Selected Issues

Abstract

Selected Issues

Labor Market Reform in Slovenia: An Assessment1

A. Introduction

1. Amid deteriorating labor market outcomes as the crisis deepened, Slovenia embarked on a labor market reform in 2013, aimed to ameliorate the immediate adverse impact on employment and address the root causes of labor market malfunction and pervasive segmentation. This chapter attempts a preliminary assessment of the effects of the 2013 reform.

2. The discussion is structured in a way that can be useful for drawing policy implications. Specifically, issues to be addressed include:

  • What were the salient features of Slovenia’s labor market that account for its unsatisfactory performance before the reform?

  • How did the 2013 reform address these shortcomings? Has the mix of the reform measures been appropriate, or is there a case for some recalibration?

  • Did the reform carry unintended negative side effects, and how should these best be addressed?

  • Have there been areas of suboptimal labor market performance that the reform neglected, and how should these be addressed?

3. The chapter is organized as follows: The next section summarizes the performance of Slovenia’s labor market during the crisis, and highlights features that could have contributed to these outcomes. This is followed by a description of the 2013 reform and its objectives. An assessment of the reform’s impact is then carried out via complementary methodologies involving: relevant macro indicators; gross labor flows; and a structural model. The final section draws some key policy conclusions.

B. Context and Background

4. Slovenia’s labor market outcomes deteriorated sharply with the onset of the 2009–13 crisis. In addition to the large adverse aggregate demand shock experienced by Slovenia, the crisis exposed underlying rigidities, which prevented a smooth labor market adjustment and did not facilitate the substantial labor reallocation that would be called for in response to the sectorally asymmetric nature of the shock. In addition, the major increase in the minimum wage introduced in 2010 added to the burden on labor market adjustment2

5. These factors contributed to worse labor market performance in Slovenia compared to other EU countries. Specifically, Slovenia experienced a sharper increase in unemployment and deeper reductions in employment that the average EU member (chart below). This deterioration has been even more pronounced when compared to relative output performance – even as the latter overstates the relative autonomous demand shock since it (partly) reflects the labor market performance itself. While Slovenia under-performed the EU average in terms of both employment and unemployment, its relative employment deterioration was much more pronounced; this feature will be taken up in this chapter’s analytical and policy discussion.

uA03fig01

Employment rate

2008–13 change, in percent

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT.
uA03fig02

Unemployment rate

2008–13 change, in percent

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT.
uA03fig03

Slovenia youth Temporary Employment

(Percent of total)

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT.

6. These aggregate trends relate importantly to labor market segmentation. Over the past decade, Slovenia’s labor market has become progressively more segmented between a “core” segment of employees on open-ended contracts enjoying heavy employment protection (both by EU standards and, especially, by the standards of countries at comparable stages of development) and a segment dominated by short-term, temporary labor contracts. The latter segment has been increasingly relied on by employers to provide a missing “margin of flexibility” in response to sectoral or cyclical shocks, with temporary employment becoming increasingly pervasive particularly for the young, the single most important pool of entry to the labor market. Thus, the share of youth (15–24) employment covered by temporary contracts had been steadily increasing in the run-up to the crisis (chart).

7. The incidence of temporary contracts continued to grow through the crisis. Following a brief reduction in the immediate wake of the 2008 global financial crisis, as employers resorted to layoffs of temporary workers to adjust total employment, the share of youth temporary contracts started rising again, eventually exceeding its pre-crisis peak. By the trough of the recession in early 2013, temporary contracts made up almost ¾ of youth employment, making Slovenia an outlier within the EU (chart).3

uA03fig04

Youth Temporary Employment

(Percent of total, 2013Q1)

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT.

8. The prevalence of youth temporary contracts rendered this segment particularly vulnerable to the crisis. Thus, the youth (and the low-skilled) experienced a particularly significant deterioration in terms of employment and unemployment rates, compared both to the overall Slovenian labor force and to corresponding EU averages. Once again, the relative employment deterioration appears to have been comparatively more substantial.

uA03fig05

Slovenia: Employment Rate

2008–13 Percent Change

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT.
uA03fig06

Slovenia: Unemployment Rate

2008–13 Percent Change

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT.

9. Pervasive labor market duality, with a high incidence of temporary employment, arguably carries significant welfare costs, from both a near- and a longer-term perspective. At typical business cycle frequencies, it could lead to excessive employment fluctuations (and sub-optimal risk sharing among different groups) – a pattern consistent with a broad class of analytical models but also with Slovenia’s crisis experience. From a longer-term perspective, it could entail sub-optimal human capital acquisition (particularly if temporary employment is concentrated among the young, as in Slovenia), impairing growth prospects. Against this background, it is not surprising that addressing duality without hampering employment prospects was a major objective of the 2013 labor market reform.

10. A number of policy measures to stem employment losses in the early phase of the crisis had rather limited impact. These efforts were codified in two intervention laws aimed at preserving jobs (via shortening working hours and allowing temporary layoffs)4 and at expanding the scale of active labor market ALM) policies, albeit from a rather low base. While Slovenia’s employment losses would probably have been even worse in the absence of these policies, their overall impact was limited and of rather short duration: with regard to ALM policies in particular, the number of persons involved remained generally low (and actually declined during 2011–12), as these programs ultimately ran into financing constraints in the face of Slovenia’s erosion of fiscal space and loss of sovereign market access.

C. The 2013 Labor Market Reform

11. The 2013 reform was relatively comprehensive in nature, and aimed at addressing the root causes of Slovenia’s labor market problems. Specifically, the reform’s main objective was to limit duality by increasing employers’ flexibility in handling open-ended contracts while strengthening legal protections for temporary employees. Overall, the reform entailed a range of policies targeted at both the open-ended and temporary segments of the labor market, aiming to expand the scope of the former via lowering employment protection and offering fiscal incentives for open-ended employment contracts, while reducing some of the advantages (to employers) of temporary contracts.5

12. Reforms to open-ended contracts included the following features:

  • Simplification of individual dismissal procedures – changes included allowing temporary layoffs and compensation in cash instead of reintegration and a shortening of dismissal notice periods;

  • Reduction of firing costs – entailing shorter notice periods (from 120 to 60–80 days) and lower severance payments;

  • Some (mostly temporary) exemptions from social contributions – these included: a 2-year exemption from contributions to unemployment insurance for new employees hired on open-ended contracts; exemptions for part of the employer pension and health contributions for hiring younger or older workers on open-ended contracts; a 2-year exemption from social contributions for hiring young, previously unemployed persons.

13. Reforms to temporary contracts included the following features:

  • Limits on the scope of repeated (so-called “chain”) contracts – with temporary contracts allowed to be renewed up to two years for the same work, after which the employee had to be offered an open-ended contract or dismissed;

  • Increase in some social contributions and severance pay for employees hired on temporary contracts;

  • Student work made subject to social contributions – effective from early 2015 on.

14. While overall changes in employment protection legislation under the reform were non-trivial, in some respects they arguably did not go far enough. The most substantial reduction in employment protection relates to the regime governing individual dismissals of workers under open-ended contracts: in this area, regulations have become significantly less restrictive, with the relevant index falling slightly below the OECD average – although still exceeding some other transition economies and some countries of comparable income levels (chart). On the other hand, regulations concerning temporary contracts were tightened, and remain somewhat more restrictive than the OECD average. More importantly, the regulatory regime regarding collective dismissals was not changed by the reform, and remains relatively restrictive.

uA03fig07

EPL Index - Individual Dismissals, 2013

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: OECD

15. Post-reform aggregate labor market outcomes improved somewhat, but the picture looks less bright when adjusted for the cycle. Thus, Slovenia’s employment growth and reduction in unemployment both actually fell short of the EU average Figure), despite Slovenia’s stronger post-2014 output recovery relative to the EU.

uA03fig08

Employment Rate

2013–16 change, in percent

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT.
uA03fig09

Unemployment Rate

2013–16 change, in percent

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT.

16. On the other hand, the reform had some of the intended impact on labor market duality, but this effect appears to have been largely transitory. Thus, while new contracts in the period leading up to the reform had been predominantly temporary, in the immediate wake of the reform this pattern reversed, with new open-ended contracts dominating. Specifically, over the first 12 post-reform months, the annual growth of new open-ended contracts substantially outpaced temporary contracts, especially for the youth segment of the labor market; on the other hand, the near-zero post-reform total new youth contracts suggest unintended side effects of the reform that will be taken up in subsequent discussion. However, this break with previous trends appears to have been largely confined to the immediate post-reform window, with the relative incidence of the two contract categories broadly reverting to the pre-reform picture in subsequent periods.

uA03fig10

New Employment Contracts, total

Post-reform, 12-month percent change

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: IMAD.
uA03fig11

New Employment Contracts, youth

Post-reform, 12-month percent change

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: IMAD.

17. Alternative summaries of post-2013 data broadly confirm a mostly transitory impact of the reform on labor market duality. Year-on-year changes in new contracts could exaggerate the picture for the second and third post-reform years, as they may be distorted by high base of new open-ended contracts (and low base of new temporary contracts) of the immediate post-reform widow. To correct for this, we looked at the evolution of the share of open-ended and temporary new contracts in total new contracts, pre- and post-reform (chart below).6 While the picture is naturally less dramatic compared to year-on-year changes, it confirms the steady decline of the reform’s impact on labor market duality. It could also be argued that focusing on new contracts would tend to bias the results, as temporary contracts, by their nature, tend to be terminated and renewed more often than open-ended ones. Looking instead at shares of temporary contracts in the total stock of contracts rather than the flow of new ones would correct for such differential job churning. In this regard, data confirm that progress in reducing duality has been declining post-reform: while the share of temporary employment in total employment edged down in 2014, it resumed its upward trend subsequently, reaching an all-time high by 2016 (chart).

uA03fig12

Slovenia: Temporary Employment

(percent of total employment]

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

18. While illustrative, the above trends provide an incomplete picture of the reform’s impact on labor market dynamics. The following sections attempt a more systematic assessment, via complementary methodologies drawing on macro evidence, in particular gross labor market transitions.

uA03fig13

Total Permanent and Temporary New Contracts

(Percent of total new contracts)

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: IMAD.
uA03fig14

Youth Permanent and Temporary New Contracts

(Percent of total new youth contracts)

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: IMAD.

D. Reform Impact—Macro Evidence

19. This section draws on some standard macro relations to assess the post-reform functioning of Slovenia’s labor market. Specifically, two broad questions are of interest in this context:

  • Has growth become more “employment-friendly” since the reform?

  • Has labor market matching improved since the reform?

Employment-growth relationship

20. Low employment-growth elasticities and a highly procyclical labor productivity can reflect labor market distortions. In particular, high employment protection can induce firms to hoard labor during contractions and rely heavily on overtime during expansions – these patterns appear to have characterized Slovenia in the pre-reform period. Given the reform’s emphasis on loosening employment protection regulations, it is of interest to explore whether the employment-output relation has shifted in the expansionary post-reform period relative to past trends.

21. Available evidence does not suggest that growth became more “employment-friendly” post-reform. Given the sensitivity of the output-employment relation to the cycle, we distinguish between 3 sub-periods: the crisis period (2009–13); the post-reform expansion period (2014–16); and an equal 3-year pre-crisis expansion period (2006–08). The hypothesis of interest is whether the post-reform expansion leads to higher employment growth relative to the pre-crisis (and pre-reform) expansion. While the data confirm broad employment-output co-movement over the full sample, there are indications of shifts in the relation over the sub-periods considered. Calculating average sub-period employment-output elasticities to rule out transitory year-specific effects does not provide support to a salutary impact of the reform (chart): if anything, the calculated elasticity for the latest sub-period is slightly lower when compared with the pre-crisis expansion, although with so few data points it is not possible to measure the statistical significance of the finding. The sharp drop in the elasticity relative to the crisis period suggests that extensive layoffs during the crisis mainly via termination of temporary contracts gave way to more intensive labor utilization (as opposed to significant new hires) as the economy transitioned to recovery, despite the reform.

uA03fig15

Output and Employment Growth

(Annual percent change)

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT.
uA03fig16

Average Employment Elasticity

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT.

22. The above evidence raises doubt on whether the reform has achieved adequate progress in lowering effective employment protection. Certain caveats are certainly warranted. It is possible that differences in labor intensities across sectors that expanded strongly before and after the crisis (e.g., construction vs. export-oriented manufacturing) partly account for the observed differences in employment elasticities across sub-periods. However, a cursory look at relative labor intensities between construction and manufacturing makes it unlikely that they constitute a significant part of the explanation. With these caveats, however, one is left with suggestive evidence that the loosening in employment protection under the reform has not been adequate to clearly impact output-employment dynamics over the cycle.

Matching job seekers and vacancies

23. Improved labor market flexibility should yield better labor market matching between unemployment and vacancies. This would imply that, at least eventually, one would expect a reduction in unemployment for a given level of vacancies, i.e., an inward shift of the Beveridge curve. This section examines Slovenia’s experience in the post-reform period against such a benchmark.

24. In addition to a spike in cyclical unemployment, the 2009–13 crisis appears to have been associated with a sharp deterioration in labor market matching. This is reflected in a major outward shift of the Beveridge curve that started in early 2009 and continued through the trough of the recession in mid-2013 (chart). While the magnitude of the shift is striking, some matching deterioration would have been expected given the large asymmetric sectoral shocks associated with the crisis – with demand shifting sharply away from some large nontradable sectors (notably construction). Under these conditions, it should have been expected that it would take some time for the labor market to restore matching between the new skills demanded and the unemployed.

uA03fig17

Slovenia: Beveridge Curve

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT.

25. What is more surprising is that labor market matching has apparently failed to improve despite the increased flexibility fostered by the reform. In fact, the Beveridge curve is at its end-2011 position almost 4 years after the reform and 3 full years into the recovery. While Slovenia is not alone failing to achieve an improvement in labor market matching well into its recovery, quite a few countries of comparable income levels and flexible labor markets have seen their Beveridge curves shift back in as they exited their recessions (charts on page 42). In addition, Slovenia’s Beveridge curve appears to have become relatively steep, implying an unfavorable vacancy-unemployment tradeoff: it now takes a sizeable increase in vacancies to bring about a moderate decline in unemployment.

26. Some robustness analysis suggests a more nuanced assessment of Slovenia’s labor market mismatches, but does not change the broad picture. Close observers of the Slovenian economy caution that vacancy data should be used with care, as they were estimated (rather than directly reported) during the 2014–15 period. Even though excluding the period in question would still suggest substantial remaining mismatches by end-2016 relative to the pre-crisis period, it would be worthwhile to check the robustness of this conclusion by using an alternative indicator of employer job postings that does not suffer from such measurement issues. One such measure is the Labor Shortage Indicator (LSI), which measures the share of employers who report labor shortages, and which has been consistently reported in Slovenia throughout the period. Clearly, this indicator is not a perfect substitute for a proper vacancy rate: it does not provide information on the level of vacancies, and moreover, it covers only the industry. With these caveats in mind, the adjacent chart presents an alternative Beveridge curve using the indicator in question as proxy for job openings. While the relation between LSI and unemployment is much noisier relative to the standard Beveridge curve, it conveys a somewhat more benign picture of Slovenia’s labor market matching, with this alternative Beveridge curve showing signs of shifting back since the trough of the recession. Nonetheless, even this indicator suggests that substantial mismatches remain: by end-2016, for given LSI, unemployment was still some 2½ percent higher relative to the pre-crisis period. This suggest the basic conclusions are reasonably robust.

uA03fig18
uA03fig18

Slovenia: Alternative Beveridge Curve

(Based on LSI)

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT

27. While insufficient liberalization may be partly responsible, structural skill mismatches are probably also contributing to the stickiness of the Beveridge curve. As also argued above, insufficient progress towards loosening employment protection could be rendering employers in the sectors facing increased demand still reluctant to hire from the pool of the unemployed, preferring instead to utilize their existing work force more intensively. However, the size of the shift in the Beveridge curve relative to the pre-crisis period suggests that it is unlikely that this is the whole story. Rather, the major shifts in the structure of the economy since the onset of the crisis may have generated structural skill mismatches that may not be easily corrected via market forces alone. While this hypothesis is difficult to test conclusively, the changing composition of vacancies in terms of sectors as well as qualifications sought by employers provides suggestive evidence. In that case, while the types of policies pursued under the 2013 reform probably need to go further in the direction of liberalization, a different range of initiatives targeted at skill acquisition may need to be considered in parallel to achieve a substantial and durable improvement in labor market matching.

E. Reform Impact—Evidence from Labor Market Transitions

28. This section draws evidence from labor market flows between employment-unemployment-inactivity to achieve a more refined assessment of the impact of the 2013 reform. Compared with relying only on employment and unemployment figures, information on flows between all possible states of the labor market provides a better picture of key adjustment mechanisms at work, allows a fuller assessment on how specific policies impact the labor market, and provides intuition about transition margins in the labor market where distortions may persist and hence additional policy interventions may be called for. These advantages, and increased availability of finely detailed data, have made this type of analysis the methodology of choice to study a wide range of labor market issues – see, for example Kugler and Saint-Paul (2000) on cross-country differences in labor market adjustment; Hall (2005) on distinguishing between labor market models; Fujita and Nakajima (2016) on identifying the cyclical impact on labor market adjustments; Diamond and Sahin 2016) on the relation of gross hires to labor market tightness in a “matching function” framework.

29. The results of this type of analysis are usually summarized in terms of relative transition frequencies. Transition frequencies capture the size of flows into a particular labor market state in a period from a (possibly different) labor market state in the previous period, as a percent of the previous period’s population size of the originating state:

Tij,t = 100*(Flow from state j to state i in t)/(Population of state j in t-1), i, j ∊ {Employment, Unemployment, Inactivity}

In a stochastic setting, these transition frequencies correspond to the more familiar conditional probabilities: Tij,t would correspond to the probability πij,t of transitioning to state i in t conditional on being in state j in t-1.

30. The analysis relies on a recent cross-country database on labor market transitions developed by Eurostat. The database draws on a large representative longitudinal sample from the European Union’s Statistics on Income and Living Conditions SILC) survey, which provides data on labor market transitions for 24 EU countries over the 2003-16 period; this combination of time series and cross country dimensions is particularly well-suited for policy analysis, as it allows one to contrast transition frequencies pre- and post-reform, while also offering a comparative perspective against a sample of similar countries with a large common cyclical component. This database has been used extensively in OECD research on a broad range of labor market issues – for recent examples see Cournède et. al. (2016) and Garda (2016). A particularly attractive feature of the SILC database for the issues at hand is that it provides a decomposition of total employment into self-employment, open-ended employment, and temporary employment, which will be relied on extensively in the discussion that follows.

31. As elsewhere, increases in cyclical unemployment in Slovenia are not dominated by layoffs of previously employed workers but rather by inability of new labor market entrants to find jobs. Indeed, Slovenia’s transition frequencies from employment to unemployment are virtually identical between the crisis and the 2014-16 recovery period (chart). This is also a central finding of Hall (2005) for the US, and has been confirmed in studies of other countries as well. While Hall (2005) interprets this result as raising doubt about the importance of wage rigidities in explaining cyclical unemployment, for the issues at hand it is perhaps more important to interpret it as cautioning that simply looking at employment and lay-off series tells us very little about unemployment dynamics, as most of the action relates to new entrants to the labor market unable to get jobs. Thus, one has to look at transition frequencies from all labor market states to capture the mechanisms at work. This should also be kept in mind in assessing the impact of policy reforms, and highlights the usefulness of the methodology.

uA03fig19

Transition Frequencies: Employment → Unemployment 2009–2013

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT
uA03fig20

Transition Frequencies: Employment → Unemployment 2014–2016

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

32. Regarding the reform’s goal to address labor market duality, transition analysis confirms a substantial but transitory shift from temporary to open-ended contracts on impact. Indeed, the transition frequency from temporary to open-ended employment jumps by almost 7 percentage points between the pre- and post-reform periods, with Slovenia moving from the middle to the upper quartile of the sample distribution (chart). At the same time, this increase relates only to the year after the reform (a 13 percentage point jump in transition frequency), as the transition frequency essentially returns to pre-reform levels during 2015–16 (charts).7 While uncertainty about the robustness of the recovery may have played a role, this result points to a potential problem with the chosen mix of the reform policies: with the effective loosening in employment protection viewed as inadequate, employers are mainly responding to the (temporary) fiscal incentives offered under the reform, but return to the pre-reform contract structure once other considerations outweigh these incentives.

uA03fig21

Transition Frequencies: Temporary employment → Permanent employment 2009–2013

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT
uA03fig22

Transition Frequencies: Temporary employment → Permanent employment 2014–2016

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

uA03fig23

Slovenia: Transition Frequency: Temporary employment → Permanent employment

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT

33. Transition analysis reveals further unintended consequences of the reform’s policy mix. The (albeit transitory) shift from temporary to open-ended employment is not in fact the whole story. At the same time, there is a structural shift from temporary employment to unemployment, moving Slovenia from the middle of the sample distribution pre-reform to near the top of the distribution post-reform (chart) – the relative constancy of transition frequencies between the two sub-periods implies a substantial structural deterioration (given the very different cyclical position of the two sub-periods), adding to the NAIRU. This shift mainly reflects the impact of the cost equalization between temporary and open-ended contracts, or the provision that after two temporary contracts the employer has to either offer an open-ended contract or dismiss the worker. Both provisions, while theoretically sound, reduced the attractiveness of temporary workers and weeded out the ones with low productivity, while the loosening of employment protection, aimed to induce employers to offer open-ended contracts, was perceived to be insufficient.8 Another major related shift concerns the transition from self-employment to open-ended employment, with the corresponding transition frequency dropping by some 7 percentage points from pre- to post-reform (charts below): it appears that previously temporary workers replaced part of the self-employed who would normally have been offered an open-ended job. Finally, one observes a major shift from open-ended employment to self-employment, with Slovenia moving from the bottom quartile to almost the top of the distribution and the corresponding transition frequency almost tripling (charts on page 47): it appears that self-employment has increasingly become a fallback destination for those who could not find an open-ended job, possibly also reflecting employer attempts to seek additional margins of flexibility. Overall, these substitution patterns to make room for temporary workers provide further evidence that the fiscal elements of the reform (rather than looser employment protection) are the dominant influence behind employer behavior, pointing to potential problems with the reform’s policy mix.

uA03fig24

Transition Frequencies: Temporary employment → Unemployment 2009–2013

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT.
uA03fig25

Transition Frequencies: Temporary employment → Unemployment 2014–2016

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT
uA03fig26

Transition frequencies: Self-employed → Permanent employment

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT
uA03fig27

Transition frequencies: Self-employed → Permanent employment 2014–2016

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT
uA03fig28

Transition frequencies: Permanent employment → Self-employment 2009–2013

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT
uA03fig29

Transition frequencies: Permanent employment → Self-employment 2014–2016

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT

34. Transition analysis identifies a number of other areas in which Slovenia is an outlier and which suggest underlying distortions that may warrant policy intervention. Perhaps the most striking is the persistently very low transition frequency from inactivity to employment, with Slovenia consistently at the very low end of the sample distribution with an average transition frequency under 5 percent (chart), and which has apparently not been impacted by the reform. While outside the purview of the 2013 reform, the scale of the problem would warrant focusing on the tax wedge at the entry level of the labor market and consider policy options to correct distortions.

uA03fig30

Transition Frequencies: Inactivity → Employment 2006–2016

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT.

35. A final noteworthy point to observe is the steady rise in transition frequencies from unemployment to unemployment. This translates into a persistent lengthening in average unemployment duration, which, while not out of line with EU trends, is all the more striking given the healthy recovery of the past three years. This steadily increasing incidence of long-term unemployment has recently come into policy focus – see OECD (2016). While policies to foster further labor market liberalization and correct unintended consequences of the recent reform would go some way towards addressing the problem, they may need to be supplemented by targeted initiatives to upgrade the skills of the long-term unemployed.

uA03fig31

Transition Frequencies: Unemployment → Unemployment 2009–2013

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT
uA03fig32

Transition Frequencies: Unemployment → Unemployment 2014–2016

Citation: IMF Staff Country Reports 2017, 126; 10.5089/9781484300749.002.A003

Source: EUROSTAT

F. Reform Impact—A Structural Model

36. This section draws on a simple structural model to shed additional light on the impact of the 2013 reform. Structural modeling has the advantage of grounding some of the stylized facts on labor market transitions of the previous section in a rigorous framework, thus allowing an evaluation of the causal content of the hypotheses of the previous section. It also allows for micro-founded welfare analysis that can inform a normative assessment of the components of the 2013 reform.

37. The model uses as a starting point the analytical framework of the seminal Blanchard and Landier (2002) paper (B-L henceforth). The B-L framework explicitly addresses labor market dualism by exploring the implications of introducing temporary contracts alongside highly-protected open-ended contracts. In this setting, B-L motivates temporary contracts as a screening device in an environment of worker heterogeneity in terms of productivity and asymmetric information, whereby firms cannot observe worker-specific productivity prior to employment. In this setting, allowing temporary contracts gives employers the option to hire workers in entry-level, temporary jobs, find out how good the resulting matches are, and decide whether to keep them in higher-productivity, permanent jobs.

38. Under plausible parametrizations, the B-L framework could imply that introducing temporary contracts may entail perverse labor market outcomes. Two underlying effects work in opposite directions: on the one hand, employers should now be more willing to hire workers to see how they perform; on the other hand, access to temporary contracts could render employers more reluctant to keep workers in permanent jobs – even if a match turns out profitable, a firm may still prefer to fire the temporary worker while the firing cost is low and take a chance with a new worker. Plausible parametrizations led B-L to conclude that this latter, perverse outcome is quite likely. In that case, labor market outcomes would include: more low-productivity entry-level jobs and fewer regular jobs (i.e. more pronounced labor market duality), lower overall productivity and output, and higher turnover in entry-level jobs. From a welfare perspective, the B-L results are even sharper: even if unemployment is reduced, workers may still be worse off as they may have to go through many spells of unemployment and entry-level jobs before obtaining a regular job.

39. For the purposes of this chapter, we made a few extensions to the B-L framework. The B-L modeling assumption of prohibitively high employment protection in open-ended contracts appears unduly restrictive, while the B-L framework leaves out some policy instruments that were important components of the 2013 reforms. Accordingly, B-L was extended in two directions:

  • Fully endogenize open-ended contracts in the firms’ optimization problem: By treating layoff costs on open-ended contracts as large but finite, firms are modelled as jointly determining the terms for regular contracts (employment/wages) together with those for temporary contracts. This modelling strategy is analytically more satisfactory, and renders the analysis more relevant to the features of the 2013 reform.

  • Incorporate payroll taxes on regular and temporary work as additional policy variables: In addition to constituting important features of the 2013 reform, inclusion of payroll taxes into the analysis can shed light on the likely labor market impact of the income tax reform currently being implemented by the authorities.

    While these modeling extensions substantially complicate the analysis relative to the B-L model, they also allow much richer welfare and policy evaluation.

40. The mechanics of the extended model deviate from B-L in some important respects. As in B-L, the extended model (details and derivations in the Annex) follows the search-theoretic tradition, whereby workers with heterogeneous (and ex ante unobservable) productivity are matched with potential employers. Worker productivity is subject to random shocks that can give rise to layoffs; however, in an important departure from B-L, the model allows layoffs both out of temporary and out of regular workers (albeit at a higher cost relative to the former). Firms and workers interact in a repeated Nash game and optimize on the basis of the probability distribution of the shocks and of preference, technological, and policy parameters – the latter including layoff costs and payroll taxes – to determine labor flows and wages. In this richer environment, the model endogenously pins down two key productivity thresholds: (i) the match-specific productivity at which firms are indifferent between ending temporary jobs and converting temporary into permanent jobs; (ii) the match-specific productivity at which firms are indifferent between dismissing and retaining workers under open-ended contracts. In turn, these thresholds are an important determinant of firms’ optimal wage and employment decisions under the two contract types.

41. A number of specification choices simplify the model’s solution method. These include imposing risk neutrality on both workers and firms, allowing flows into employment from the pool of unemployed only (i.e. ruling out inactivity-employment transitions), and ignoring transition dynamics (e.g. via adjustment costs), in order to facilitate closed-form solutions that are amenable to comparative statics analysis. These restrictions, which are shared with B-L, do not qualitatively affect the main results of interest. The higher complexity of the model necessitates an additional, somewhat less innocuous, restriction of setting the expected utility of the unemployed to zero.

42. The model’s equilibrium conditions highlight the joint determination of the main variables of interest across contractual arrangements. The model determines endogenously steady state unemployment, as well as wages and employment for workers under temporary and open-ended contracts. In contrast to B-L, where permanent employment is largely exogenous to factors affecting temporary contracts, in the extended model the equilibrium wage and employment terms of both types of contracts are determined as the solution of a single optimization problem, and thus depend on the full range of relevant variables: both productivity thresholds defined above, as well as the model’s policy variables and structural parameters. This feature of the model has an important bearing on its policy implications.

43. The model’s policy implications provide strong causal support to the main hypotheses advanced in the previous section to explain post-reform labor market transitions in Slovenia. While some of the B-L results are confirmed, the richer environment of the extended model leads to the rejection of others, and also allows assessment of policy changes (specifically as regards payroll taxes) that, by construction, the B-L model cannot address. In particular:

  • The B-L result on the impact of lowering layoff costs on aggregate labor market outcomes continues to hold under the extended model: Lower layoff costs for open-ended contracts tend to increase both hiring and dismissals, with an ambiguous net impact on aggregate employment and unemployment.

  • On the other hand, in contrast to B-L, the extended model removes the ambiguity regarding the impact of lowering layoff costs on labor market segmentation: Lower layoff costs for open ended contracts lead to higher conversions of temporary jobs to regular jobs, as well as a narrowing of regular-temporary wage differentials, thus implying an unambiguous reduction in labor market duality.

  • A reduction in payroll taxes for regular contracts unambiguously reduces unemployment as well as labor market segmentation.

  • Under plausible calibrations of the model,9 an increase in payroll taxes for temporary contracts tends to reduce labor market segmentation but increase unemployment10 At the same time, simulation results suggest a strong interaction with the extent of employment protection: for a given payroll tax increase, the adverse impact on unemployment is smaller, and the favorable impact on duality larger, the lower the level of layoff costs on permanent contracts.

44. Overall, the extended model is broadly consistent with Slovenia’s post-reform labor market transitions. Viewed through the lens of the model, the stylized facts of the previous section can be interpreted as a shift to a new equilibrium under the impact of the various components of the reform. Regarding labor market duality in particular, the observed improvement is consistent with the model’s predicted combined impact of lower employment protection for regular contracts and reduced (increased) social contributions on regular (temporary) jobs. At the same time, the model can account for the limited, and (mostly) transitory, improvement in labor market segmentation by the temporary nature of social contribution cuts on regular jobs, as well as by too little liberalization of employment protection restrictions in the overall reform package.

G. Policy Implications

45. The assessment of the 2013 labor market reform suggests a mixed record. While the reform design appropriately identified rigid employment protection as a key root cause behind the pervasive labor market dualism, the outcomes were mixed. The mix of policies would thus need to be recalibrated to support improved aggregate outcomes, render the reform’s objectives more durable, and correct unintended side effects. In addition, a number of important problem areas that were kept out of the reform agenda would need to be addressed.

46. A major policy message from the analysis of this chapter is that the role of faster and deeper loosening of employment protection regulations should be increased. This is key in order to ensure that the initial gains in reducing dualism are sustained, limit unintended outflows of workers under temporary contracts into unemployment, and improve labor market matching by facilitating labor mobility and reallocation in the face of major sectoral shifts. While reluctance to proceed further in this area at the time of the reform may have been understandable given fears of deeper job losses at the trough of the recession, the current environment of steady recovery should be much more favorable for further progress. Loosening of restrictions on individual dismissals has brought Slovenia’s regulatory regime to the OECD average, but more seems needed to bring it closer to that of some of its eastern and southern European peers; and progress needs to be made to loosen the regime governing collective dismissals which was not touched by the reform and remains relatively restrictive.

47. On the other hand, the role of fiscal instruments providing incentives for open-ended contracts and penalizing all forms of temporary contracts may have to be rethought. In the absence of further significant progress on employment protection, such instruments are not a good substitute: they cannot prevent suboptimal aggregate employment (via suboptimal flows to unemployment and inactivity); they cannot be relied upon to achieve sustainable progress in addressing duality; and they may distort flows of other groups into employment.

48. Improving labor market matching may require additional policy interventions. While further liberalization would be important to strengthen labor mobility and reallocation, further action may be needed to address structural skill mismatches. An intensified retraining program could be a suitable first step. More generally, an enterprise-based, apprenticeship-type system, coordinated with the school system, would be the best option in the medium term, as firms are better placed to assess needed skills in a fast changing environment and they would also optimally share in training costs – given substantial industry- or firm-specific component of acquired human capital that would lead to open-ended contracts. The authorities can support these efforts via derogations from the minimum wage and targeted fiscal incentives while the system is being set up.

49. Strengthening transitions from inactivity to employment from current very low levels is an important priority. Policies would need to ensure that the overall level of the marginal tax wedge at the entry level is appropriately low to limit disincentives – in this regard both taxes and social benefits may have to be reassessed. If action on the tax side is taken, it should preferably take the form of general reductions in labor taxes – but concentrated at the low end of the wage distribution. More broadly, a comprehensive reform strategy is essential, taking care that labor market reforms affecting other segments are mutually reinforcing.

50. Addressing rising long-term unemployment is an important challenge. Further market liberalization would help, but would need to be complemented by initiatives on the marginal tax wedge, derogations from the minimum wage for certain groups, and training initiatives where skill gaps exist (or human capital has depreciated) designed to ensure adequate mobility across sectors.

References

  • Blanchard, O. and A. Landier (2002), “The Perverse Effects of Partial Labour Market Reform: Fixed-Term Contracts in France”, Economic Journal, Issue. 112, June.

    • Search Google Scholar
    • Export Citation
  • Cahuc, P., O. Charlot, F. Malherbet, H. Benghalem, and E. Limon (2016), “Taxation of Temporary Jobs: Good Intentions with Bad Outcomes?”, CEPR Discussion Paper, No. 11628, November.

    • Search Google Scholar
    • Export Citation
  • Cournede, B., O. Denk, and P. Garda (2016), “Effects of Flexibility-Enhancing Reforms on Employment Transitions”, OECD Economics Department Working Paper, No. 1348, November.

    • Search Google Scholar
    • Export Citation
  • Diamond, P. A. and A. Sahin (2016), “Disaggregating the Matching Function”, M.I.T Department of Economics Working Paper, No. 16–12, December.

    • Search Google Scholar
    • Export Citation
  • Fujita, S. and M. Nakajima (2016), “Worker Flows and Job Flows: A Quantitative Investigation”, Review of Economic Dynamics, Vol. 22.

    • Search Google Scholar
    • Export Citation
  • Garda, P. (2016), “The Ins and Outs of Employment in 25 OECD Countries”, OECD Economics Department Working Paper, No. 1350, November.

    • Search Google Scholar
    • Export Citation
  • Hall, R. E., (2005), “Employment Efficiency and Sticky Wages: Evidence from Flows in the Labor Market”, NBER Working Paper, No. 11183, March.

    • Search Google Scholar
    • Export Citation
  • IMAD (2014), “Labour Market Performance and Challenges during the Crisis”, Economic Issues, Part II.

  • Kugler, A. D. and G. Saint-Paul (2000), “Hiring and Firing Costs, Adverse Selection, and Long-Term Unemployment”, IZA Discussion Paper, No. 134, March.

    • Search Google Scholar
    • Export Citation

Annex

This Annex presents the details and derivations of the extended model of Section F. It discusses, in turn, the firms’ optimization problem, the workers’ optimization problem, and the model’s equilibrium and comparative statics.

Firms

Firms are risk neutral and discount the future at discount rate ρ and can offer either regular or temporary contracts to fill vacancies. Denote the cost of carrying an unfilled vacancy by CV.

Temporary jobs are filled from the pool of unemployed and regular jobs from the pool of temporary workers (i.e. workers are first hired into temporary jobs). New (temporary) matches start with productivity z0 ≥ 0. Match-specific productivity of permanent jobs is a random variable z, drawn from a distribution with cumulative density function F:

z~Fon[0,zmax ],zmax >z0

Following initial match, jobs (both regular and temporary jobs) are subject to productivity shock, with instantaneous probability π. Denote new match-specific productivity by z', also drawn from F. When hit by a productivity shock, temporary jobs are either terminated or converted to regular, while regular jobs are either terminated or continued.

Firing costs and payroll taxes differ between regular and permanent jobs. Denote firing cost for regular jobs by CR, and, without loss of generality, set firing cost for temporary jobs to 0. Flat payroll taxes for regular and permanent jobs are, respectively, tRwR and tTwT, where wi, i=R,T is the wage rate.

The present discounted value to the firm of temporary and regular jobs can be written, respectively, as the Bellman equations (with E being the conditional expectation operator):

ρ VF,T(z0)=z0(1+tT) wT(z0)+π E(VF,R(z)VF,T(z0)| zz);
ρ VF,R(z)=z(1+tR) wR(z)+π E(VF,R(z)VF,R(z)| zẑ)+π (VF,T(z0)VF,T(z)CR) F(ẑ);

where:

z: Productivity threshold at which firm is indifferent between terminating a temporary job and converting it to a regular one.

z^: Productivity threshold at which firm is indifferent between laying off and retaining workers under regular contracts.

Workers

Without loss of generality, the labor force is normalized to 1. Temporary and regular workers receive, respectively, their wage WT and WR and a fraction of payroll taxes as benefits, btTWT and btRWR (0<b<1). Workers are infinitely lived, risk neutral, and discount the future at rate ρ.

It is assumed that the unemployed have zero utility, and must enter a temporary job before moving on to a regular job. In line with standard search theory, the arrival rate of temporary jobs (θ) can be written as:

θ=h/u,

where h are total hires and u is the number of unemployed.

The present discounted values (to the worker) of a temporary job, a regular job, and being unemployed, VW,T, VW,R, and U, respectively, take the form of the following Bellman equations:

ρ VW,T(z0)=(1+btT) wT(z0)+π E(VW,R(z)VW,T(z0)|zz)+π [UVW,T(z0)] F(z);
ρ VW,R(z)=(1+btR) wR(z)+π E(VW,R(z)VW,R(z)|zz^)+π [UVW,R(z)] F(z^);
ρ U=θ [VW,T(z0)U]
Equilibrium

The model assumes free entry of firms, implying that the number of vacancies is determined by the zero net profit condition:

VF,T(z0)=CV

Since firms’ net present value of regular jobs increases in productivity, the zero profit condition can be used to determine the two key productivity thresholds:

VF,R(z)=VF,T(z0)=CV(1)
VF,R(ẑ)=VF,T(z0)CR(2)

Wage setting is determined by symmetric Nash bargaining, under continuous renegotiations. The Nash bargaining conditions for temporary and regular jobs can be written as:

VF,T(z0)CV=VW,T(z0)U(3)
VF,R(z)VF,T(z0)+CR=VW,R(z)U(4)

Substituting the zero-profit condition in equation (3) yields:

VW,T(z0)=(=0)

Integrating equation (4) over z and zmax, one obtains:

E(VF,R(z)VW,R(z)|zz)=(CVCR) [1F(z)]

The above equation, together with the U=0 assumption and the temporary job Bellman equation, yields an expression for the temporary job wage, in the form of the Bellman equation:

wR(z)=[zρ (CVCR)]/[2+(1+b)tR]

With regard to regular jobs, integrating equation (4) over z^ and zmax, one obtains:

E(VF,R(z)VW,R(z)|zz^)=(CVCR) [1F(z^)]

The above equation, together with the U=0 assumption and the regular job Bellman equation, yields an expression for the regular job wage, in the form of the Bellman equation:

wR (z)=[zρ (CVCR)]/[2+(1+b)tR]

Substituting the wage equations and the zero profit condition into the firm’s net present value Bellman equation for permanent jobs, and evaluating at z^ and z, one can obtain expressions implicitly defining the two productivity thresholds. These take the form of the Bellman equations:

(ρ+π)CV={[(1+btR)z+ρ(CVCR)]/[2+(1+b)tR]}+π E(VF,R(z)|zz^)+π(CVCR)F(z^)(5)
(ρ+π)(CVCR)={[(1+btR)z^+ρ(CVCR)]/[2+(1+b)tR]}+π E(VF,R(z)|zz^)+π(CVCR)F(z^)(6)

Subtraction of equation (6) from (5), and substitution of the equation for wR into the firm’s net present value Bellman equation for regular jobs, integration by parts, and use of the threshold equations (1) and (2), yields the final expressions for the two productivity thresholds:

z^=(ρ+π)(CVCR)π zmax/ρ+(π/ρ)[F(zmax)F(z^)]
z=(ρ+π)CVπ zmax/ρ+(π/ρ)[F(zmax)F(z^)]+(ρ+π)(1+tR)CR/(1+btR)

The above expressions allow comparative statics on the two productivity thresholds with respect to changes in the policy variables (layoff costs and payroll taxes):

  • A reduction in CR reduces the difference between z^ and z – as z^ falls and z rises.

  • A reduction in tR has the same effect as long as the link between benefits and payroll taxes is not perfect (b<1); if b=1, only z^ falls.

Once the productivity thresholds have been determined, one can derive the steady-state values for unemployment (u), temporary employment (eT), and regular employment (eR). Since the flow out of unemployment has to equal the flow into unemployment plus the flow into temporary employment, one obtains:

θu=π [ eT F(z)+eR F(z^)]=π eT

Using the productivity threshold steady-state conditions and the identity u+eT+eR = 1 as inactivity is assumed away in the model) yields steady-state unemployment, temporary employment, and regular employment:

u=[π F(z)]/[π F(z^)+θ (F(z^)+π(1F(z))]
eT=[θ F(z^)]/[π F(z^)+θ (F(z^)+π(1F(z))]
eR=[θ(1F(z)]/[πF(z^)+θ (F(z^)+π(1F(z))]
1

Prepared by Ioannis Halikias.

2

The increase brought the minimum wage from 41 to 51½ of the average wage, the highest ratio in the EU; see IMAD (2014). The minimum wage increase was phased in gradually to limit the burden on the labor market.

3

It should be noted that student work makes up 15–20 percent of Slovenia’s employment in the 15–29 age group.

4

These covered almost 5 percent of the total active population; see IMAD (2014).

5

For a detailed description of the reforms see IMAD (2014).

6

Shares, rather than absolute levels, are used to correct for the impact of the cycle.

7

Note that in this framework there is no ambiguity relating to the high 2014 base, as these are conditional frequencies, i.e. they capture the probability of transitioning to an open-ended contract conditional on being in a temporary contract in the previous period.

8

There could be further unintended consequences of the reform which however cannot be captured by our data. For instance, Cahuc et. al. (2016) document that taxation of temporary jobs in France also induced a shortening of the average temporary contract duration.

9

The restriction of zero expected utility for the unemployed (needed to obtain closed-form solutions), together with the free entry/zero profit condition, have the implication that payroll taxes drop out from the model’s equilibrium conditions. In order to meaningfully analyze the impact of this policy variable, the utility of the unemployed had to be endogenized, and the model calibrated.

10

At the same time, when temporary contracts are widely used for work that would normally be done under open- ended contracts, unifying payroll taxes for both types of contracts helps avoid tax evasion.

Republic of Slovenia: Selected Issues
Author: International Monetary Fund. European Dept.