Labor Market Reforms in Europe: Towards More Flexicure Labor Markets?

Labor market segmentation refers to a salient divide between secure and insecure jobs and is related to problems in important areas, including macro�?economic efficiency, workers' wellbeing and repercussions for social cohesion. European countries have started a new wave of labor market reforms in the aftermath of the 2008/09 crisis to tackle a number of issues, including labor market segmentation. This particularly concerns reforms in: (1) employment protection, i.e. dismissal protection and restrictions on fixed�?term contracts; (2) unemployment benefit generosity and coverage; and (3) the intensity of active labor market policies.The paper provides an overview of reform patterns and tries to assess whether and to what extent these reforms have led to more or less dualized labor markets in terms of dismissal protection, the provision of unemployment benefits and access to ALMPs. In particular, we will provide some evidence on potential changes in hirings on temporary contracts.


Introduction
Labor market segmentation has been one of the most prominent topics in the debate about labor market reforms in Europe. The issue has gained new momentum with the asymmetrical impact of the 2008/09 recession on young people -many of whom are employed on a fixedterm basis -and the subsequent wave of labor market reforms implemented particularly in countries heavily hit by the crisis. This paper aims to show trends in labor segmentation patterns, provide an overview of recent labor market reforms and assess whether these reforms have deepened the segmentation or -reversing a long-standing trend -helped to overcome divides in the labor market that were often attributed to institutional dysfunctionalities.
What do we mean when we talk about labor market segmentation? The term has been used in different ways over recent decades. In this report, we will use it to refer to a salient divide in European labor markets between secure and insecure jobs. This divide has been institutionalized by the creation and expansion of non-standard forms of employment, most notably temporary employment (such as fixed-term and agency work). It does not make much sense to think of segmentation in static terms. A labor market is only segmented if there are significant mobility barriers. A large stock of temporary workers in the labor market is unproblematic if there is high mobility into secure jobs; indeed, many temporary workers make such a successful transition. However, segmentation exists if workers feel 'trapped' in unstable and insecure jobs, interrupted by repeated unemployment spells. Addressing segmentation thus means addressing (lacking) mobility into stable jobs.
To be more specific, we have to be clear about which problems resulting from segmentation are to be addressed. With some simplification, they can be found in three areas: macroeconomic efficiency, workers' well-being and repercussions for social cohesion. Regarding macro-economic effects, there are indications that the growth of temporary employment has hampered productivity growth (because investment in the workforce is discouraged) and failed to increase employment levels (because secure employment is substituted with temporary jobs) (Boeri and Garibaldi 2007;Boeri 2011).
There is also strong evidence that job insecurity and temporary contracts in particular depress workers' health and well-being (Benach et al. 2014;De Cuyper et al. 2008;Sverke et al. 2002). Naturally, the same is true for unemployment and poverty, as experiences that are inevitable if mobility out of insecure jobs is lacking. There are strong indications that the stress caused by such social problems not only impedes health but also interferes with central aspects of people's lives, including cognitive abilities, family formation and parenting (Mani et al. 2013). What makes segmentation and labor market risk particularly stressful is that many European welfare states are ill-prepared to deal with non-standard workers. Built upon the assumptions underlying the post-war employment model, eligibility criteria for social insurance benefits often disadvantage workers with short and interrupted work records (Berton et al. 2012;Clasen and Clegg 2011;Hinrichs and Jessoula 2012).
Hence, the premise of this paper is that measures to address labor market segmentation should primarily improve non-standard workers' socio-economic situation in terms of employment stability and mobility as well as social protection. Nonetheless, for the sake of social cohesion and democratic legitimacy, they should also break with the 'dualizing' reforms of the past and signal more solidarity across labor market divides. 3 To preview the message put forward in this paper, the best way to improve the well-being of insecure workers and their integration in politics is to ensure sufficient possibility for upward mobility. Of course, the question of how to ensure mobility is not always easy to answer. Second -and equally importantly -we need to ensure that insecure workers' disadvantage is not multiplied through the welfare state, namely through 'social insurance penalties' related to fragmented employment biographies. Moving to more universal systems can substantively improve the social situation of vulnerable workers and send a strong signal that society does not leave some citizens behind. Third, it is substantively and symbolically important that employers contribute to the social costs of their hiring practices.
The paper is structured as follows. We begin with a background section on patterns of temporary employment in the European Union, highlighting considerable segmentation between temporary and permanent jobs, even if upward transitions happen frequently. In the subsequent section, we review recent attempts of member states to address labor market segmentation on four dimensions: employment protection legislation (EPL), unemployment benefits, active labor market policies (ALMP) and working-time flexibility. We will assess the extent to which segmentation has been reduced or deepened through reforms, particularly regarding EPL, unemployment benefits and ALMPs. Subsequently, we briefly point to the challenges that policy-makers face in reducing labor market segmentation. A final section concludes and derives policy recommendations.

Background: Labor market segmentation in the European Union
In this report, we focus on a specific form of segmentation, namely segmentation between workers with permanent and temporary employment contracts. While this is not to deny that other potentially precarious forms of non-standard employment exist-including part-time employment, temporary agency work, and quasi-dependent self-employment -as opposed to temporary employment, part-time employment is predominantly voluntary in most European countries. 2 Moreover, while temporary agency work and dependent selfemployment can be highly precarious, they make up considerably smaller shares of the European workforce. Figure 1 presents the development of the share of workers with temporary employment contracts in the EU-28 countries. Besides the share among all wage earners, it also presents shares among young (15-29 years) and older workers (55-54 years). First of all, there is considerable variation across member states. For instance, low shares of around six per cent or less are observed in Bulgaria, Estonia, Latvia, Lithuania, Romania and the UK. Shares considerably above the EU-28 average of 14 per cent can be found in Croatia, Sweden (both 17 per cent), Cyprus (19), Portugal, the Netherlands (both 21), Spain (24) and Poland (28). 4 The solid black line in Figure 1 very clearly shows that the young are disproportionately affected by temporary employment. While this is a well-known pattern, a discomforting finding is that the gap between the young and the general population is dramatically growing in some countries, particularly Croatia, France, Italy, Netherlands, Poland, Slovenia and (in most recent years) Spain. Furthermore, there is a growing share of temporary employment among the young in Germany, although the development is difficult to assess against the background of Germany's sizable apprenticeship system, which is responsible for about half of all temporary contracts. Source: Eurostat (2015).
Temporary employment in the age bracket of 55-64 is substantially less pronounced compared to the general workforce in virtually all member states. The highest shares are observed for Cyprus, Hungary, Portugal, Spain (all around ten per cent) and Poland (18). Moreover, the development in this group is rather stable. Hence, we can conclude that the problem pressure in terms of segmentation is much stronger among young workers and thus it seems advisable to focus on this group.
Looking at stocks of temporary workers might be misleading because policy changes only affect newly recruited workers. Figure 2 thus presents the share of temporary contracts among new employment contracts (less than three months old) at two different points in time as presented by the OECD. At this level of aggregation we observe a rather strong persistence over the past years and in many member states temporary contracts among newly recruited workers remain pervasive.
A striking pattern in Figures 1 and 2 is that not a single member state managed to reverse the trend of growing temporary employment among the young. A partial exception is Spain, where many temporary jobs were lost, lowering their share in total employment. Furthermore, the decline came from an extremely high level and has been reversed in recent years, exhibiting the highest shares of fixed-term contracts in youth employment as well as recent hirings. Hence, the project of "addressing labor market segmentation" cannot build upon easily identifiable best-practice cases. One important reason for this lack of success is arguably that many of the most segmented countries face a difficult overall labor market situation. The upper panel in Figure 3 plots unemployment rates and temporary employment shares among those aged 15-24 (in 2014). With youth unemployment rates above 30 per cent and despite regulatory issues, the Mediterranean countries also suffer from a lack of labor demand and cannot offer young people a sufficient number of secure positions in the labor market. This also means that reducing temporary employment cannot be the top priority among policymakers in these countries. There is little doubt that the negative socio-economic effects of unemployment are much more severe than those of being in a temporary job.
Temporary employment does not produce segmentation if it is merely a transient experience that leads to stable labor market inclusion. However, if temporary workers remain in insecure positions for a long time, we can conclude that they are 'trapped' in insecure and precarious jobs. This point is best illustrated by examining the upward mobility of temporary workers. The European Commission has recently presented year-to-year transition rates from fixedterm to permanent contracts (see Figure 3). Furthermore, the OECD (2015, p. 188) has recently provided numbers for the long-term probabilities of temporary workers obtaining a permanent contract. Low transition possibilities are found in a couple of countries, which can thus be characterized as having strongly segmented labor markets. Here, the share of involuntary fixed-term jobs in all temporary contracts is the highest (European Commission 2016). However, there also is a group of countries in which labor market context is less disadvantageous and temporary employment are nonetheless widespread among the young. Some of these countries -such as Austria or Germany -are characterized by large shares of apprentices among fixed-term contract holders.   The lower panel in Figure 4 plots youth unemployment rates and the probabilities that temporary workers will obtain a permanent contract over a ten-year period (unfortunately, data is only available for a limited number of European cases). The plot reveals significant variation in transition rates: whereas virtually all temporary workers move to permanent jobs in Austria, Estonia and Germany, around half of the workers in Italy and Spain are still in temporary jobs, even after ten years. The plot also reveals that the prospects for upward mobility strongly correlate with general labor market performance. While this is unsurprising, it is important to emphasize that labor market segmentation in many European countries is closely related to the macro-economy and insufficient demand for labor. 1) Segmentation -measured as the share of temporary employment -is a persistent feature of many, but not all European labor markets. To date, no member state has substantially managed to reverse the trend.
2) Segmentation is more severe in member states with high youth unemployment. This concerns the share of temporary employment as well as the prospects of mobility into permanent jobs.
3) Even if many temporary workers make a successful transition into permanent employment, segmentation does exist. Considerable shares of temporary workers appear to be 'trapped', even in the long run.

Measures taken to address labor market segmentation and their impact
How have member states responded to labor market segmentation? In this section, we present measures taken over recent years in the fields of EPL, unemployment benefits and ALMPs. To reduce complexity, we will focus this review on countries with substantive shares of temporary workers and the most significant policy changes.

Employment protection legislation
It is often argued that segmentation results from institutional mobility barriers. Because dismissals are costly in strictly regulated labor markets, employers make use of temporary contracts that allow for a 'cheaper' adjustment of the workforce. Many policy experts have recommended deregulating EPL for regular contracts, although this has proven difficult for political reasons. Instead, many member states have deregulated the use of temporary contracts: a strategy that has contributed to segmenting the labor market (e.g. Boeri 2011; Kahn 2010).
Thus, there are two possible strategies to address segmentation through EPL reforms, which are not mutually exclusive: 1. de-regulating permanent contracts; and/or 2. re-regulating temporary contracts.
As mentioned above, deregulating permanent contracts has been difficult to implement politically. However, in the aftermath of the Great Recession, some of the most segmented and crisis-ridden counties in Southern Europe implemented substantial deregulation, including France, Greece, Italy, Portugal, Slovenia and Spain (OECD 2013;Game et al. 2015).
Shown in Figure 5, EPL reforms -as well as reforms in other policy areas -can be mapped using the European Commission's LABREF database (Turrini et al. 2015) and the ILO inventory on labor market reforms (Game et al. 2015).

Figure 5: Employment protection reforms in the EU (LABREF)
In line with this we can observe major changes in the aggregate EPL index developed by the OECD and available until 2013. Here, it has to be noted that some of the most recent and probably most substantial reforms have not yet been incorporated.

Dismissal protection reform
Regarding employment protection in case of open-ended contracts, Greece, Portugal, Slovenia and Spain's post-crisis reforms considerably reduced notice periods and severance pay for (newly hired) permanent workers (see Figure 6). In Italy, building upon earlier, more partial reforms, the recent 'Jobs Act' stipulated that employers no longer have to reinstate workers dismissed for invalid economic reasons; whereby this reform was combined with strong employer incentives for new hirings on open-ended contracts, including conversions from temporary to permanent contracts, but falling short of a single type of contract removing the distinction between permanent and temporary types of employment (Picot and Tassinari 2015;Sestito and Viviano 2016). In Spain, dismissals were facilitated by extended and clarifying the reasons for separation as well as limiting severance pay through the 2010 and 2012 reforms, while strengthening the employment protection of temporary workers and encouraging internal flexibility. While the reforms in Greece, Italy, Portugal and Spain arguably bring substantive reduction of EPL for permanent workers, reforms in France and the Netherlands were more subtle. In France, workers who do not accept collectively agreed wage and working-time adjustments in economic crises can now be dismissed. In the Netherlands, the 2015 Work and Security Act simplified EPL without necessarily lowering it clarifying the cases in which courts or the Public Employment Agency decide upon the validity of dismissals. It has also introduced a streamlined formula for severance pay, which will considerably reduce the amount in some cases, while increasing it in others. Hence, the Work and Security Act brings a clarification rather than deregulation of EPL. In the other European countries with high shares of temporary workers (Finland, Poland and Sweden), no substantial reforms of EPL for permanent workers have been undertaken in recent years.

Temporary employment
The second possibility to address segmentation through EPL reform is re-regulating temporary contracts. Indeed, there have also been some -albeit less far-reaching -changes in this direction (OECD 2013;. One trend is to make the use of temporary contracts more expensive for employers, while at the same time offering monetary incentives to allow for transitions into permanent jobs. In Slovenia, the 2013 labor market reform raised employers' unemployment insurance contributions for temporary jobs. However, if the temporary job is converted into a permanent one, employers are exempted from contributions for a limited time. The social partners in France have agreed upon a similar reform. Since May 2013, employers pay increased unemployment insurance contributions depending on the length of the contract. Contributions increase by 3 percentage points if the contract runs for less than one month, 1.5 if one to three months and 0.5 for longer contracts. Hence, the goal is to make employers contribute to the costs of extremely flexible hiring practices. Moreover, employers in France are exempt from contributions for a limited period if they hire younger or older workers on permanent contracts. In Italy, since 2012 employers have had to pay a contribution 1.4 percentage points higher for temporary rather than permanent workers, although this is reimbursed if the job is later converted into a permanent one (see also the more recent subsidy for permanent hirings). In Spain, from 2015 onwards, temporary workers receive a higher severance pay at the end of their contract. The payment corresponds to the salary of twelve days per year of employment with the firm and is considerably lower than for permanent workers. Finally, in the Netherlands, social protection for temporary and freelance workers was improved with an Act from 2013, addressing the boundary of dependent work and (sometime bogus) self-employment.
Other restrictions of temporary work were implemented in the form of a lower maximum duration. In Slovenia, the 2013 reform limited the time for which an employer can use temporary contracts for a specific job to two years. In the Netherlands, the maximum duration was limited from three to two years in 2015. However, collectively agreed deviations are still possible, whereby it remains to be seen what the reform means in practice. 12 Some countries have made the regulation of temporary work more permissive in recent years. Italy has gone furthest in this direction, whereby a recent reform has abolished the need to indicate a valid reason for temporary employment. It has also extended the maximum duration as well as the possible number of renewals. However, firms are not allowed to use temporary contracts for more than 20 percent of the workforce. The maximum duration of temporary contracts has also been increased in Greece and Spain.
To summarize, on the face of it, there have been some important changes in the field of legislation on employment protection, an institution that was considered highly pathdependent before the crisis. Overall, there is a trend in some of the most segmented countries towards lower dismissal costs for permanent workers in combination with a moderate (and sometimes ambiguous) re-regulation of temporary contracts. Overall, this tends to narrow the regulatory gap between permanent and temporary contracts at least to some extent. Nonetheless, have these changes contributed to reducing segmentation? Given that many of the reforms have only been implemented in the past three to five years -in a situation with limited labor demand in most of the countries -it might still be too early to evaluate their impact on hiring practices with currently available data. However, revisiting Figure 1 casts some doubt on their effectiveness, showing that the share of temporary contracts has even increased in the countries that deregulated EPL during the crisis. Between 2010 and 2014, the share of workers aged 15-29 increased by roughly seven percentage points in Italy and Spain, three in France and Portugal and one in Greece. Of course, we do not have a counterfactual situation, thus making it difficult to assess the exact influence of the reforms. In most countries, the pattern of permanent vs. temporary hirings has not dramatically changed. However, rather a long-term declining trend in permanent hirings seems evident (e.g. Austria, Belgium and Luxemburg), while a cyclical component seems to dominate in countries such as Ireland or Latvia. Table 1 shows the summary statistic for an initial OLS regression model presented in Table 2. The dependent variable "share of temporary employment" is already discussed in Figure 7. Explanatory variables are GDP growth and the annual unemployment rate both measured in percent as well as the average annual wage. We further control for changes in the employment protection legislation (EPL) for temporary and permanent employment. Table 1 shows that on average 54 percent of new employment contract in one year are temporary. However, as already shown Figure 7, the range of temporary employment strongly varies among EU countries. It is further shown that within the period from 2005 to 2013, EPL slightly decreased. 14 The results of the OLS regression are presented in Table 2. In a first step, we include the macro indicators introduced above into the regression. Subsequently, we use country as well as year dummies to account for country-and year-specific fixed effects. The results show that in this simplistic model the annual unemployment rate is robustly and significantly positive related to the share of temporary employed people. Accordingly, the more people who are unemployed, the higher the share of temporary employment, which is a very intuitive result. Furthermore, an increase in EPL for temporary employment is significantly related to a decrease in temporary employment, which might also be expectable. However, after controlling for country and year fixed effects, the coefficient becomes insignificant, whereby all the other variables are less robust or insignificant. The result suggests that a more in-depth analysis is needed to explain differences in the share of temporary and permanent employment.  Studies on single countries based upon national data can provide additional evidence. Using most recent regional data from the Italian region of Veneto to analyse the effect of the Jobs Act, Sestito and Viviano (2016) show that the 2015 reform combining reduced dismissal costs and regulatory uncertainty with relatively generous hiring incentives regarding the employment of permanent workers has changed firm behavior in the expected direction, departing at least to some extent from hiring and firing concentrated among temporary employees (see also Gama et al. 2015).

Unemployment benefits
Flexicurity -or a well-managed form of labor market flexibility -also means access to reasonable social benefits and avoiding a double disadvantage of workers with short employment spells who often not only lose their job but also lack proper access to unemployment insurance benefits. The coverage of temporary workers by unemployment insurance schemes is an important aspect contributing to labor market segmentation. In many countries, the formulas by which benefit generosity and eligibility are calculated tend to disadvantage workers with short employment spells. Moreover, since benefits are often tied to former earnings, the wage penalty of temporary workers translates into worse unemployment protection. The problem is particularly severe in many Southern European countries that lack a universal safety net. For instance, Berton et al. (2012) show that nonstandard employment comes with a significant penalty in social insurance systems in Italy.
Modernizing the welfare state towards universalism is a difficult task in times of severe budget constraints. A recent step towards more universal benefits was undertaken by the introduction of Assicurazione Sociale per l'Impiego (ASPI) in Italy in 2013. ASPI replaces the rather fragmented system of unemployment benefits and makes the system more generous.
Interestingly, the reform also introduced Mini-ASPI, a special benefit with lower contribution requirements, specifically designed for including workers with short employment spells. While benefit generosity will be the same as for ASPI recipients, duration of Mini-ASPI will be lower and conditional upon the actual contribution record (Berton et al. 2015). In May 2015, the Italian government replaced ASPI and Mini-ASPI with a new scheme (NASPI) that uses rather permissive eligibility criteria close to Mini-ASPI (13 weeks of contributions in the four years before unemployment). At the same time, a new unemployment benefit scheme -DIS-COLL -was piloted for atypical workers such as project-based collaborators ("co.co.pro") in parallel with a new unemployment assistance scheme (Picot and Tassinari 2015). With the goal of better reflecting the situation of temporary workers and extending coverage, the reform represents a remarkable step and warrants careful evaluation once fully effective. It has the potential to serve as a model for countries facing similar challenges as Italy. Moreover, in France, in terms of access to unemployment benefits for workers with relatively short employment spells, since 2009, four months of employment over the last 28 months suffice to receive unemployment insurance benefits for at least four months. Finally, in Spain, access to unemployment benefits was facilitated for temporary and younger workers as well as selfemployed, albeit -similar to other countries -combined with stronger work incentives and activation policies for the unemployed. Figure 8 shows that massive differences in the share of recent unemployed people receiving unemployment benefits persist, although in some countries such as Croatia, Estonia, France, Italy, Latvia and Spain the coverage share grew during and after the crisis (albeit still far from being comprehensive) while it dropped in the UK, Sweden and Hungary. As an approximation to benefit generosity for the least well-protected dismissed workers, we can refer to the maximum duration of unemployment benefits as calculated by the European Commission (2016) (see Figure 9). Aside from Italy, we do not observe an expansion of unemployment benefit generosity for the least advantaged unemployed: in fact, in some countries, the benefit duration has been cut. To conclude, the picture regarding the development of protection through unemployment benefit looks somewhat ambiguous, with notable country differences in terms of both in levels and changes. Despite some expansion, benefit systems continue to be highly segmented in most countries.

Active labor market policies
Another possibility to address labor market segmentation is expanding ALMPs and humancapital investment in particular. In segmented labor markets, vulnerable workers have a higher risk of becoming unemployed and thus are more likely to enter into contact with public employment services (PES). Access to unemployment benefits also tends to facilitate access to ALMPs as benefit recipients remain within the reach of PES (European Commission 2015). In principle, this provides the opportunity to prepare vulnerable workers better for the labor market and improve their future employment outcomes (e.g. job stability or wages). Effective ALMPs and PESs can thus contain the consequences of higher unemployment risks by enabling quick and sustainable re-employment.
Evaluation studies show that skill-oriented ALMPs are particularly effective in this regard: they improve the chance of being in employment over the long run and hence clearly contribute to reducing segmentation (Card et al. 2015), while lock-in effects into training measures are less of an issue in times of recession.
Hiring incentives are another important element of ALMPs as they can set effective incentives to employers to hiring job-seekers. Many countries with segmented labor markets have a tradition of subsidizing hirings on a temporary basis as well as conversions into permanent positions. This can have strong short-term effects and encourage continued employment.
Job search assistance and activation policies require effective PESs. Sanctioning practices are only effective in the short run, whereas they do not lead to better long-term outcomes.
Hence, expanding investment in training and other forms of potentially effective ALMPs appears to be a plausible strategy towards improving long-term labor market integration. In fact, during the crisis, the intensification of ALMP has been on the agenda in many European countries, although budget constraints and capacity issues have influenced the actual practices (see Gama et al. 2015). To assess which member states followed this route, Figure  10 presents ALMP expenditure for training per person looking for work in 2007 and 2013, the most recent available data point. Expenditure is expressed in purchasing power standards, whereby small and large economies are comparable. The figure reveals that there is no general trend towards increased spending on training. Among the more segmented countries, the trend rather seems to proceed towards less spending per unemployed person. In some of the most segmented countries, this decline comes from a rather low pre-crisis baseline, namely in Greece, Italy, Poland and Spain. Furthermore, in Ireland, the Netherlands and Portugal the numbers point to a marked reduction in training expenditure. Among the countries with moderate-to-high shares of temporary workers, Austria, Finland, Germany and Sweden have increased training expenditure. In sum, we cannot identify a clear trend towards more investment and training. Similar trends can also be identified when looking into participation data.   In a similar fashion, spending on hiring incentives per jobseeker (as classified in the LMP database) was more limited in 2013 in most countries as compared to 2007, with Sweden and Denmark being the only notable exceptions.
PESs have also grown in importance, albeit falling short of providing sufficient individualized support in situations of high caseloads typically observed in countries with steeply increasing unemployment rates. Many countries have also reformed their PES structures over recent years, e.g. France and the UK. Regarding expenditure per unemployed, spending only increased in Germany (with a declining number of unemployed in the denominator), Denmark, Sweden and Estonia. It was neither increased nor stabilized in those countries with high unemployment and segmented labor markets such as Greece, Italy, Portugal and Spain. In particular, mobility-enhancing measures such as training, hiring incentives and intensive job counselling can help to reduce the risk of exclusion from employment. However, there are important limitations to the contribution of ALMPs towards tackling segmentation.
First, ALMP participation is always selectively allocated by the PES, whereby not all participants among the labor force will benefit to the same extent, while effective ALMPsoften allocated to the 'best risk' unemployed -may simply privilege some job-seekers over others, effectively creating an additional element of segmentation. This is particularly relevant in situations with high and increasing caseloads of job-seekers. Second, training the unemployed is most useful for persons with rather low or deteriorated skills, such as in the case of the long-term unemployed. It is more doubtful how effective it is for young people with recent educational experiences and those who have frequent employment spells in the form of temporary employment. European data shows that workers with low-skills -the prime beneficiaries of training -reflect the minority of all temporary workers in all countries (despite accounting for considerable shares in some segmented labor markets, most notably Portugal and Spain). Quite to the contrary, in some countries a sizable share of the temporary workers is highly skilled, as Figure 13 shows. Usually between onequarter and one-third of all temporary workers have tertiary education and this group is unlikely to benefit from ALMP intervention in the form of additional training. Third, temporary, targeted hiring incentives may only provide limited relief as it may also result in churning, aggravating the problem of excessive turnover among labor market entrants or those with longer sequences of temporary jobs, particularly when the macroeconomy is in a critical status and employment protection for permanent workers is rigid.
National policy-makers should thus carefully evaluate the mobility patterns of people with different skill backgrounds to decide which groups should actually be targeted by training and hiring incentives (which are rather costly policy interventions). In sum, ALMPs can make a considerable contribution to improving long-term labor market integration, although, they are not a general remedy for segmentation particularly in countries where there are insufficient jobs for skilled workers under given economic and institutional conditions. Therefore, in a situation of recession and austerity, ALMP may not be particularly effective in shortening average unemployment.

Complementarities
Complementary channels of flexibility play an important role, both at the conceptual level and with respect to observable reform trajectories. Internal flexibility by way of working-time adjustments and internal reorganization can help to stabilize employment and avoid job losses in times of crisis. Collective bargaining -particularly concerning wages -is also an important factor influencing job creation and job stability.
The use of temporary workers reflects employers' need for numerical flexibility. However, adapting staff levels is only one way to achieve it. A functional equivalent to hiring and firing is internal numerical flexibility; for instance, in the form of working-time accounts. Such accounts allow accumulating surpluses (or deficits), which are compensated with time off (overtime) later on. As such, working-time accounts help employers to respond to workload fluctuation by using temporary workers. Germany is often cited as an example with workingtime flexibility having greatly increased in the past decades (Eichhorst and Marx 2011) and playing an important role in reducing the impact of the crisis on jobs (in conjunction with government subsidies in the form of a short-time work scheme). It also should be stressed that the availability of working-time flexibility in Germany complements (and possibly limits) the use of temporary employment without making it entirely redundant. In fact, during the 2008/09 crisis, working-time adjustments could stabilize employment of about 350,000 workers in the core manufacturing workforce while the marginal workforce -comprising temporary agency workers -was reduced by a similar number. Hence, internal flexibility was combined with external flexibility, mirroring the existing segmentation of the German labor market.
Working-time flexibility is difficult to induce through public policy and it rather depends on firm-level practices and industrial relations. The feasibility and spread of such practices also strongly depends on the sector and the size of firms. It is thus better to examine actual firm practices rather than legislation. Figure 14 presents the shares of companies with ten employees or more that allow their workers to accumulate overtime. Admittedly, this is only a crude indicator that says nothing about management's and workers' discretion in using accumulated hours, although it provides a rough picture of how widespread flexible working time is in the member states.
The results reveal considerable variation in the use of flexible working time (although this can be partly related to different composition of national economies by sector and firm size). Interestingly, many countries with high shares of temporary employment rank in the bottom of the distribution and show a recent decline in the use of working-time accounts, including Croatia, Cyprus, Italy, Portugal and Spain. However, the fact that for instance Poland and Sweden combine a wide spread of working-time flexibility with very high shares of temporary employment underlines that such flexibility is not a remedy against segmentation per se. Further research will be necessary to determine which forms of working-time flexibility exactly 22 provide alternatives to temporary employment. Nonetheless, it seems clear that internal and external flexibility are linked, and that developing internal flexibility does not necessarily prevent a segmentation between the core and margin. . Note: The surveys cover companies with ten or more employees.

Challenges facing Member States in addressing labor market segmentation
In this section, we briefly point to the challenges that policy-makers face in reducing labor market segmentation, which can be grouped into three categories.
Political preferences: if segmentation is addressed though reducing the social rights of 'insiders', it can be difficult to organize democratic majorities for such proposals. EPL is a case in point, whereby it is unsurprising that deregulation only took place under severe reform pressure. Moreover, expanding social protection for vulnerable workers could be met with resistance from the majority with a low unemployment risk (the net payer) owing to the considerable costs of such reforms. However, it also should be noted that the political economy literature greatly exaggerates the extent to which public opinion is divided between insiders and outsiders. There is considerable solidarity across the insider-outsider divide (Marx 2015), to which policy-makers can appeal. Macro-economy and budget constraints: as explained above, in some countries there simply seems to be insufficient labor demand to offer secure jobs to the entire workforce. In a context of high unemployment and economic uncertainty, it is understandable if workers and policymakers view temporary employment as the lesser of two evils. The situation is exacerbated by the fact that that (political and economic) room for stimulating the economy is very limited in the context of the European debt crisis. Apart from EPL and industrial relation reforms, substantially improving the situation of the unemployed and temporary workers costs money: in the form of training for those with insufficient skills, wage subsidies, decent jobs in the public sector and adequate social protection (against unemployment and in old age). Needless to say, funding such reforms is particularly challenging in times of austerity. Policy-makers will have to look for ways to make employers contribute to the social costs of their hiring practices, without making temporary contracts so expensive that hiring is discouraged.
Evasion and non-compliance: it is a reasonable assumption that employers will always look for legal loopholes to reduce labor costs. Despite little systematic knowledge, anecdotal evidence from countries such as France, Italy and Spain also suggests that there is considerable noncompliance with labor law and that labor inspectorates are insufficiently equipped to deal with this situation. Hence, re-regulation of temporary employment might either be ineffective (if compliance is low) or employers will respond by relying on other precarious forms of employment. In many member states, a trend towards even more unregulated (economically dependent) self-employment or 'contracts for work and labor' already seems to be emerging. Another risk is that employers respond to stricter regulation with undeclared work, which would imply strong precariousness and loss of tax revenues. Regulation regarding these forms of employment has to be designed in a careful and smart way, minimizing incentives for circumvention.

Conclusions and points for discussion
The positive message of this report is that, in the aftermath of the 2008/09 crisis and rising youth unemployment, EU member states have started to take the issue of segmentation more seriously. There are plenty of policy initiatives with the goal of overcoming labor market divides, although the more pessimistic message is that they have achieved little success to date: temporary employment keeps rising and youth unemployment remains worryingly high. Admittedly, it is probably too early to evaluate some of the very recent reforms, also given the difficult macro-economic environment. Nonetheless, one should also consider the possibility that the steps taken so far are insufficient to substantially improve the labor market situation.
Accordingly, which policy options exist and how suitable are they against the background of the identified challenges?
While banning or prohibitively regulating temporary contracts might appear as a solution, there is clear evidence that temporary employment can serve as a bridge to permanent jobs. Eliminating this option would also create strong incentives to use potentially more 'precarious' types of work such as economically dependent self-employment. Indeed, there is little doubt that negative repercussions of temporary employment are dwarfed by the devastating sociopsychological as well as political effects of long-term unemployment. To the extent that temporary contracts serve as entry points into the labor market for young workers or longterm unemployed, banning them probably does more harm than good.
Another frequently proposed policy solution is the deregulation of permanent contracts. This solution definitively has some value since it would remove the labor market rigidity that leads to a visibly unfair distribution of job security and market risks and limits transition possibilities to permanent jobs. Under such a regime, permanent contracts would be less permanent, although the reliance on temporary contracts would also decline. Hence, a vulnerable worker would not be less vulnerable if his or her temporary contract is replaced with a flexible 'permanent' employment contract, although the chance of being hired (and eventually dismissed) on a permanent contract would be higher. However, if deregulation does not go far enough, incentives to hire on temporary contracts remain. Member states will thus have to try to find a moderate level of EPL that balances both goals, whereby recent EPL reforms can be interpreted as steps in this direction. Similar reservations can be formulated for the frequently proposed "single-employment contract". Here, the idea is to replace temporary and permanent contracts with a unified legal framework in which dismissal protection is phased-in with tenure. However, the problem is that such a framework produces thresholds in dismissal costs below that employers' prefer to replace a worker with a new one whose protection starts from zero. This effectively means that newly recruited workers still face the same insecurity, at least for some time. In addition, if thresholds are implicit and employerspecific, workers do not even know how long they can stay with the firm and when they make a transition into secure employment. Depending on how the single-employment contract is modelled, it is quite possible that it makes the situation of workers with short tenure even more precarious than it would be on a temporary contract. Finally, in the real-world situation, it is difficult to imagine an arrangement without any option to have temporary contracts under certain conditions, meaning that some duality will remain.
A factor that can mitigate effects of temporary contracts is the prospect of making a successful transition into stable employment. On the other hand, temporary contracts are a real social and political problem if workers see themselves in a 'dead end'. For policy-makers, this means that ALMPs (and training in general) should be at the core of any strategy to improve the situation of low-skilled temporary workers who frequently experience unemployment. At least in countries with relatively well-functioning labor markets, governments should make every effort to ensure that lacking or obsolete skills do not 'trap' workers in chains of unemployment and precarious jobs. However, such efforts are somewhat less promising in countries where there are insufficient jobs, even for university graduates. The use of subsidies to encourage conversions of temporary into permanent contracts is a plausible and quite 25 widely used alternative, although it needs careful monitoring and evaluation. However, we can expect that mobility-enhancing investments in human capital are most promising if employment protection is reformed in a way whereby mobility is encouraged and transitions to long-term employment are facilitated.
Another important concern for policy-makers should be avoiding multiple disadvantages through the welfare state. The modernization of unemployment and pension systems towards more universal schemes would not only improve the socio-economic situation of temporary workers but also increase the legitimacy of the dual employment model. If society compensates temporary workers for their higher risk by offering decent and accessible social security, this could lessen feelings of political marginalization or exclusion. As the fiscal situation in most affected member states does not leave much leeway for social policy expansion, the politics necessary for such changes are complex and probably involve zero-sum conflicts. However, the recent Italian reforms towards more universal unemployment benefits demonstrate that these are not insurmountable hurdles.
However, an important objection to 'fixing' the temporary employment issue with social benefits and labor market policies is that turnover costs are shifted from employers to society. Another way of compensating temporary workers for their risk is granting them 'insecurity bonuses', as achieved in France in the form of the prime de précarité. In this system, employers have to pay a bonus to the worker (amounting to ten per cent of the total gross wage paid under the contract) if the employment relationship is not continued after the contract has expired. As mentioned, since 2013 there has also been a higher employer contribution rate to the unemployment insurance fund in case of short fixed-term contracts. This system ensures that employers contribute to the social costs of their hiring practices and increases incentives to use temporary contracts responsibly. In addition, it might reduce the perception among temporary workers being disadvantaged in several dimensions (unemployment risk and lower wages). For such a system to be effective, it would be important that, first, the bonus is sufficiently high to substantively affect employers' incentives and workers' material situation (albeit without totally discouraging hiring). Second, it has to be ensured that there are no legal loopholes that allow circumventing the bonus. Third, -and a more general point -the equalpay principle has to be vigorously enforced through monitoring by unions and labor inspectors as well as through severe sanctions in case of non-compliance; otherwise, employers can cancel out the insecurity bonus by lowering temporary workers' wages, a practice that already seems to be widespread.
An alternative way to provide temporary workers with employer-financed social security is the Austrian Mitarbeitervorsorgekassen. In 2003, severance payments were replaced with a system in which employers contribute on a monthly basis to an account for each worker. When the employment relationship is terminated, workers can withdraw the money from their account (instead of a severance payment) or carry it over to their next job and use it as pension savings. An advantage of this model is that also self-employed can be integrated, which is the case in Austria since 2008. Besides providing additional financial security to 26 insecure workers, the system also overcomes stark differences in separation costs. However, one has to note that there is no incentive to employers to refrain from dismissals.
There is no magical formula to address labor market segmentation, with fiscal, economic and political circumstances significantly limiting room for maneuvre. Member states should nonetheless continue to carefully modernize labor law and welfare states, even if it is only possible in an incremental way. Radical deregulation of permanent contracts or re-regulating of temporary contracts should be avoided. Policy-makers should also resist the temptation to cut ALMPs and training programs in particular. Given its favorable long-term consequences, training programs targeted at the most needy should be an important pillar in any strategy to address segmentation. Member states should continue to experiment with fiscal incentives to convert temporary into permanent contracts, while carefully evaluating the schemes and making their survival conditional upon success. Finally, it will be important to move towards a better social protection of temporary workers. For economic and symbolic reasons, employers will have to make a relevant contribution to this. Austria, France and Italy exemplify different approaches that could be developed into more broadly applicable strategies to address segmentation.