Assessment and recommendations

Latvia’s labour market has recovered from the large shock of the financial crisis

In the aftermath of the financial crisis, Latvia had experienced one of the largest increases in unemployment – 15 percentage points – among all OECD countries and falling real as well as nominal wages. During a strong recovery, Latvia’s unemployment rate halved between 2010 and 2015. At 8% in 2018, it was close to pre-crisis levels but remains above the OECD average. Latvia’s employment and participation rates have risen beyond pre-crisis levels and OECD averages, respectively reaching 63% and 78% in 2018. However, most sectors have not fully regained the absolute levels of employment they exhibited before the crisis, substantial numbers of unemployed persons have left the labour force or emigrated from Latvia over the past decade, and high unemployment rates persist in some regions and demographic groups, such as youth and older men.

In recent years, instances of unmet labour demand have become more frequent, and workers with certain skills as well as in some services are in short supply. Strong wage growth has resumed, partly reflecting the minimum wage increases from 2007 onwards, which resulted in the highest relative increase in minimum wages observed in OECD countries in the period 2007-2016. However, the wage premium for workers with a tertiary education, compared to workers with upper secondary education, remains lower in Latvia than in many other OECD countries. Comparatively few new hires (35% in 2016) have a fixed-term contract, and over-qualification is rare. Estimated employment in the shadow economy reached the lowest level in 2017 since 2009.

Long-term unemployment is one of the main challenges for Latvian labour market policy

The rate of long-term unemployment was 3.3% in 2017 (as a share of the labour force), almost twice the OECD average (1.8%) but one of the lowest among the countries that were heavily affected by the financial crisis. In 2017, about two-fifths (38%) of all unemployed persons in Latvia were long-term unemployed (i.e. for at least 12 months) and 15% had been unemployed for four years or more. The share of long-term unemployed is higher among men than among women (45% compared to 37%).

Discouraged workers – often regarded as hidden unemployment – do not engage in job search because they believe that no work is available for them. Despite recent decreases, the number of discouraged workers remains comparatively high in Latvia. They represented 4% of Latvia’s non-employed working-age population in 2017, the third-highest share in European OECD countries. Six out of ten discouraged workers have been without employment for at least 12 months.

Latvia’s menu of Active Labour Market Policies (ALMPs) has expanded in recent years, but comparatively few unemployed persons register and participate

A number of Active Labour Market Policies (ALMPs) have been introduced since 2012, typically targeting disadvantaged groups. This includes: a programme promoting job seekers’ mobility across Latvian regions; a programme for the motivation and rehabilitation of (long-term) unemployed persons; a programme with aiming to prevent unemployment of older workers by raising their skills; and, in the context of the EU Youth Guarantee, several programmes targeting unemployed youth. Overall, the focus has shifted away from public works since 2012 towards employment incentives and rehabilitation. Further ALMPs are being developed through Latvia’s Inclusive Employment Strategy 2015-2020. This strategy has three overarching policy goals: reduced barriers to employment for disadvantaged groups; better of balancing labour supply and labour demand; and creating an institutional and tax environment that facilitates employment. Specific goals notably include improving ALMPs, promoting regional labour mobility, and targeting groups with a high risk of unemployment. The results of this OECD Review seek to inform the evaluation of the progress made on implementing the Inclusive Employment Strategy and form the basis for the next Employment Strategy.

Only 11% of registered unemployed persons participated in an ALMP measure in 2016, one of the lowest shares among European OECD countries. Latvia’s public spending on ALMPs (0.19% of GDP) is substantially below the OECD average (0.53% of GDP). The role that ALMPs can play is also limited by low registration of job seekers with Latvia’s public employment service, the State Employment Agency (SEA). Roughly half of all unemployed persons – a low share in comparison with other European OECD countries – were registered in the past few years, and just three-quarters of long-term unemployed persons were registered. Many unemployed persons therefore cannot be reached for support with job search or participation in ALMPs. Job seekers who are not currently unemployed account for a very small share (0.26%) of persons registered with the Latvian public employment service compared with other European OECD countries. Estimates further suggest that the number of registered vacancies in Latvia may have been below 20% of all hires for most of the past decade. The public employment service could use web-scraping methods to include unregistered vacancies that are advertised elsewhere.

Profiling of unemployed persons and sanctions could be used more effectively

In order to select and prioritise participation in ALMPs among the registered unemployed persons, Latvia’s public employment service uses a profiling tool (introduced in 2013) in combination with an individual action plan. However, most of the information used for profiling relies on self-declared answers from the unemployed person, who may refuse to answer at all. In this context, a greater role could be given to statistical information from the data collected by the SEA on job prospects and the caseworker’s assessment. While profiling currently places persons in one of 39 groups, a much smaller number of groups coupled with a greater focus on those most at risk of long-term unemployment may be more effective in practice. Most importantly, the groups should be linked to much more differentiated service streams in terms of employment and training support. This way, the available resources can be better targeted at high-risk groups, while very limited support for low-risk groups could largely or entirely be provided through online services. The example of the Netherlands could be useful when deciding who to support exclusively via online services and how these services should be provided. The individual data used for profiling could also be used to generate automated referrals to specific vacancies given the characteristics of the unemployed person.

Sanctions for unemployed persons are strict in Latvia in comparison with other OECD countries, except for one aspect: the first job offer may be rejected without consequences. Sanctions lead to the loss of the entire benefit and of access to ALMPs, which likely reduces the role they can play in practice. Less severe sanctions might be more readily applied and could effectively change behaviour while maintaining access to benefits and ALMPs in principle. In 2016, almost all sanctions were applied due to missing appointments or deadlines, while sanctions played almost no role in the context of job search and ALMP participation. Sanctions were especially likely to apply to unemployed persons not receiving unemployment benefits and often occurred around the time when unemployment benefits expire.

Unemployment benefits decline over time and the level of social assistance is low

Unemployment benefits depend on years of contribution and decline over time. Those who have adhered to the system for at least 30 years receive up to 65% of prior average wages. Stepwise lower replacement ratios apply with fewer years of contribution, down to 50% for 9 years or less. The full benefit is paid for the first three months, three-quarters are paid for the next three months, and half is paid for the last three months. Unemployed persons may otherwise be eligible for means-tested benefits that raise their income to a Guaranteed Minimum Income (GMI) level of EUR 53 per month per person in 2018 plus housing benefits. Despite this assistance, the income of targeted households relative to median incomes in Latvia remains low compared with targeted households in most other OECD countries. The income of a single person receiving these benefits amounted to only 10% of the median income in 2016, and the income of a married couple receiving benefits amounted to 14% of the median income for those without children and 22% for those with children.

Dependency on unemployment and GMI benefits is limited

Unemployment benefits are provided for up to nine months and gradually decline in value after the first three months. During the period from January 2012 to October 2017, one third of unemployment benefit recipients exhausted the nine-month duration of their benefits, while about half of beneficiaries left unemployment benefits within six months. This reflects both the higher chances of finding employment during this period and the reduction of the benefit to 50% of the granted amounted after the sixth-month mark.

The GMI is a mean-tested benefit offered to persons in need and can be combined with unemployment benefits, if the latter are sufficiently low. In 2017, only 0.5% of the population aged 15 and above received GMI benefits, down from 4% during the economic crisis. More than half of all GMI beneficiaries received the benefit for 6 months or less. The probability of leaving the GMI benefit in a given month declines with time spent on the benefit, falling below 8% after 19 months. Although the number of GMI recipients has decreased since the crisis, the share of those who rely on the benefit for 19 months or more has increased and reached 17.5% in 2016-17. Reliance on GMI benefits is higher for persons aged 55 and older (they represent half of all GMI beneficiaries) and for persons with disabilities (17% of all GMI beneficiaries). Moreover, old age and disability status are correlated with reliance on GMI benefits for longer periods and this correlation has become stronger over time.

Co-operation of municipalities and the public employment service has strong potential for the activation of social assistance recipients

Recipients of GMI benefits are required to register with the SEA as well as the municipalities that are responsible for the delivery and management of social assistance. GMI beneficiaries made up one-fifth of all registered long-term unemployed persons over the period 2012-2017. The often complex situations of long-term unemployed persons require a holistic approach and call for effective co-operation between the SEA and municipal social services. A pilot programme introduced in 2013 intensified efforts to place persons who had been registered unemployed for 20 years or more, while encouraging the SEA and municipal social services to explore various forms of practical co-operation. The outcomes of this pilot indicated large potential benefits from closer co-operation: 40% of the pilot programme’s participants – who comprised (very) long-term unemployed persons – took up employment, compared to just 16% of a group of comparable non-participants. While the intensity of co-operation thus far depends on the initiative and efforts of individual staff members in both institutions, a regular format for these exchanges has not been set up across Latvia. This should be established to embed this co-operation more firmly in the institutional architecture.

Growing numbers of persons with disabilities may need additional support to return to sustainable employment

The number of recipients of disability benefits has grown more strongly in Latvia than in almost all other European OECD countries since 2007. The stock of disability benefit recipients (aged 18 or more) increased by 17% between 2012 and 2017, reaching 106 200 persons in October 2017. This may reflect high exposure to risk factors during the crisis and poor attitude to health checks and medical visits. A comparatively high tax wedge limits the labour supply incentives for benefit recipients. The increasing numbers of disability benefit recipients may also be driven by the higher generosity of disability benefits and the limited duration of unemployment benefits. Although the vast majority of disability benefit recipients (85%) are persons of working age, their labour market outcomes are relatively poor. Only one quarter of them were employed in 2017 and 9% were registered as unemployed. When only disability benefit recipients of working age are considered, their employment rate is higher, at 36% in 2017. Additionally, only 16% of the disability benefit spells recorded during the observation period from January 2012 to October 2017 were short spells of less than 12 months. Moreover, this is an underestimation of the actual share given that many observed spells are censored: some spells that appear to last less than 12 months because they are cut short by the lack of data after October 2017 may actually last longer than 12 months.

The probability of leaving disability benefits is low, typically less than 1.5% at any time. A clear spike is observed at 12 months of benefit receipt (2.6% probability of leaving the benefit), when eligibility is reassessed. The majority of those who left the benefit took up the old-age state pension, which offers a more generous benefit than the disability scheme, at least for persons with moderate disabilities (group III). About one third of all disability benefit recipients aged 15-64 went back to employment one month after the end of their disability benefit spell. These transitions concern mainly persons, possibly with less severe health conditions, who remained active in the labour market while receiving the disability benefit. Inactivity during one’s disability benefit spell is indeed a strong predictor of subsequent labour market outcomes. Less than 2% of all disability benefit recipients who were not employed during the last 6 months of their disability spell were employed immediately after the end of their spell.

Training has remained an important component of Latvia’s menu of ALMPs

Despite other changes in the landscape of ALMPs in Latvia in the past five years, training has continued to be a vital strategy for connecting unemployed people with good jobs. Between 2012 and 2017, the number of training participations per year – a figure which potentially counts the same individual more than once if they participated in multiple trainings – averaged 6 600 for formal vocational training (henceforth “formal training”) and 15 900 for non-formal training. During this period, 74 700 individuals were registered with the SEA in any given month, on average. Formal trainings typically aim to build specific new skills such as social care, project management or welding, with participants working towards a professional qualification. Formal training takes between 22 and 202 days to complete, lasting 91 days on average. By contrast, non-formal trainings build more general skills, including languages and Information and Communications Technology (ICT) skills, but do not necessarily lead to a formal qualification. Non-formal training lasts 42 days on average.

Formal and non-formal training are distinguished from so-called “Measures to Improve Competitiveness” (MICs), which have far wider coverage but are considerably shorter in terms of duration and contact time. Between 2012 and 2017, there were around 80 000 participations per year in MICs. These MICs typically comprise short courses and workshops, concentrating on how to write CVs, how to succeed during interviews, and how to network effectively. MICs only last 1 day and require just 7 hours of contact time on average. Given their size and their wide coverage, the main analysis of training (in Chapter 3) does not focus on evaluating the effects of MICs, except in conjunction with formal and non-formal training measures.

Specialised econometric techniques are needed to evaluate training

The design and improvement of successful training programmes, and indeed all ALMPs, relies on solid empirical evidence, which typically comes from monitoring and evaluation of existing programmes. This allows policymakers to build both the efficiency and effectiveness of ALMPs. In Latvia, there are sizeable opportunities for this type of evidence-based policymaking, given the extent of detailed and linkable administrative data on individuals’ participation in ALMPs, their labour market outcomes, their background characteristics, and their receipt of social assistance.

The central problem when evaluating ALMP measures is to compare what actually happened to participants – in terms of their subsequent labour market outcomes – with what would have happened had they not participated in the ALMP measure. This counterfactual cannot be observed in practice, so evaluators typically try to approximate it by comparing participants with a comparison group of individuals that did not participate. However, simply comparing those individuals who participated in training with those who never participated in training does not produce reliable estimates of the effects of training in the context of ALMPs in Latvia, because individuals begin their training at very different times throughout their unemployment spells. It is only those individuals that spend a sustained spell in unemployment who will be assigned to training. Individuals that spend only a short spell in unemployment and who quickly find work themselves are unlikely to be assigned to training. This latter group is likely to have better labour market outcomes in the future, regardless of their non-participation in training, so they do not serve as a suitable counterfactual for training participants.

This review uses specialised econometric techniques to estimate the effects of training accurately. It looks at individuals who have spent a certain number of months in unemployment, and compares those who begin training in a given month with those who are still waiting for training, another ALMP measure, or some other way out of unemployment. The quality of these comparisons is improved by controlling for key observed characteristics – such as age, education, and some aspects of individuals’ employment histories – between those that begin training and those who are still waiting.

The main outcomes on which the review focusses are individuals’ chances of employment and individuals’ earnings upon (re)entering employment, after a given number of months after training starts. The earnings estimates represent the effects on the flow of earnings measured at a given time (a certain number of months) after the start of training. They do not capture the stock of earnings that may be foregone while training is being completed and returning to work is not possible (or is much more difficult).

Both formal and non-formal training have positive effects on individuals’ labour market outcomes

Formal and non-formal training increased individuals’ likelihood of finding a job and increased earnings among those who found a job. The point estimates of the employment effects were overall larger for formal training than for non-formal training, and these differences between formal and non-formal training were statistically significant over all time horizons. Looking at labour market outcomes 18 months after the start of training (by which time training will have finished) formal training participants experienced a 7.7 percentage point increase in their employment chances while non-formal training participants experienced a 4.9 percentage point increase. By contrast, the point estimates for the earnings effects were somewhat larger for non-formal training, with these differences between formal and non-formal training being statistically significant over most time horizons. After 18 months, non-formal training participants experienced a 5.8% increase in their monthly earnings compared with a 2.2% increase for formal training participants (although this does not take into account the stock of earnings foregone whilst participating in training). Nevertheless, it is important to note that the effect on earnings can only be estimated for those individuals that successfully gained a job. This may dampen the estimates of the effects of formal training on earnings, relative to non-formal training, because formal training increased the likelihood that individuals had a job by more. As such, more formal training participants – with lower earnings potential – would have been included in the estimation of the effects of formal training on earnings, making this estimated effect appear lower.

The positive effects on earnings and employment emerge relatively quickly, and persist over a long time horizon. For employment, the effects are positive and significant after 12 months for formal training and after just 6 months for non-formal training. The slower onset of any effects for formal training is consistent with their taking longer to complete. Similarly, for earnings, it takes 18 and 12 months for positive and significant effects to appear for formal and non-formal trainings respectively (notwithstanding any earnings foregone during the training itself). All of these effects remain positive with very little decay even after 36 months, demonstrating the persistence of training’s impact.

These results warrant a certain level of optimism about the effectiveness of Latvia’s training programmes for the unemployed. Indeed, comparing the results of this review to recent meta-analyses, Latvia’s training programmes for the unemployed appear successful relative to similar programmes in other countries. This relative success may partly be down to contextual factors, especially given the large pool of individuals that became detached from the labour market during the financial crisis and the skills shortages that persist in some sectors. However, certain aspects of training for the unemployed in Latvia may have helped to foster good results: for one, the breadth of formal and non-formal training courses offered chimes with cross-country evidence that ALMPs that are in some sense adaptable are more likely to succeed.

While trainings’ effects are heterogeneous, there are positive effects for virtually all sub-groups

Certain sub-groups appear to benefit more than others from Latvia’s training programmes – at least in terms of the point estimates – with three results meriting particular attention. Firstly, while women benefit slightly more than men from formal training in terms of the chances of finding a job, the inverse is true for non-formal training, with these differences widening over time. Secondly, younger workers – those aged less than 30 years old – experienced weaker employment effects, at least from formal training. Thirdly, the long-term effects on the employment chances of social assistance recipients are substantially larger than those who did not receive social assistance.

Nevertheless, one of the most striking features of the results is that training’s effects remain positive even when the sample is divided up according to individual characteristics. Thus, rather than justifying increased specialisation in the way that ALMP measures are assigned per se, the results indicate that training has the potential to work for many different types of unemployed individuals.

Combining training with other ALMP measures increases its impact on labour market outcomes

When training is combined with other ALMP measures, there is a larger impact on participants’ labour market outcomes. Firstly, providing Measures to Improve Competitiveness (MICs) before training slightly improves the effects on employment and earnings for both formal and non-formal training. The SEA is currently seeking to reduce the number of MICs by bundling some of them in with non-formal training courses, so as to improve the quality of training programmes and MICs overall. Yet it will be important to ensure this reform does not limit access to MICs for those participating in formal training, either by maintaining some independent MICs to which all registered unemployed individuals have access or by combing MICs with formal training as well as non-formal training. Secondly, providing mobility support – a programme in which individuals may receive EUR 100 per month to cover the costs of travel to or accommodation at training sites more than 15 kilometres away – increases, sometimes sizeably, the effects of both formal and non-formal trainings. Indeed, in one particularly striking result, receiving non-formal training with mobility support yields a 23.1% increase in earnings after 18 months, compared with an increase of 5.5% for those receiving non-formal training without mobility support. In part, this may be because those training participants taking up mobility support are more motivated than those who do not (and such differences in motivation were not captured by the control variables included in the analysis). More fundamentally, however, mobility support is likely to improve the match between specific trainings and participants, offering a larger boost to participants’ labour market outcomes. The individual action plans should develop combinations of ALMPs that harness these complementarities and respond better to the widely differing needs and constraints faced by beneficiaries.

Training is provided through vouchers to promote choice and competition

Since 2011, training for the unemployed has been provided through a voucher system in Latvia. The vouchers consist of a physical document, issued at the local SEA branch office, which carries a cash-equivalent value and can be redeemed at pre-approved training providers. Vouchers afford participants more choice over the type of training that they do and over their training provider, potentially improving the match between voucher recipients’ needs and the training that is actually provided. In addition, voucher systems may improve the quality and performance of training providers by promoting competition. This last point is of particular importance for Latvia. Prior to 2011, there were notable examples of training providers having extremely long procurement contracts, such that performance deteriorated over time. The introduction of the voucher system therefore sought to take this procurement step out of the equation, instead using vouchers to allocate training transparently.

Some voucher recipients may need more support to ensure they can effectively exercise choice

Caseworkers in voucher systems face a difficult balance between supporting clients in using their vouchers and not overly interfering such that they effectively make choices on their clients’ behalf. On the one hand, caseworkers are likely to have a good understanding of clients’ needs given their regular interactions with them and their experience of assigning clients to different ALMP measures. On the other hand, assuming that voucher recipients know their needs best, it may be difficult to reap the full benefits of having a voucher system – in terms of choice and competition – if caseworkers are too heavy-handed in their support. In Latvia, the SEA already gives caseworkers many opportunities to support voucher recipients to make choices effectively during their regular meetings. Caseworkers are able to recommend particular occupations or types of training programmes from the full list of programmes for which vouchers are eligible.

Certain disadvantaged groups may need more support from caseworkers in order to exercise effective choice. In Latvia, age and language abilities appear to influence rates of voucher redemption, which may serve as a proxy for individuals’ ability to exercise effective choice. Individuals lacking at least basic Latvian language skills are 7 percentage points less likely to redeem their vouchers than those with basic (or higher) Latvian language skills. Similarly, individuals aged 15-24 years are 7 percentage points less likely than those aged 55 years or more to redeem their vouchers. However, young people may be directed to long courses organised by the Ministry of Education and Science outside the voucher system, which increases non-redemption rates. In any case, restricting caseworkers’ opportunities to support clients’ choices over how they use their vouchers may be counterproductive among those sub-groups who are already struggling to redeem their vouchers.

Special effort is needed to promote competition between training providers in certain remote and rural areas

Training providers are not spread evenly across Latvia, meaning that some voucher recipients may struggle to find suitable training options locally. There are large clusters of rural municipalities, especially in the Kurzeme and Zemgale regions, which contain no officially accredited training providers at all. All other things equal, this may hamper choice for voucher recipients and limit competition between training providers in such areas.

One way to address this issue would be to boost the number of training providers in areas that currently have relatively few training providers, but such an approach does not currently seem to be tenable in Latvia. Indeed, the SEA is in the process of reducing the number of accredited training providers, a reform which is motivated in part by the fact that it is difficult to fill certain training classes (especially in remote and rural areas), which lengthens the time voucher recipients must wait to begin training and increases possible lock-in effects. To do this, the SEA is making the selection criteria for becoming an SEA-accredited training provider more stringent, which also serves to boost quality amongst those providers that survive.

With the number of training providers declining, support for regional mobility for training participants is even more important and may need to be enhanced, to make sure all voucher recipients can access a wide range of training providers. Although overall receipts of mobility support grew in 2016 and 2017, just 9% of formal training participants and 7% of non-formal training participants received mobility support. The expansion of mobility support may need to go further to promote effective choice, especially in areas with relatively few training providers.

Latvia’s voucher system may leave individuals unsure about their status, potentially compounding lock-in effects

Vouchers are technically valid for a very short period in Latvia, at just 14 days. However, individuals are assigned to the training voucher programme in advance of actually receiving the voucher. During this time, they are expected to search for suitable training providers and training programmes, potentially at the expense of searching for a job. The period between assignment to the training voucher programme and actual receipt of the voucher can be very long: 46 days on average for formal trainings and around 96 days on average for those non-formal trainings focused on foreign languages and ICTs.

While it is unlikely that waiting times can be fully eliminated, the specific procedure for assigning vouchers in Latvia may compound lock-in effects. Although voucher recipients-to-be are told of their assignment to the training voucher programme in advance, it is not clear (1) whether they fully believe the voucher will arrive after this waiting period and (2) whether they know how long the waiting period will be (especially if it depends on classes filling up). As such, the intensity with which they should search for training providers during this period may be unclear, potentially making this search longer and less effective. An alternative would be to disburse the voucher earlier but then make vouchers valid for longer, as in other countries. In Germany, for example, training vouchers for the unemployed are typically valid for three months. This may help to clarify the status of voucher recipients, ensuring that they devote sufficient effort to finding a suitable training provider, even if there is a subsequent waiting period before training starts.

Large differences between regions imply gains to inter-regional mobility

According to several indicators, the divide between regions in Latvia is one of the strongest among OECD countries. The unemployment rate, the youth unemployment rate and the share of long-term unemployed among all unemployed all vary considerably across Latvia’s regions, with densely populated Riga and the surrounding Pieriga region exhibiting the lowest levels. In addition, the same two regions exhibited the highest supply of job vacancies in 2016, relative to total employment in the region. Within regions, rural areas appear to offer less favourable labour market conditions: while rural areas accounted for 39% of all employed persons in 2016, they also accounted for 45% of the registered unemployed, 51% of recipients of guaranteed minimum income benefits and 56% of discouraged workers.

The observed long-run trends in migration flows between Latvian regions highlight the attractiveness of urban and suburban areas, especially the surroundings of Riga but also regional centres. Compared with other European OECD countries, regional mobility in Latvia reaches an intermediate level: in recent years (2013-16), an annual estimate of 1.3% of the population of working age (15-64) changed their region of residence. The mobility of unemployed persons in Latvia is somewhat higher (estimated at 1.4%). Young unemployed persons (15-34) exhibit a rather high willingness to move within Latvia: in 2016, 26% were willing to move to another region for a job, compared with an EU average value of 20%. However, their willingness to commute for a job is comparatively low (44% compared with an EU average of 64%). Rapidly increasing rent levels in Riga and the surrounding region of Pieriga likely discourage some regional mobility.

An ALMP programme contributes significantly to greater regional mobility of unemployed persons

In 2013, the SEA introduced an ALMP programme that offers support with taking up distant job offers (at least 20 km from the current residence) or with attending distant training measures, by reimbursing costs for transport or housing. The cumulated number of participants approached 9 200 by the end of 2017, and participants are typically younger than 35, unmarried and not highly educated. Using three eligibility rules for participation in the programme, the OECD has carried out an impact evaluation of this programme. In a difference-in-difference approach, the evaluation examines whether the job-related mobility of eligible groups of unemployed persons increased more strongly after the introduction (or extension) of the programme than the job-related mobility of comparable yet ineligible unemployed persons.

The results suggest an overall positive effect of the programme (and its extension) on the job-related mobility of unemployed persons. The introduction of the programme appears to have increased the probability that unemployed persons move in a particular month to take up a job by one-half. Offering the programme under the Youth Guarantee and extending eligibility to public-sector jobs had further, albeit smaller, positive effects on job-related mobility of unemployed persons. Specific analyses for certain groups of unemployed persons confirm that the programme and its extension have also raised the job-related mobility of recipients of social assistance. Job-related mobility of two other target groups, persons with disabilities and residents of the region of Latgale, appears positively affected by the introduction of the programme but unaffected by its extensions. Further results indicate that job-related mobility of unemployed persons declines with unemployment duration: the probability of moving elsewhere to take up a job is highest during the first three months of unemployment, then falls steadily as the duration increases up to a duration of 13-18 months, after which this probability stays roughly constant.

While families might need more support for mobility, young and single persons should face greater obligations to move

The limitation of financial support under this programme to EUR 400 limits expenses for the SEA but seems low for couples and especially for families who often face substantially higher up-front costs of moving. One way that the SEA could allow for higher amounts is to arrange for small loans from a third party. To increase take-up of the programme, stricter mobility requirements could apply to unemployed persons who are single or whose partner does not hold a local job. In particular, young unemployed persons without family commitments at the place of residence should be expected to move anywhere in Latvia in order to take up employment, given that support for mobility is available. Under current rules, offers of jobs that cannot be reached within one hour on public transport can be declined without consequences. In addition, whether or not an unemployed person has access to a car could be included as a criterion for a distant job offer being suitable or not.

Currently, unemployed persons only become eligible for support with regional mobility after an unemployment duration of two months. As unemployed persons might often be able to shift out the beginning of the job beyond the two-month mark, this condition might have little relevance for eligibility but tend to delay employment. Instead of the two-month waiting period, eligibility for the programme could be tied to poor job prospects as established by profiling. Some participants in the programme move to take up high-skilled jobs, and public support for their mobility is probably not necessary in these cases and such job offers could be non-eligible for mobility support. Linking eligibility for the programme to a certain profiling outcome, for example, would exclude some who do not need support for mobility, freeing up resources for those who do.

The programme for entrepreneurship can access an untapped potential of entrepreneurs among unemployed persons

To some extent, policymakers can also address unemployment across Latvia’s regions by fostering entrepreneurship and start-ups in order to generate sustained employment growth within the regions themselves. The number of self-employed persons has been on the rise in most of Latvia’s regions. In Pieriga, it rose by 40% between 2012 and 2016, although there are concerns that this may reflect the introduction of the microenterprise tax law that is thought to be associated with an increase in bogus self-employment. Growth in the number of firms is around 5% per year both in urban and in rural areas in Latvia, which is one of the highest values among OECD countries covered by these data. With respect to entrepreneurship in the digital economy, Latvia’s well-developed broadband infrastructure can help rural regions catch up with cities.

Survey data from 2015 suggest that 3% of all unemployed persons in Latvia would like to become self-employed, more than the average for EU countries. However, only 2% of unemployed persons in Latvia moved into self-employment in 2016. The discrepancy highlights an untapped potential of entrepreneurs, which appears larger than in any other EU country. Latvia’s ALMP programme for entrepreneurship and self-employment can therefore play an important role for reducing unemployment across regions. The programme assists participants with the formulation of business plans, provides feedback, and supports the implementation of approved business plans with grants of up to EUR 3 000 as well as monthly stipends at the level of the minimum wage. Between 2012 and 2017, participants whose business plan was approved exhibited higher employment rates four or more months after the end of the programme than participants whose business plan was not approved (65%-70% compared with around 60%). Out of 377 persons who received a grant for their start-up or self-employment in the period 2008-2014, 71% were still in this business two years later, a survival rate comparable to those observed in similar programmes in other European countries. The programme also offers good chances of obtaining grants to participants with disabilities and long-term unemployed persons. Participants who are residents of regions outside Riga are at least as likely as other participants to receive a grant.

A well-targeted programme of subsidised employment is offered to vulnerable groups

Latvia devotes one fifth of its ALMP expenditures to a programme offering subsidised employment in the private sector for the most vulnerable groups of unemployed persons. The subsidy covers up to 50% of the total wage cost and the subsidy should not exceed the minimum wage (or 1.5 times the minimum wage for some persons with disabilities). The maximum duration of the subsidy is six months for youth and 12 months for most other unemployed persons, although the duration can go up to 24 months for the long-term unemployed (LTU), persons with disabilities, and some other vulnerable groups. The subsidy is paid to a broad range of employers under the condition that they hire a candidate from the pool of eligible unemployed persons registered with Latvia’s public employment service. In order to minimise substitution effects, the vacancy should be first advertised for a minimum of four months before a subsidised employee can be hired and the selected candidate should not have been an employee of the same firm in the past year. The same person can participate in the programme more than once, but a minimum of one year should occur between two spells of subsidised employment. During the period from January 2012 to October 2017, about 10% of employment subsidy participants had participated in the programme more than once. Nevertheless, this estimate represents the lower bound of the extent of repeated participation, as participation before January 2012 and after October 2017 cannot be observed in the available data.

A total of 9 000 persons participated in the programme between 2012 and 2017. Persons with disabilities represented close to one third of all the participants in employment subsidies, while young persons (aged 20-29) and persons aged 55 or more represented 28% and 11% respectively. Tight targeting of subsidies is a key determinant of their success, along with the size of their indirect effects such as deadweight losses (persons who benefit from the programme but who would have found a job even without participation), lock-in effects, and displacement effects. The existing evidence from such programmes in other countries suggests that subsidies should not be provided at the very start of the unemployment spell in order to minimise deadweight losses. In that sense, Latvia’s targeting of the LTU is in line with well-established facts in the empirical literature. Moreover, by targeting vulnerable groups, who would otherwise have limited chances of finding employment in the absence of the subsidy, this programme minimises deadweight losses.

Participation in the subsidised employment programme implies some administrative burden for employers

Heavy bureaucratic procedures and stringent conditions imposed on employers, may jeopardise their willingness to participate in the programme, especially if the extent to which they benefit is considered to be low. In Latvia, participation in subsidised employment entails some administrative burden for employers that could potentially be reduced. Firstly, to be selected for participation in the programme, employers have to prepare and submit to the SEA a list of documents, including those certifying their compliance with tax and other duties. Many of these documents can easily be requested by the SEA through the online system and should not require additional work from the hiring firm. Secondly, employers are required to submit to the SEA monthly reports on the hours worked by the subsidised employee so that the subsidy can be calculated and paid. Again, this can be a substantial administrative burden especially for small-sized companies (up to four employees) who represent close to half of the employers participating in this programme in Latvia. Finally, the requirement to assign a qualified supervisor for every unemployed person hired through the programme is welcome but can also constitute an obstacle for some employers. Although the supervisor is paid a wage supplement from the state budget, the requirement that his/her qualifications should match those needed for the subsidised position is likely to be an additional hurdle for small businesses.

The subsidised employment programme has a positive effect on employment

The probability of obtaining subsidised employment among eligible persons is very low. It is close to 3% for persons who have been unemployed for at least 12 months and even lower for youth and older unemployed persons. For persons with disabilities, the likelihood is 6.6%.

The evaluation of employment subsidies’ effectiveness conducted in this review relies on comparing the labour market outcomes of participants (the intervention group) with similar eligible unemployed persons who did not receive employment subsidies (the comparison group). Two sets of estimations are implemented because the time that must be spent in unemployment to become eligible for employment subsidies differs for certain groups: some individuals (including young people) must wait just six months to become eligible while others must wait 12 months from the start of the unemployment spell. The second group is likely to capture the LTU, including those unemployed persons who have not been treated as a priority by caseworkers and those who have spent time participating in other ALMPs. It should also be noted that some groups – including persons with disabilities – become eligible for employment subsidies immediately after registering. In any case, the intervention group comprises those persons who received the employment subsidy within 6 months of becoming eligible for participation in the programme. To increase the comparability of the intervention and comparison groups, the estimations include a rich set of controls for personal characteristics, household composition, and location.

Participation in the programme is associated with a higher probability of employment, which declines quickly over time. Two years after the point at which they become eligible (six months of unemployment), programme participants are 16 percentage points more likely than similar unemployed individuals in the control group to be employed. The estimated effects are large but are not very different from those found in similar settings in other countries and when similar econometric techniques are applied. They reflect to a great extent the fact that many of these persons still hold a subsidised job two or even three years after becoming unemployed.

When individuals who remain in (or return to) subsidised jobs are excluded from the analysis, the estimated effect of the programme is smaller (i.e. ten percentage points at 24 months after they become eligible for person who become eligible after six months of unemployment) but remains positive and statistically significant up to four years after the moment when the individual became eligible. Separate analyses conducted for the different target groups show that when only non-subsidised jobs are considered, the estimated effect of the programme is positive and significant up to four years after the clock start for youth and for older unemployed persons (to a lesser extent).

However, no effect is found for persons with disabilities. This result seems in line with anecdotal evidence that employers tend to let persons with disabilities go after the end of the subsidy. The short-lived increase in the maximum duration of the subsidy for persons with disabilities in 2014 led to some increase in the average duration of the subsidy for this group, but was not correlated with improved labour market outcomes. More in-depth analysis would be required to fully understand the absence of an effect for persons with disabilities, but this is not possible given the small number of persons affected by the change. However, the lack of any correlation between the longer subsidy duration and improved labour market outcomes suggests that such programmes may have limited effectiveness for this group as a whole. One option for consideration would be to differentiate the treatment of persons with disabilities according to their assessed degree of disability and barriers to work. For those with severe disabilities, a longer subsidy duration could be considered, while for those with milder disabilities, a shorter subsidy duration could be coupled with strong social services. Moreover, the level of the subsidy could vary according to the unemployed person’s assessed disability and could also change over time.

Latvia’s rich administrative data should be used for regular monitoring and impact assessment of ALMPs

Latvia has a remarkable administrative data system in place, which makes it possible to link individual-level data from various sources and, in turn, analyse important labour market policy questions as well as many other socio-economic research and policy questions. This review has benefited from enormous efforts from the SEA and its data operator, UNISO, the State Social Insurance Agency, the Office of Citizenship and Migration Affairs, and ZZ Dats who maintain the municipal information system data base with the support of Latvia’s 118 (out of 119) municipalities and who agreed to extract their data on social assistance. A rich set of administrative data was provided to the OECD and was linked by the OECD team, which allowed for an in-depth and rich analysis of the impact of selected ALMPs to be carried out. Crucially, the linked administrative data made it possible to track individuals over relatively long time horizons, allowing both the short-term and longer-term impacts of programmes to be identified, and thus providing a better understanding of the mechanisms through which ALMPs may operate. Moreover, the detailed information on the participation of registered unemployed persons in all types of ALMP measures and on their interactions with the SEA allowed the review to explore how the effects of different elements of labour market policies interact. At the same time, having information on each individual’s personal characteristics made it possible to control for observable differences between those participating and those not participating in a programme, reducing bias in the estimated impact of each programme that was evaluated.

Efforts to maintain detailed and linkable administrative data should continue in order to facilitate regular monitoring and evaluation of the effectiveness of activation measures. This data collection can also serve to answer other policy-relevant questions going beyond the field of activation policies. Such efforts require investment in human resources to build the necessary technical skills. This could be easily achieved in Latvia where investment in ICT skills has been high. Lessons from other OECD countries (e.g. Estonia, the Netherlands, Norway, Germany, Flanders in Belgium) could be used to further boost Latvia’s capacity in this field.

Key policy recommendations

General operation of the State Employment Agency

  • Establish a regular and structured format for co-operation between the SEA and the municipalities, following the positive results of the 2013 pilot targeting the very long-term unemployed.

  • Reconsider the requirements for the registration of vacancies so as not to unduly discourage employers for doing so i.e. by requiring detailed wage information. Wage information in ranges could be requested instead.

  • Introduce possibilities for less severe sanctions in case of refusal of job offer, e.g. temporary benefit reductions.

  • Require young unemployed persons without family commitments to accept job offers from anywhere in Latvia. For all unemployed persons, take access to a car into account in case-by-case decisions of whether a job offer is suitable.

  • Revise the profiling tool of unemployed persons and its use along the following lines:

    • Reduce the number of groups and link them to differentiated service streams.

    • Ensure the profiling tool is available and used at the moment of registration with the SEA.

    • Rely more on easily available and reliable statistical information than self-declared information from the unemployed person.

  • Consider providing online services to unemployed persons who are more likely to resume work quickly according to the outcomes of the profiling tool.

Training programmes

  • Enhance the ongoing programme to support regional mobility for those receiving training, especially in areas such as Kurzeme and Zemgale where there are relatively few training providers, to allow voucher recipients to access a larger pool of training programmes.

  • Consider reforming the specific procedure for assigning vouchers to limit lock in effects by:

    • Shortening or eliminating the period when registered unemployed individuals are aware that they have been assigned to the training voucher programme but have not yet received their voucher.

    • Lengthening the time for which vouchers are actually valid past the current two-week period, to support effective choice.

  • Consider additional ways to reduce the lock-in effects by allowing training to happen alongside job search.

  • Continue to consolidate the provision of training for the unemployed to reduce waiting times, but carefully monitor the effects this has on choice for voucher recipients and on competition between training providers.

  • Ensure caseworkers have the capacity to provide special support to those voucher recipients most in need of help in exercising effective choice, including individuals without a basic command of the Latvian language and young people.

  • Ensure that ongoing reforms that bundle together Measures to Improve Competitiveness with non-formal training do not reduce access to such measures for formal training participants.

Programme for subsidised employment

  • Reduce the burden on employers by using the IT system to transmit the monthly information required on hours worked to calculate the amount of the subsidy to be paid.

  • Consider differentiating the conditions of the employment subsidy for persons with disabilities according to the degree of assessed disability, barriers to work and/or work capacity, e.g. by:

    • Extending the duration of the subsidy and possibly reducing its amount for persons with severe disabilities for whom this is a unique way to get a job. In these cases, the subsidy should also be considered as a tool for social inclusion.

    • Reducing the duration of the subsidy for those with milder disabilities and strengthening the employment and social services provided to them during the period of subsidised employment.

  • Maintain the programme for temporary public works, which can be scaled up to serve as a safety net in difficult economic conditions.

Programme promoting regional mobility of unemployed persons

  • Explore how greater support for mobility can be offered to families, e.g. through small loans that the State Employment Agency arranges with a credit providers.

  • Abolish the limitation of the programme to persons who have been unemployed for at least two months but link eligibility to profiling outcomes that determine the need for such mobility support and exclude certain occupations and/or highly paid jobs from those eligible for support.

Data collection

  • Continue the investment in building and maintaining a well-functioning data infrastructure.

  • Develop a mechanism for automatic monitoring and evaluation of the outcomes of ALMPs with minimal human resource requirements on a regular basis.

  • Use the experience acquired and lessons learned through the OECD Review and other data analyses conducted in the past to build the capacity to perform systematic impact evaluations of ALMPs, internally or in co-operation with experts.

End of the section – Back to iLibrary publication page