key assumption of this model is that the effect of a given covariate is the same through time. Thus, the two patterns in Figure 4 should be more or less parallel for each country, and Figure 4 seems to indicate a violation of this assumption. (5) To overcome this problem we add another variable that is the interaction of unemployment benefits and time to take into account the time-varying effect of benefits and solve the problem of non-proportionality. The key variables we use in our estimation are the amount of benefits, the interaction of benefits with time and the three variables that we use as proxy of household wealth and financial conditions: (1) the mortgage payment, (2) taxes paid on wealth (divided by the equivalized size of households) and (3) a qualitative variable that represents, according to the individuals, whether the household is 'having problems in making ends meet'. This latter variable takes the form of a dummy that is one if the household experiences great difficulty or difficulty in making ends meets. As we are particularly interested in comparing the effect of these variables among the different countries, we allow for country-specific coefficients for the five key variables. We also add several other control variables and in particular we include age, gender, education and region of origin. (6) Finally, to account for unobserved heterogeneity we added a variable that measures months spent in unemployment in 2006 and that should capture unobservable characteristics of the individuals that make them more likely to stay in unemployment. (7) The use of this variable that takes into account unobservables also helps us to mitigate the effect of self-selection into benefits receipt. Table 3 presents the results for estimations of the Cox models. A positive coefficient implies a positive relationship with the probability of finding a job and thus a negative relationship with unemployment duration. Model (i) reports the estimation results from the parametric regression whereas Model (ii) adopts a semi-parametric approach, and we use the latter to check the robustness of our results. The results from Model (i) on the effect of the UI schemes indicate two patterns depending on the country. In Finland and in Italy being on the UI schemes produces a statistically significant effect on unemployment duration, but this effect changes with time. At first, being registered for the UI implies a higher probability of finding a job, but this effect decays with time. Apparently, the employment counseling received at the moment of the registration of benefits exerts a positive effect on the quality and the quantity of search. However, after this initial effect, the boosts to effort seem to dissipate and the standard effect of benefits on the reservation wage and search effort predominates. It is important to stress that this is true even in Finland where employment counseling and checks on active search take place each month. Apparently, even when these aspects of UI are carried out efficiently, only the initial counseling exerts a relevant effect, whereas the following checks on effort are not effective. The situation is different in Poland where employment services are carried out less extensively and in fact the UI does not display a significant improvement on re-employment probabilities It is important to stress that even in Poland benefits do not significantly reduce employment probability. The results on the role of financial condition are less clear cut. In the case of Italy, mortgages, wealth and economic problems have the expected effect on unemployment duration so that individuals facing less financial pressure stay unemployed longer. However, in the case of Finland this set of variables is not significant. A possible explanation is that unemployed workers in Finland do not face serious liquidity constraints so that their actual degree of wealth is not particularly relevant in the determination of unemployment duration. This interpretation is also supported by the data we described in the previous section that showed that unemployed workers are equally distributed among households of all wealth classes and do not appear to face severe economic problems. Finally, the results for Poland are mixed: mortgages have a positive effect on re-employment, suggesting that financial pressure has an influence on duration. However, household wealth is not relevant and having major problems in making ends meet actually has a negative effect on re-employment. These latter results suggest that in a country that is facing serious economic problems and where the welfare system is not well developed, the actual pressure to seek work is not mitigated by household wealth. Two things should also be considered from this point of view. First, household wealth is relevant only in Italy, a country that is known for strong family ties so that the family can act as a safety net for unemployed workers. Second, looking in more details at the Polish data it turns out that individuals who declare that they make ends meet with 'some problems' have a strictly higher hazard rate than those that have major problems or no problems (whose hazard rate is particularly low). (8) This suggests that having some problems does induce a reduction of the reservation wage and an increase in search effort, whereas having too serious problems might have a discouragement effect. Other, more standard, results are also found. Younger and older individuals stay unemployed longer as do individuals with lower education or living in less productive areas. Interestingly enough, gender does not have a statistically significant effect. Finally, the variable indicating that workers had been previously unemployed is clearly significant, allowing us to take into account some of the unobserved heterogeneity. In Table 3, we also report the value of the logarithm of the ancillary parameters. Since the logarithms of these parameters are significantly greater than zero, the very parameters are non-negative. Under the assumption of a Weibull distributed hazard function, this implies a non-decreasing hazard function, something that supports the robustness of our results. In fact, in survival analyses, a decreasing hazard function may be a signal of self-selection into being unemployed and a sign that we are not including relevant variables in the regressions. (9) We further check the robustness of our results using the estimation from Model (ii), where a semi-parametric specification was used. Almost all the results are confirmed with one important exception. The interaction of time and benefits for Finland is negative but not significant. However, the significance of this variable is quite close to the threshold level, being different from zero with 86% probability. In addition, since no systematic difference is detected between the two models, the first specification should be correct and more efficient. Given that this interaction is negative in both specifications, that in the second is almost significant and that the first specification appears to be correct and certainly more efficient, we believe that the negative effect should be considered relevant. We can now sum up our findings. UI has a complex effect on unemployment duration. When coordinated with employment counseling it provides incentives for more and better job search and overcomes any negative effect due to increase of reservation wage and the reduction of search effort. However, after some time, the latter effects take over and UI seems to increase duration. This is true also in Finland, where employment services and search requirements takes place through the duration of benefits. As for liquidity constraints they appear to be relevant in Italy and partly in Poland suggesting that they are relevant in countries where wealth and social welfare are limited, unlike the case of Finland. CONCLUSIONS Our analysis focused on the effect that UI schemes and liquidity constraints have on re-employment probabilities and unemployment duration and tried to tackle the different mechanisms through which these operate by comparing different countries. Using a sample of newly unemployed individuals from Finland, Italy and Poland in 2007, we estimate Cox hazard models. The results show how UI schemes may have an initial positive effect on the re-employment probabilities but, even in schemes that offer continuous employment services and expect strict and periodical search requirements, this effect decays through time. This pattern is also consistent with the results for Poland, whose UI scheme that does not provide well developed employment services and fails to increase re-employment probabilities. The role of liquidity constraints was also explored, and the results we obtain are quite different depending on the country we are observing. In Italy and partly in Poland individuals with more financial pressure have shorter unemployment duration whereas in Finland these aspects are not significant. Apparently, the individuals' financial conditions are less relevant in countries where, like Finland, wealth and social welfare are high. Interestingly enough, household wealth is only relevant for Italy, a country that is known for its strong family ties. From a policy point of view our analysis suggests that unemployment benefits should be given together with employment services though strict search requirements may not be effective in promoting search effort over longer spells of unemployment. Given this pattern it is advisable for benefits to be quite generous but not particularly long as, with time, they necessarily appear to increase unemployment duration. Acknowledgements The author would like to thank for their useful comments the participants of the International Workshop 'Crises, Institutions and Labour Market Performance: Comparing Evidence and Policies' held in Perugia, Italy in November 2011. Suggestions from Giuseppe Croce were particularly useful. The analysis is based on data obtained from European Commission, Eurostat, longitudinal EU-SILC/2011/08. Eurostat has no responsibility for the results and conclusions, which are those of the researcher. The usual disclaimer applies. REFERENCES Asbenfelter, O, Ashmore, D and Deschenes, O. 2005: Do unemployment insurance recipients actively seek work? Evidence from randomized trials in four US States. Journal of Econometrics 125(1-2): 53-75. Atkinson, A and Micklewright, J. 1991: Unemployment compensation and labor market transitions: A critical review. Journal of Economic Literature 29 (4): 1679-1727. Bloemen, H. 2002: The relation between wealth and labour market transitions: An empirical study for the Netherlands. Journal of Applied Econometrics 17(3): 249-268. Card, D, Chetty, R and Weber, A. 2007: Cash-on-hand and competing models of intertemporal behavior: New evidence from the labor market. The Quarterly Journal of Economics 122(4): 1511-1560. Chetty, R. 2005: Why do unemployment benefits raise unemployment durations? Moral Hazard vs. Liquidity. NBER Working Papers 11760. Chetty, R. 2008: Moral hazard vs. liquidity and optimal unemployment insurance. Journal of Political Economy 116(2): 173-234. Eurostat. 2007: Description of SILC user database variables: Cross-sectional and longitudinal. Version 2007.1 from 01-03-09. Eurostat, Unit F-3: Luxembourg. Grambsch, P and Therneau, T. 1994: Proportional hazards tests and diagnostics based on weighted residuals. Biometrika 81(3): 515-526. Klepinger, DH, Johnson, TR and Joesch, JM. 2002: Effects of unemployment insurance work-search requirements: The Maryland experiment. Industrial and Labor Relations Review 56(1): 3-22. Manning, A. 2009: You can't always get what you want: The impact of the UK jobseeker's allowance. Labour Economics 16(3): 239-250. Nickell, S and Layard, R. 1999: Labor market institutions and economic performance. In: Ashenfelter, O and Card, D (eds). Handbook of Labor Economics. Elsevier: Amsterdam. Vol 3: 3029-3084. OECD. 2007: OECD employment outlook 2007. OECD Publishing: Paris. OECD. 2009: OECD employment outlook 2009. OECD Publishing: Paris. LORENZO CORSINI Department of Economics, University of Pisa, Via C. Ridolfi, 10, Pisa 56124, Italy. E-mail: lcorsini@ec.unipi.it (1) There is a large number of studies on the effect of UI schemes, though these analyses are seldom carried out from a comparative perspective. A good review is contained in Atkinson and Micklewright (1991) More recent studies focus on the role of eligibility criteria on the search effort and unemployment duration and their conclusions are mixed. Klepinger et al. (2002) use data from Maryland and show that stricter criteria improve search efforts and reduce unemployment duration. Ashenfelter et al. (2005), using data from different American states, conclude that stricter search criteria do not affect the access to benefits. Manning (2009) uses difference ill differences estimations using data for the United Kingdom in 1996 and shows that criteria affect the access to claims but stricter criteria discourage workers from meeting the search requirement and thus do not facilitate the transition to an employment. As for the relationship between benefits maximum duration and unemployment length, Nickell and Layard (1999) suggest that these two aspects are positively correlated. (2) The equivalence scale takes into account the age component. A full description of the scale can be found in Eurostat (2007). (3) The exact form of the question in the EU SILC questionnaire was: 'Concerning your household's total monthly or weekly income, with which degree of ease or difficulty is the household able to make ends meet?' (4) The extreme scarcity of mortgages in Poland is not a peculiarity of unemployed individuals: a similar frequency is also found in the population as a whole. (5) We also performed the test for proportionality assumption proposed by Grambsch and Therneau (1994) on the residuals from an estimation of the Cox hazard model. The test rejected the assumption of proportionality related to (and only to) the unemployment benefits. (6) In particular, we use age and age squared to take into accounts the non-linear effect of ageing. Education enters the regression as a dummy, which is one if individuals have at least upper secondary education (ISEC Degree 3 or higher) and zero otherwise. The region of origin is expressed as a dummy, which is one if individuals come from a region whose GDP is <75% of the country average. Different measures for education and place of origin were also tried, but these two were the most significant. (7) Since our sample consists of newly unemployed, the months spent in unemployment in 2006 necessarily belongs to another spell of unemployment and thus are not already included in the actual unemployment duration. (8) In particular, in Poland, we observe that the share of individuals that found a job where: 19.9% if declaring to make ends meet 'with great difficulty', 22% if declaring 'with difficulty', 29.6% if declaring 'with some difficulty', 27.8 if declaring 'fairly easily', and 14.28 if declaring 'easily' or 'very easily'. (9) In fact, declining hazard functions might capture the desired effect of passing of time but also the permanence in unemployment of individuals whose unobserved characteristics self-select them into remaining unemployed.
Table 1: UI schemes characteristics
Ranking among Voluntary or Requiring
OECD in terms compulsory previous
of overall employment
generosity
Finland 9th/29
Unemployment Voluntary Yes
insurance
Unemployment Compulsory No
assistance
Italy 27th/29
Unemployment Compulsory Yes
insurance
Poland 21st/29
Unemployment Compulsory Yes
insurance
Amount of Maximum
benefits duration
Finland
Unemployment Basic benefit of 17% of 100 weeks
insurance average national wage +
45% of previous
earnings exceeding
basic benefit
Unemployment 17% of average wage Unlimited
assistance
Italy
Unemployment 50% of previous earnings 7 months
insurance (40% during the 7th month)
Poland
Unemployment 24% of the average 6 months
insurance national wage, adjusted
on past length of
employment
Notes
Finland
Unemployment
insurance
Unemployment Amount is reduced
assistance according to household's
income
Italy
Unemployment Older individuals get
insurance extended duration
Poland
Unemployment Duration is extended
insurance to 12 or 18 months
in regions with high
unemployment rate
Source: OECD Employment Outlook (2009)
Table 2: Active employment services and search requirements
Placement efforts IAP
at initial registration
Finland Law requires ES Yes
office to check
for suitable vacancies.
Workers' application
is compulsory
Italy Law does not require Yes
ES office to check
for suitable vacancies
but actual efforts
vary according to
ES office
Poland Law requires ES Not compulsory
office to check by law and
for suitable rarely put
vacancies. Workers' into practice
application is
compulsory
Reporting Further interviews
requirements during unemployment
Finland Reports on search Compulsory by law,
activity through an their actual frequency
in-person counseling depends on the IAP
interview are
compulsory at least
once a month
Italy Active job search is Not compulsory by
required for workers, law, but they may
but actual checks are be included in the IAP
not required by law
and depend on local
ES procedures
Poland Reports on search Not compulsory
activity are not
required
Source: OECD Employment Outlook (2007)
Table 3: Cox estimations of re-employment probabilities
(i) (ii)
Parametric cox Semi-
(Weibull parametric cox
distribution)
Age 0.111 *** 0.112 ***
(0.0301) (0.0285)
Age squared -0.00134 *** -0.00136 ***
(0.000389) (0.000378)
Gender -0.0172 0.0140
(0.129) (0.111)
Education 0.309 * 0.209
(0.165) (0.139)
Less productive region -0.366 ** -0.314 **
(0.157) (0.136)
Months of unemployment in 2006 -0.0413 ** 0.0461 ***
(0.0162) (0.0143)
Finland
Unemployment benefits 0.000383 *** 0.000251 ***
(5.57e-05) (6.10e-05)
Interaction of benefits with time -0.000527 *** -0.000175
(0.000167) (0.000120)
Payments for mortgage 1.05e-05 3.18e-05
(0.000129) (8.24e-05)
Wealth 0.000743 -0.000214
(0.00131) (0.00114)
Problems in making ends meet -0.00163 0.0122
(0.421) (0.323)
Log of ancillary parameter 0.5347 ***
(0.081)
Italy
Unemployment benefits 0.000120 *** 9.08e-05 ***
(1.86e-05) (1.98e-05)
Interaction of benefits with time -7.49e-05 *** -4.09e-05 *
(2.05e-05) (2.43e-05)
Payments for mortgage 8.39e-05 * 7.55e-05 **
(4.38e-05) (3.73e-05)
Wealth -0.000775 * -0.000689 **
(0.000494) (0.000427)
Problems in making ends meet 0.336 * 0.263 *
(0.190) (0.150)
Log of ancillary parameter 0.3797 ***
(0.0.45)
Poland
Unemployment benefits -0.000130 -0.00533
(0.000872) (0.00478)
Interaction of benefits with time -0.00101 0.00333
(0.000494) (0.00299)
Payments for mortgage 0.000395 * 0.000383 *
(0.000238) (0.000212)
Wealth -0.00630 -0.00628
(0.00634) (0.00559)
Problems in making ends meet -0.406 * -0.356 *
(0.216) (0.196)
Log of ancillary parameter 0.4601 ***
(0.0685)
Observations 1,202 1,202
Standard errors in parentheses. Significance level: * P<0.1,
** P<0.05, *** P<0.01.
Figure 1: (a) Share of unemployed individuals that found a job
within 2007; (b) Share of unemployed individuals receiving benefits
a b
Finland 0.51 0.77
Italy 0.40 0.62
Poland 0.24 0.32
Note: Table made from bar graph.
Figure 2: (a) Share of unemployed individuals paying a mortgage;
(b) Share of unemployed individuals whose household is within the top
quartile of wealth distribution; (c) Share of unemployed individuals
whose household has major problems in making ends meet
a b c
Finland 0.28 0.23 0.10
Italy 0.15 0.16 0.45
Poland 0.03 0.14 0.57
Note: Table made from bar graph.
|
|
||||||||||||||||||||||
Printer friendly
Cite/link
Email
Feedback
Reader Opinion