Chapter 5. Activating Latvia’s most vulnerable groups
This Chapter analyses the labour market situation of Latvia’s most vulnerable groups, mainly the long-term unemployed, young and old unemployed persons and persons with disabilities. First, it presents recent trends in unemployment, disability and Guaranteed Minimum Income benefit recipiency and examines the extent to which certain population groups are dependent on these benefits. Second, it conducts an in-depth assessment of Latvia’s employment subsidy programmes that focus on the most vulnerable groups. Finally, the Chapter discusses Latvia’s Public Works Programme that was extensively used during the economic crisis as an income support measure combined with activation support.
Introduction
This chapter focuses on the policy challenges for the activation of the most vulnerable groups of job seekers in Latvia, notably the long-term unemployed, youth, old unemployed persons and persons with disabilities. First, it analyses trends in unemployment, disability and social assistance (GMI) benefit recipiency and highlights changes in recent years and since the economic crisis. Second, the Chapter analyses benefit dependency and draws the profile of persons who rely continuously or in a repetitive manner on benefits. Third, the Chapter focuses on the programme of subsidised employment offered by Latvia’s State Employment Agency (SEA, Latvia’s Public Employment Service or PES). It conducts a descriptive analysis of the programme and its participants and an econometric evaluation of the impact of the programme on the labour market outcomes of participants. Fourth, a brief description of the Public Works Programme (PWP) is presented, highlighting its role as a measure against poverty (combined with activation) used during the economic crisis.
Latvia’s most vulnerable groups and factors shaping their situation
Latvia’s PES recognises the status of two groups of persons as eligible for support: unemployed persons and persons seeking employment. Both groups are registered with the SEA and look actively for work. Meanwhile persons seeking employment have to comply with less strict requirements and are entitled only to some Active Labour Market policies (ALMPs). Unemployment benefit receipt differentiates the two groups as well. This Chapter looks at the different people receiving support from Latvia’s social protection system. A detailed description of the system and the eligibility conditions for the different benefits are provided in Chapter 2 of this Review.
Characteristics of LTU versus other unemployed persons
Table 5.1 presents the characteristics of all unemployed persons, by the length of their unemployment spell. The analysis focuses on unemployment spells that start and end within the observation period between January 2012 and October 2017. This is to resolve the problem with censored spells (those that start or end outside the observation period) which would distort the distribution of persons across spell length.
Persons who are unemployed for short periods of time differ in many ways from those who have been unemployed for longer periods (12-24 months, or more than 24 months). Women represent a higher share of persons who are unemployed for 6 months and more, and for short spells of 0-2 months men represent 47% of all the unemployed. Persons aged 45 years and above are over-represented among the long-term unemployed (46% of those unemployed for one to two years and 50% among those who have been unemployed for two years or more). In contrast, young persons are more represented in groups with short unemployment spells. Persons aged 15-24 represent one third of all the persons who have unemployed for up to three months and 28% of those with unemployment spells of 3-6 months. Likewise, unemployed persons aged 25-34 represent close to one quarter of all the unemployed with short unemployment spells (spells of 0-3 and 3-6 months). Persons of Slavic origin represent 38% of persons who have been unemployed for more than one year, whereas they represent 31-32% of those with shorter unemployment spells.
In terms of educational attainment, there are no major differences across the unemployed grouped by the duration of their unemployment spell. Persons with professional vocational education represent 31% of all unemployed persons with long unemployment spells (24 months or more), while they represent 24% of those with short spells of less than 6 months. In contrast, persons with higher education are more represented among unemployed persons with long unemployment spell durations than among those who have been unemployed for less than 6 months. Highly educated unemployed persons may be more likely to be eligible for unemployment benefits and the size of the benefit they are entitled for may be higher than for the unemployed with a lower education level. As a result, the former would afford to wait before accepting a job, whereas the latter may not be able to wait and return to employment as soon as any job opportunity arises.
The breakdown by region confirms the large regional differences that exist in Latvia as highlighted in Chapter 2 and discussed in detail in Chapter 4. Persons living in the Latgale region represent 29% of persons who have been unemployed for at least two years, whereas they represent just 15% of unemployment spells of up to one year. Persons with disabilities account for 12% of unemployed persons with unemployment spells of 1-2 years and 14% of those who have been unemployed for more than two years, while they represent only 4-5% of those with short unemployment spells of up to 6 months.
Benefit dependency is likely to be low, except for disability benefits
Latvia is characterised by a relatively low coverage of the various benefits offered to the population in need (a discussion on this can be found in OECD (2016[1])). Figure 5.1 presents the latest trends in the numbers of benefit recipients as a share of the population aged 15 and above for the following benefits: social assistance which includes GMI (Guaranteed Minimum Income) and housing benefits (these two components are also shown separately in Figure 5.1), disability benefits and unemployment benefits. Overall, the low coverage of the different benefits coupled with the short average time spent on benefits (as described in more detail later in this section) may hint to relatively low benefit dependency in Latvia.
The only exception to this general observation concerns the disability benefit. The stock of disability benefit recipients has gradually increased and the probability of leaving the disability benefit is low. This trend is related to population ageing and the adverse impact of the economic crisis on the health status of the Latvian population, especially that of older cohorts (The World Bank, 2015[2]). Moreover, public awareness of rights and benefits for occupational patients has improved and as a result, more persons with occupational diseases register officially than in the past, particularly since the beginning of the economic crisis (State Labour Inspection, 2018[3]; State Labour Inspection, 2015[4]; State Labour Inspection, 2010[5]).
In contrast, the reliance on social assistance benefits has been declining since the economic crisis when it played a key role in supporting the groups in need. In 2017, social assistance was received at any one month by no more than 2.3% of the population aged 15 and above, where GMI in cash or in kind was paid to only 0.5-0.7% of residents included in the data of the Office of Citizenship and Migration Affairs (OCMA) (Figure 5.1). Fluctuations in the coverage of housing benefits are explained by the cyclical nature of the according payments and reflect the number of persons for whom the local governments made any payments related to rent or house subsidy during the according month, not the actual number of entitled persons1.
The disability benefits considered in this section comprise various types of state disability related benefits, such as disability pension, state social security disability allowance, benefits for persons with disabilities in need of care etc. However, transport compensation for persons with disabilities with mobility limitations is not included because the available administrative data report payments made by the SSIA every six months, but do not allow indicating the number of entitled persons precisely.
Note: GMI: Guaranteed Minimum Income. The figure covers the period January 2012 to October 2017 and reports the stock of benefit recipients. The data on disability-related benefits are available for persons aged 18 and over only, and therefore, the reflected indicator is slightly underestimated as recipients aged 15-17 are not accounted for. Social assistance benefits comprise GMI paid in cash or in kind as well as rent and housing subsidies (shown together with the housing benefit and also shown separately in this figure). The disability benefits accounted for in this graph, and further in this section, comprise various types of state disability-related benefits, including disability pension, state social security disability allowance, benefit for a person with disabilities in need of care etc., but do not include transport compensation for persons with mobility limitations since the administrative data contain information on according payments made by SSIA once in six months, but do not allow us indicating precisely the number of entitled persons.
Source: Latvian Social Insurance Agency, Latvia’s Municipal Information System Database and OECD estimates.
The number of disability benefit recipients (aged 18 and over) increased by 17%, from 90 500 in January 2012 to 106 200 in October 2017 (Figure 5.2). This trend is driven by a number of factors, including relatively high exposure to various risk factors and rather poor attitude to occupational safety and health (OSH) in Latvia, that characterises both employers and workers. For example, about a half of all enterprises in Latvia admit not conducting a risk assessment of thework environment; all employers in Latvia tend to minimise their business costs by e.g. reducing their investments related to occupational safety and health (Employers’ Confederation of Latvia, 2013[6]). Overall, the Latvian society is relatively poorly informed on issues related to work environment and risk assessment: in 2013, only 36.5 % of adult respondents surveyed answered that they are well informed on these issues; others either admitted that they knew nothing or had only heard something about this (Employers’ Confederation of Latvia, 2013[6]). As a consequence, almost half (46.5%) of the working-age population (15-64) has work-related health problems resulting in severe limitations in daily activities (self-reported data). This share is twice that of the EU average (22.3%) (Eurostat, 2013[7]).
Another factor is related to relatively poor doctor visit culture (especially among males) in Latvia and heavily delayed medical examinations. According to the estimations by the Riga Stradins University, the average spell between the very first symptom of an occupational disease and the medical conclusion reaches ten years (Employers’ Confederation of Latvia, 2013[6]). This situation is compounded by relatively limited access to health care services which is related to a number of reasons, such as the relatively long waiting lists and the inability to afford health care observed among the less well-off population groups (Karanikolos et al., 2016[8]). During the crisis, groups with a prolonged disease history often opted for applying for disability benefits only after losing their jobs and exhausting their unemployment benefit entitlement (The World Bank, 2015[2]). Over the period from 2008 to 2011, the number of disability benefit recipients aged 18 and over increased on average by nearly 3 700 persons or by 6.5% on an annual basis. Over 2012-16, the average annual increase of beneficiaries was still high at about 3 000 persons or 3.2%, whereas in 2017, a slowdown was observed (Figure 5.2). Overall, since the beginning of 2012 the number of persons receiving monthly disability benefits increased by 17%.
Inflows into disability benefits have steadily increased between 2012 and 2017 (Annex Figure 5.A.2). Inflows by persons after retirement (65 and over) are driving this increase, while those by persons of working age (18-64) are either stable or somewhat declining (during 2017).
Note: The figure covers the population aged 18 and over. Disability benefits comprise various types of state disability related benefits, including disability pension, state social security disability allowance, benefits for persons with disabilities in need of care etc. It should be noted that disability benefit recipients covers only part of persons with disabilities of retirement age. Persons who had disability before retirement and who after reaching retirement became recipients of old-age pension (with continuous disability) are not included.
Source: Latvian Social Insurance Agency and OECD estimates.
In 2017, 85% of adult disability benefit recipients were persons of working age2 (Figure 5.2). Similarly to the situation in the pre-crisis and crisis period (The World Bank, 2015[2]), disability is associated with relatively low labour force participation. In 2017, only 34-36% of working age persons with disabilities were employed and another 8-9% were registered as unemployed. Nonetheless, a positive development is observed during the period covered in Figure 5.2. The share of employed persons among those with disabilities has increased.
Disability benefit recipiency tends to have a long-term dimension. Almost all disability beneficiaries had only one spell during the period from January 2012 to October 2017 (Figure 5.3) and that spell was relatively long. Only 16% of all beneficiaries had short disability benefit spells of up to 12 months. However, the actual share of such spells is even lower since a large number of the observed spells are left or right censored.
Note: The figure covers the population aged 18 and over due to data limitations that do not allow including persons between 15 and 18 years old. The period covered is January 2012 to October 2017. Disability benefits comprise various types of state disability related benefits, including disability pension, state social security disability allowance, benefits for persons with disabilities in need of care etc. Close to 91% of all disability benefit spells are censored. The number of months on disability benefit as well as the number of spells are calculated using all spells, including censored ones. In case of 1-2 months short interruptions, spells were merged.
Source: Latvian Social Insurance Agency and OECD estimates.
These trends reflect a low probability of leaving the disability benefit as shown in Figure 5.4. Given that official disability status is usually related to illnesses and chronic conditions, the probability of leaving the disability benefit is very low at any point of the spell, and shows a typical declining pattern over time and as the disability benefit spell becomes longer (Figure 5.4). Spikes in the probability of exit are observed every 6 and 12 months, when the eligibility for disability benefit grant is usually reconsidered.3 The highest probability of exit (2.6%) is observed at the end of a full year of benefit receipt, while the probability of exit is reduced to half as soon as the first year has passed.
Note: The figure covers the population aged 18 and over due to data limitations that do not allow including persons between 15 and 18 years old. The period covered is January 2012 to October 2017. Disability benefits comprise various types of state disability related benefits, including disability pension, state social security disability allowance, benefits for persons with disabilities in need of care etc. Close to 91% of all disability benefit spells are censored. The number of months on disability benefit as well as the number of spells are calculated using all spells, including censored ones. In case of 1-2 months short interruptions, spells are merged into one spell.
Source: Latvian Social Insurance Agency and OECD estimates.
The main reason for leaving the disability benefit scheme is because of a transition to the old age state pension (Figure 5.5). Although about one third of all 18-64 year old beneficiaries were employed one month after the end of their disability benefit spell, those were mainly persons who remained active in the labour market during their official disability period, possibly because they had less severe health conditions.
Non-employment during the disability benefit spell is a strong predictor of non-employment after leaving the disability benefit. Less than 2% of all persons who were not employed during the last six months of their disability benefit spell transited into employment right after exiting the benefit (the following month). The respective share six months after leaving the disability benefit is only moderately higher. About two thirds of the persons who were not employed during (at least) the last half year of their disability benefit spell were pre-pension age persons with disabilities who made the transition to state pension right after or soon after they exited the disability benefit scheme.
For the younger groups of disability beneficiaries, exit from the scheme is not strongly associated with employment. This is explained by a number of factors. First, withdrawal from the labour market during the disability period has a substantial negative impact on further employability. Second, persistent health deteriorations often become a serious barrier both for employability and willingness to look for a job. In Latvia, severe and moderate limitations in activities are found to have a strong negative effect on the willingness to work among men and women aged 50 and above as well as on the employability of those willing to work (The World Bank, 2015[2]). Furthermore, both age and health discrimination may play a role.
Note: The figure covers disability benefit recipients aged 15-64, who exited the benefit scheme during January 2012 and October 2017. Disability benefits comprise various types of state disability related benefits, including disability pension, state social security disability allowance, benefits for persons with disabilities in need of care etc. Close to 91% of all disability benefit spells are censored. The number of months on disability benefit as well as the number of spells are calculated using all spells, including censored ones. In case of 1-2 months short interruptions, spells were merged. The year refers to the month following the last month of the benefit spell (or the sixth month after the last month of the benefit spell in the lower part of the figure).
Source: Latvian Social Insurance Agency and OECD estimates.
Unemployment benefits cannot lead to benefit dependency
As described in Chapter 2 of this Review, unemployment benefits in Latvia are provided for a maximum of nine months and gradually decline after the first three months, providing incentives to the unemployed to look actively for work shortly after falling into unemployment. This early phasing out of unemployment benefits also create the incentives for the unemployed to participate in ALMPs which offer a stipend that complements their unemployment benefit while raising their chances of finding work. These could include for instance training programmes and special work experience programmes for youth and the regional mobility support provided during the participation in the ALMPs (training programmes are discussed in Chapter 3 of this Review, whereas the regional mobility programme is discussed in Chapter 4).
Because of these features, reliance on unemployment benefits can only be of temporary nature. During the period from 2012 to 2017, slightly more than one third (34%) of all unemployment benefit recipients received the unemployment benefit for the entire length of their entitlement (Figure 5.6, Panel A). Although the probability of an early exit from the unemployment benefit scheme is rather low (Annex Figure 5.A.1), in total about a half of recipients exited the benefit (and the status of registered unemployed) within six months. This possibly reflects the reduction of the monthly payment to 50% of the granted amount after the sixth month. The vast majority (77%) of unemployment benefit recipients had only one unemployment benefit spell (Figure 5.6, Panel B). During the incomplete six-year period analysed in Figure 5.6, only a quarter of all beneficiaries had two or more unemployment benefit spells.
Note: The sample includes beneficiaries of unemployment benefits aged 15-64, from January 2012 to October 2017. About 20% of all unemployment benefit spells were censored. The number of months on benefit was calculated after removing all censored spells, whereas the number of spells was calculated using all spells, including censored ones. In case of short interruptions, spells were not merged. While the information on the dates and therefore the exact length of unemployment benefit spells is not available, the number of months on Panel A reflects the number of months when a person received the unemployment benefit. Therefore, spells which seem to last ten months correspond to unemployment benefits for which payments were made by the SSIA in ten different months but they truly correspond to nine-month unemployment benefit spells.
Source: Latvian Social Insurance Agency and OECD estimates.
Figure 5.7 reproduces the analysis in Figure 5.6 for the four triennial periods covered by the available data. In the period between 2015 and 2017 the share of beneficiaries who had only short unemployment benefit spells (up to three months) was higher than that during the pre-crisis period (2006-08), and the share of beneficiaries who remained on unemployment benefits for the full nine-month period was slightly lower than in 2006-08.
However, as shown in Figure 5.7 (Panel B), the average length of the unemployment benefit spell has been relatively stable and has decreased only slightly since the economic crisis (from 6.8 months in 2009-11 to 6.3 months in 2015-17). The rigidity of this indicator is explained, to a large extent, by changes in the normative acts and a set of measures put in place in response to the crisis. On the one hand, these measures expanded coverage of unemployment benefits during the economic recession by lowering the minimum insurance contribution period necessary for unemployment benefit entitlement.4 On the other hand, these measures aimed at restricting spending via reduced monthly benefit payments for the groups of unemployed with contribution history of less than 20 years.5
Note: The sample includes beneficiaries of unemployment benefits aged 15-64, from January 2012 to October 2017. The number of months on benefit was calculated after removing all censored spells. In case of short interruptions, spells were not merged. While the information on the dates and therefore the exact length of unemployment benefit spells is not available, the number of months on Panel A reflects the number of months when a person received the unemployment benefits. Therefore, spells which seem to last ten months correspond to unemployment benefits for which payments were made by the SSIA in ten different months, but they truly correspond to nine-month unemployment benefit spells. Calculations for the period 2015-2017 are based on periods of 34 months given that year 2017 in the available data is incomplete (January-October).
Source: Latvian Social Insurance Agency and OECD estimates.
Dependency on GMI benefits seems limited
GMI beneficiaries are required to register with the SEA and sign an agreement (a “cooperation plan”) with the municipal social services that are responsible for the management and delivery of social benefits. The benefit is granted for three to six months for persons registered with the SEA but only for one month for those who have not registered. In those cases, the benefit can only be extended if the person acquires the status of registered unemployed in the meantime. The social situation of GMI beneficiaries and their families is reassessed every three months to determine continuation of the benefit or its suspension. Benefit receipt is suspended if a person fails to respect either the individual job search plan agreed with the SEA (and loses its status as registered unemployed) or the cooperation plan signed with the municipal services.
GMI can be received by people who satisfy the means-tested eligibility threshold (EUR 49,80 per month per household member in 2013-17). This implies that in a large household, it is possible to establish GMI eligibility even if there is one employed member. The remaining household members have to be registered with SEA, with exceptions for several groups defined by the Law on Social Services and Social Assistance.6 GMI and unemployment benefits can be received simultaneously only if the unemployment benefit is relatively low so that the average income per household member is below the threshold. This is more likely to be true in the last three months of the unemployment benefit spell when the unemployment benefit is reduced to half the size during the first six months of unemployment benefit receipt.
The number of GMI beneficiaries is small and declining. In 2017, the percentage of people receiving GMI benefits reached just 0.5% of the population aged 15 and more, four times smaller than that in 2012 (Figure 5.1). According to estimates by Gotcheva and Sinnott (2013[9]), even during the economic recession, reliance on the GMI was relatively low with no more than 4% of the population receiving GMI benefits during the crisis.
The duration of GMI spells and the number of GMI spells during the period from January 2012 to October 2017 are presented in Figure 5.8. They are calculated using all spells, including censored ones which represent about 37% of all benefit spells. In case of short interruptions of one or two months, spells were merged and considered as one spell. The average time spent on the GMI is relatively short. The spikes that are shown every three months (at three, six, nine, etc. months) indicate the reassessment of GMI eligibility every three months. From 2012 to 2017, one third of GMI recipients received the benefit for up to three months and another 25% for up to six months. The majority of beneficiaries have had only one GMI spell during this period.
Note: GMI: Guaranteed Minimum Income. The sample includes GMI beneficiaries aged 15-64 years old, from January 2012 to October 2017. The number of months on GMI as well as the number of spells was calculated using all spells, including censored ones which represent about 37% of all benefit spells. In case of 1-2 months short interruptions spells were merged and considered as one spell.
Source: Latvia’s Municipal Information System Database and OECD estimates.
More beneficiaries use GMI as either a short-term support measure or as longer term support
Figure 5.9 presents the duration of GMI benefit receipt for the four two-year periods covered by the data: 2010-11, 2012-13, 2014-15 and 2016-17. During the 2010-11 period, only a quarter of GMI recipients received the benefit for more than one year, whereas this share increased to one third in the last two periods (2014-15 and 2016-17). The distribution of GMI recipients by the time spent on the GMI benefit has indeed changed over time. On the one hand, the share of persons who used GMI as a very short-term support measure (for no more than three months) has increased in comparison with the economic recession (2010-11). On the other hand, the share of GMI beneficiaries who received the benefit for 19 months or longer has become relatively more important in the recent years.
Note: The figures comprise recipients of Guaranteed Minimum Income (GMI) benefits, aged 15-64 from January 2010 to October 2017. The number of months on GMI was calculated using all spells, including censored ones which represent about 37% of all benefit spells. In case of 1-2 months short interruptions spells were merged and considered as one spell.
Source: Latvia’s Municipal Information System Database and OECD estimates.
Although the number of GMI recipients has decreased, the share of GMI beneficiaries who rely on the benefit for an extended period of time has grown. The probability of leaving the GMI declines with the time spent on it and after 18 months (without interruptions exceeding two months) the probability of exiting the scheme is particularly low and does not reach 8% at any point (Figure 5.10). The spikes observed in Figure 5.10 are explained by the reassessment of eligibility implemented by the social services every three months.
The increasing share of persons who receive the GMI for long periods highlights the need to analyse their characteristics and understand the barriers that make them dependent on social assistance with the aim to identify the policies that would help bring them closer to the labour market.
First, a comparison between GMI beneficiaries and non-beneficiaries (Annex Figure 5.A.3) shows that GMI beneficiaries are on average older than non-beneficiaries. Close to half of all GMI beneficiaries are 50 years old or above, while the respective share is only 29% among non-beneficiaries. Men are relatively more likely than women to receive GMI. Persons with disabilities represent 17% of all GMI beneficiaries, whereas their share among GMI recipients is only 6%. Old age and disability status are also correlated with reliance on the GMI benefit for longer periods (Figure 5.11). Longer GMI receipt is also more prevalent than shorter GMI spells among single-person households and urban areas.
Note: Guaranteed Minimum Income (GMI) beneficiaries aged 15-64 are included in this analysis and covers the period from January 2012 to October 2017.
Source: Latvia’s Municipal Information System Database and OECD estimates.
Old age and disability status have become more important in explaining receipt of GMI benefits for longer periods (24 months or more7) (Figure 5.12). In 2016-17, 61% of those who spent the entire period on the benefit were 50 and above, which is 23 percentage points higher than in 2012-13. Job prospects for younger groups are better, while older age discouraged unemployed persons are typically more difficult to activate (Ferré, Immervoll and Sinnott, 2013[10]; The World Bank, 2015[2]). Raising employment rates for the pre-retirement age population requires special efforts, especially after long inactivity periods.
Close to 45% of all persons who spent close to two years on the GMI benefit scheme in 2016-17 had the official disability status. This share is more than double their share in 2013-14, and close to four times higher than the share of persons who stayed on GMI benefits for up to six months in 2016-17 (Figure 5.11). Although the disability benefit exceeds the income threshold used to determine GMI benefit eligibility, according to the SOPA data, about half of all GMI beneficiaries lived in two or more person households, therefore per capita income was below the defined threshold.
Men and persons living alone relatively more often than before are found among those who spend longer time on the GMI benefit (Figure 5.12). Ethnic minorities are more often dependent on GMI than ethnic Latvians: more than 2/3 of all beneficiaries who stayed on the scheme for the entire biennial period (2016-17) were non-Latvians, of whom nearly a half had no Latvian citizenship.
Note: This figure includes all 15-64 year old recipients of Guaranteed Minimum Income (GMI) benefits, from January 2016 to October 2017. The number of months on the GMI benefit is calculated using all spells, including censored ones. In the case of 1-2 month short interruptions, spells were merged. Education is based on the SEA data and supplemented using SOPA data. Household size is derived from the SOPA data. Age and other personal characteristics indicated on this figure refer to January 2016.
Source: Latvian State Employment Agency, Latvian Office of Citizenship and Migration Affairs, Latvia’s Municipal Information System Database and OECD estimates.
Note: This figure includes all 15-64 year old recipients of Guaranteed Minimum Income (GMI) benefits, from January 2012 to October 2017. For 2016-2017, the total number of months spent on the benefit is limited to 22 months. The number of months on the GMI benefit is calculated using all spells, including censored ones. In the case of 1-2 month short interruptions, spells were merged. Education is based on the SEA data and supplemented using SOPA data. Household size is derived from the SOPA data. Age and other personal characteristics indicated on this figure refer to January of the first year of the two-year period.
Source: Latvian State Employment Agency, Latvian Office of Citizenship and Migration Affairs, Latvia’s Municipal Information System Database and OECD estimates.
People transition between benefits and labour market statuses
This section presents labour market transitions taking place between January 2016 and January 2017. The possible statuses retained for this analysis (presented in Table 5.2) are selected on the basis of their frequency and relevance for the analysis in this Chapter. More specifically, the analysis considers a combination between labour market statuses (employment, unemployed and out of the labour force), benefits (disability benefits, GMI, unemployment benefits) and employment-related activation measures (employment subsidies and public works programmes).
The majority of employed persons are still in employment one year later, but some changes are taking place in terms of the benefits they receive while in employment. For instance, 29% of employed persons receiving GMI and disability benefits transition into full disability benefit one year later, whereas close to half of employed GMI beneficiaries are off the benefit one year later. The vast majority of recipients of disability benefits who are in employment remain in employment one year later (81%), and an additional 4% are employed without relying on a disability benefit.
A positive result is found for unemployed persons who participate in the employment subsidy programme (the programme is discussed in detail in the next section). More than one quarter of the registered unemployed who are in subsidised employment and do not receive any benefits, are employed without support one year later. An additional quarter of them are still on subsidised jobs one year later, mainly groups that are entitled to a subsidy for up to two years.
These descriptive results are less positive for disability beneficiaries. Close to one third of disability benefit recipients who are registered with the SEA as unemployed continue to receive the benefit one year later, but are out of the labour force. Disability benefit recipients who are out of the labour force are the least likely to change status than any other group. Only 5% of them return to employment (with disability benefits) and another 5% are out of the labour force without access to disability benefits one year later. For disability beneficiaries who are registered with the SEA, the most common transition is to GMI benefits while maintaining their status as registered unemployed.
GMI beneficiaries are the most likely to change status than most other groups considered in Table 5.2, highlighting the temporary nature of this benefit as already discussed in the previous section. Among GMI beneficiaries who are registered with the SEA and also receive unemployment benefits, one third are employed one year later and no longer receive the GMI.
Programmes for subsidised employment
Latvia has a well-developed system which offers support for employment in the private sector (employment incentives) to the most disadvantaged unemployed groups. These consist of the following groups: persons with disabilities, persons who have been unemployed for at least 12 months, persons aged 55 or above, persons who have at least one dependent and persons who have obtained a status of refugee or alternative status. The programmes in place in Latvia are in line with the European Council recommendations on establishing a Youth Guarantee (European Union, 2013[11]) and the long-term unemployed (European Union, 2016[12]) which call for well-targeted employment subsidies for these groups of the unemployed. The programmes offered under the Guarantee programme were offered in a similar form prior to the introduction of the Youth Guarantee and are expected to continue after the end of 2018.
Latvia’s tight targeting of these measures to the most vulnerable groups is in line with the evidence on the effectiveness of subsidy programmes. Most existing papers find the largest effects of wage subsidies on the most vulnerable groups of unemployed (Card, Kluve and Weber, 2018[13]). These programmes provide support to groups for which other less costly and radical solutions are unlikely to work. In addition, Latvia offers a number of services and measures that aim to help disadvantaged persons prepare themselves for the labour market through counselling, specialised training (discussed in Chapter 3) and concrete support.
Activities offered to the long-term unemployed have been intensified over time and additional activities were introduced in 2016 including individual and group consultations, health checks, determination of professional suitability, motivation programme, social mentoring, and therapies for dependent persons.
Two main trends can be observed in Table 2.1 in Chapter 2 of this Review, which presents the number of participants in all the different categories of ALMPs in Latvia. First, temporary public works were scaled up during the crisis but were subsequently reduced during the recovery period, and second, there was a sizeable increase in participation of young persons in activation measures, related to the introduction of the Youth Guarantee.
Participants in subsidised employment programmes represented 3.3% of all participants in ALMPs in 2017, excluding participants in general support measures and the public works programme. This is similar to their share in 2012, but less than half the share in 2015 (8%) and well below the share in 2016 (5.6%). In contrast, participants in temporary public works represented 13% of all ALMP participants in 2017 (excluding participants in general support measures), down from 52% in 2012, when this scheme was used intensively to provide support against poverty to households heavily hit by the economic crisis and enable them to stay close to the labour market (income support combined with activation).
Spending on employment incentives represented about one fifth of all spending on ALMPs in Latvia in 2016, slightly above the OECD average share of 19% (Figure 5.13). The notable increase in 2015 reflects the new European Social Funds made available for the Youth Guarantee. Spending on employment subsidies as a share of GDP was fairly low in 2016, about 0.04%, versus 0.1% in OECD countries on average. Spending on direct job creation has gradually declined from 34% of all ALMP expenditures in 2011 to 10% in 2016.
Note: OECD is an unweighted average and excludes Iceland and Lithuania for all years; the United Kingdom for years 2012-2016; and France, Greece, Italy and Spain in 2016.
Source: OECD/Eurostat Labour Market Programme Database, https://doi.org/10.1787/data-00312-en.
This chapter focuses on the two main employment subsidy programmes: i) the programme of subsidised employment for vulnerable groups (Pasākumi noteiktām personu grupām) which targets persons with disabilities, individuals older than 55 years and the long-term unemployed; and ii) the programme of subsidised workplace for young unemployed (Subsidētā darba vieta jauniešiem), offered under the Youth Guarantee between 2014 and 2018 (Table 5.4). The exact eligibility conditions for these two programmes are presented in Table 5.3. In addition to this programme, young unemployed persons have access to a number of programmes offering them employment opportunities in the private sector, but the conditions (level and duration) of the monthly support provided under those schemes are less generous than the main employment subsidy programme for youth. Therefore, their impact is not examined in this chapter.
Employment subsidies are available for vulnerable groups and youth
The programme for vulnerable groups is a wage subsidy paid to employers, which can cover up to 50% of the total wage cost. It cannot exceed the minimum wage set by the government (EUR 430 from 1st January 2018). For persons with disabilities, the subsidy is higher and may be up to 1.5 times the minimum wage, except when it concerns persons employed in low-skilled jobs for which the maximum wage subsidy cannot exceed the minimum wage. The amount of the subsidy is fixed and does not depend on the degree of disability and the disability group the person belongs to.
The subsidy is granted for up to 12 months for persons who have been unemployed for at least one year, those who are 55 years old or older and refugees (see Table 5.3 for a list of the eligibility conditions and programme characteristics). It is granted for up to two years for persons who have been unemployed for at least two years and those who have been unemployed for 12 months and have either at least one dependent member or are older than 55 years. For persons with disabilities, the subsidy is also granted for up to two years. An eligible person can only participate in the programme again after one year has passed following the end of their previous participation.
This programme aims to integrate a total of 5 177 disadvantaged unemployed persons into the labour market in the period from February 2015 to December 2022. The programme is running with support from the European Union (European Support Fund for EUR 30 million) and some contribution from the state budget (EUR 3.7 million) as well as private funding (EUR 14 million) paid in the form of wages by employers.
The SEA allocates the number of places on the employment subsidy programme – and the corresponding resources – to each local office, according to a number of criteria. These include the number of unemployed who are registered in the specific local office in the target groups over a 12-month period (notably persons who have been unemployed for at least 12 months, those who have been unemployed for at least 24 months, and unemployed persons aged 55 or more). For unemployed persons with disabilities, the number of subsidised jobs at each SEA local office is calculated taking into account the number of registered unemployed with disabilities in the local office and the number of subsidised jobs from the previous year that are ongoing. The demand from the local office is also taken into account. From 2019, staff workload and the number of registered vacancies in the office are also being taken into account in the calculation of the number of subsidies allocated per local office.
This funding mechanism enables the SEA to achieve an effective allocation of resources based on the needs of the clients of the specific local office while minimising the burden on local offices related to budget planning. Prior to the introduction of this mechanism, the number of subsidised jobs granted to local offices was based on the degree of cooperation between the local offices and employers in the area, which created many imbalances and was not successful in serving the needs of the unemployed. Moreover, there is some degree of flexibility in the system that allows heads of local offices to request additional subsidised jobs to be assigned based on unexpected changes in the local labour market. The SEA central office examines these requests and may decide in favour of the allocation of additional funds for this programme in a specific local office in need.
Once the number of subsidies and corresponding budget has been assigned to the local office, a call for applications is issued by the local SEA offices targeting employers who would be interested to participate in the programme. All types of enterprises, except medical institutions and education establishments, as well as self-employed persons, societies (with the exception of political parties) or foundations and agricultural services cooperatives are eligible to benefit from support under this programme. The employers submit their demand and accompanying documents8 to the local SEA office, confirming their compliance with tax and other duties. A Committee set up at the local SEA office reviews all applications and selects the employers who will participate in the programme. At a second stage, the SEA and the potential employer jointly select among the eligible unemployed persons based on skill requirements for the position.
Subsidies are provided only for new jobs that have been vacant for a minimum of four months. Moreover, the selected beneficiary should not have been an employee of the specific employer during the past year. These requirements are monitored and enforced by the SEA and the restrictions imposed are in line with similar restrictions in other OECD countries which aim at minimising the possible displacement effects and deadweight losses of employment subsidies. If the employer is found to breach one of the requirements attached to the contract, the employment subsidy is terminated and the employer is excluded from participation in this measure for two years.
Employers are required to assign a qualified supervisor (who is paid a wage supplement) for every unemployed person hired through the employment subsidy programme. The supervisor is in charge of supporting the employee and helping them acquire the skills required for their job. Supervisors can be responsible for no more than two unemployed persons. The degree of involvement of the supervisor depends on the complexity of the work and the profile of the employee, with youth and persons with disabilities usually receiving more in-depth support. For unemployed persons in high-skilled occupations, supervisors should demonstrate that their qualifications match with the occupation and type of work performed by the unemployed person and that they have relevant professional experience.
In the case of persons with disabilities, certain adjustments in the workplace (e.g. related to access and daily working conditions) are necessary to enable the person to access the workplace and fulfil his/her obligations. An expert’s view is typically required to identify the necessary adjustments. Following this assessment, the SEA can decide to pay up to EUR 711 to cover part of the cost for adapting the workplace and the involvement of experts required for persons with disabilities. Regional mobility support (see Chapter 4 of this Review) and sign language interpreters can also be provided to persons with disabilities, according to their needs.
A well-developed system of support for unemployed youth
Well before the introduction of the Youth Guarantee by all the Member States of the EU, Latvia had in place a number of programmes, which aimed to help youth acquire their first labour market experience and make a successful transition from school into the labour market. OECD (2015[14]) discusses in detail the programmes and services available for unemployed and inactive young persons registered with the SEA as well as the challenges that Latvia faces in reaching out to young NEETs to improve their connection with the labour market.
The main programme offering subsidised employment to youth is “Subsidized workplace for young unemployed” (Subsidētā darba vieta jauniešiem), that offers a monthly wage subsidy to employers who employ an unemployed youth for up to six months. Unemployed persons aged 15-29 who have been unemployed for at least six months, or who have not completed their full-time education more than two years ago and have not yet got a permanent job, are eligible for participation in this programme. The subsidy level is the same as for the programme targeting the vulnerable groups. For young persons with disabilities, it is equal to the minimum wage for low-skill jobs (elementary occupations) and cannot exceed 1.5 times the minimum wage. For all other young unemployed, the subsidy covers 50% of the total wage and cannot exceed the minimum wage.
As for the programme for adults, youth cannot be offered two subsidised jobs in a row: there is a minimum requirement of one year between two employment subsidy spells. As for the programme for vulnerable groups, some exceptions are permitted for interruptions that last less than half of the intended time of participation..
Young unemployed can also receive support through the programme “First work experience for youth” (Pirmā darba pieredze jaunietim), which offers youth the opportunity to gain work experience for up to one year in new jobs. Employers receive EUR 200 during the first six months and EUR 160 during the last six months of the programme, while a higher subsidy is paid for persons with disabilities. Additional expenses are covered for supervisors (50% of the minimum wage for the first three months), for adapting the workplace to the needs of persons with disabilities and involving relevant experts such as assistants, sign language experts etc.
For unemployed youth wishing to acquire some work experience in Non-Governmental Organisations (NGOs), the programme “First work experience for youth in NGOs” (Darbam nepieciešamo iemaņu attīstība nevalstiskajā sektorā) offers limited support to participants (EUR 90 a month) for up to six months. One of the aims of this measure was to offer opportunities for work experience to youth when job opportunities were scarce. A total of more than 4 000 young unemployed individuals have participated in this programme since the beginning of the Youth Guarantee in 2014. A similar programme was also running prior to the Youth Guarantee, with participation of about 1 000-1 400 persons per year.
Moreover, workshops in vocational schools are offered to unemployed youth to help them make career decisions. These workshops give youth the possibility to try up to three different professions (on average two weeks experience per occupation) and offer them a monthly allowance. The strength of this programme is the combination of practical experience (minimum 60% of the time) with theoretical courses.
Young persons, like workers of all ages, can receive support to accept training and job offers outside the area they live in. This mobility programme is described in detail in Chapter 4 of this Review where its impact is estimated.
Finally, the programme “student summer work” offers students in secondary education the possibility to acquire work experience during their summer holidays. Students also receive career guidance. Municipal and other public institutions account for a large share of the employers. This programme attracts the largest number of participants among youth, reaching 4 239 persons in 2016 and 4 975 in 2017.
Only a small share of employers use the subsidies
On the employers’ side, subsidies reduce the financial costs or risks associated with unknown productivity of the person to be employed. As with employment services, this is a scheme which is particularly relevant to youth entering the labour market for the first time, and whose (perceived) marginal productivity may be below market wages. Employment subsidies may also serve to lower the costs to employers of providing on-the-job training to youth. Such training subsidies offer the possibility of expanding the number of work-based training places for disadvantaged young people.
To avoid displacement (substitution) effects, wage subsidy programmes in OECD countries are available only for newly created jobs and/or impose a minimum period during which jobs should be advertised before the subsidy beneficiary can be hired. In some countries, the employment subsidies become available only when the total employment at the firm level has actually increased, to ensure beneficiaries do not displace other workers (Boockmann, 2015[15]). There is no such conditionality attached to the programme in Latvia but there is a requirement for the vacancy to be advertised for at least four months before a subsidised employee can be hired. In addition, the selected candidate should not have been an employee of the specific firm in the past year.
There is not sufficient evidence in the existing literature regarding the displacement effects of subsidies. Van Reenen (2004[16]) examines these effects in the case of the New Deal for Young People in the United Kingdom, finding little or no evidence of substitution effects against older unemployed persons.
Heavy bureaucratic procedures and stringent conditionalities which are sometimes attached to employment subsidies deter employers may lead to low participation rates of employers, especially when the amount of the benefit is relatively small. For example, stringent conditions relative to the perceived value of the subsidy by employers in the case of the French Contrat Jeune en Entreprise led to a very low take up of the programme (Roger and Zamora, 2011[17]). This programme offers a hiring subsidy to school dropouts for a three-year period and employers are required to retain the workers for the entire duration of the subsidy (except for reasons related to professional misconduct). In the case of the German Immediate Action Program for Lowering Youth Unemployment or Jugend mit Perspektive (JUMP), which combines a relatively generous benefit to employers, the conditionality of no early dismissal and a post-participation retention period (of half the period of the subsidised job) is not found to discourage employers from participating (Caliendo, Künn and Schmidl, 2011[18]).
In the case of Latvia’s employment subsidy programmes, the share of employers receiving support is small but has somewhat increased over time, reaching 1% of all employers in 2017 from 0.85% in early 2012 (Figure 5.14). For employers using this measure, subsidised employees represented on average about 30-33% of their staff in 2017.
Source: Latvian State Employment Agency and OECD estimates.
Retail trade firms represent the vast majority of employers using hiring persons on employment subsidies (70% in 2017), followed by farmers and fishermen and foundations and associations. In terms of sectors, 18% of these employers are in agriculture, forestry and fishing and another 17% are in trade. Manufacturing and other services account for close to one quarter of all employers using subsidised labour.
Participation in the scheme is associated with certain administrative burden for employers during and at the end of the programme, in addition to the process for applying for participation in the scheme described in the previous pages. Employers have to submit a monthly report to the SEA on the hours worked and wages, so that the subsidy can be calculated and paid, which can be time consuming. Time can be cut if this process is automated so that there is no need to involve administrative staff at the firm that hires the subsidised employee. The administrative burden that employers have to bear both before and during participation in the programme can be particularly constraining for small businesses, especially the first time they apply for participation in the programme. Between 2012 and 2017, small businesses of up to four employees represented close to half of all firms that participated in the scheme. Somewhat less than half of these firms were very small with one or two employees who are unlikely to have any dedicated administrative staff and accountants who can deal with the exchanges with the authorities regarding the requirements and conditions for participation in the scheme. For these firms in particular, it is important to minimise the administrative burden by using Latvia’s well-developed IT system and interconnected databases: for example, such systems could be used to transmit the monthly information required on hours worked to calculate the amount of the subsidy to be paid.
Repeated participation is kept to a minimum
As in many other OECD countries, Latvia imposes a minimum gap of one year between two participations in employment subsidies for each person. This restriction is lifted for breaks which are due to reasons beyond the control of the unemployed person and for breaks that have lasted less than half of the intended period of participation. These exceptions allow treating the two participations as one single spell, interrupted for some well-defined reason.
Data from the SEA covering the period from January 2012 to October 2017 show that one out of ten beneficiaries of employment subsidies participate in the measure more than once (Figure 5.15). This is clearly a lower bound of repeated participation as programme participation prior to 2012 and post 2017 is not recorded in the data. In contrast, the share of persons participating twice in the subsidy programme targeting youth is very low (3%), most likely because of the age limit that restricts the time during which a young unemployed person can benefit from this programme. When other youth employment-related measures are considered, repeated participation goes up to 7%. For other employment-related support measures (which have not been included in the categories discussed above), close to one third of participants have multiple participations, with 14% of persons participating at least three times during the 5.5 year observation period.
Note: Includes all participations between January 2012 and October 2017.
Source: Latvian State Employment Agency and OECD estimates.
From the side of the employers, there is no obvious indicator to assess whether some firms use the employment subsidies as a permanent way to hire workers. An analysis focusing on firms that hired at least one subsidised employee between January 2012 and October 2017, finds that these firms used subsidised employees for 28% of the period they had any recorded employment (with the median being 16% of the period with recorded employment). These firms used the subsidies on a continuous basis for 12 months on average (the median is seven months).
Data-related limitations need to be acknowledged
Despite the rich and detailed data compiled by the Latvian authorities and provided for the analysis in this Review, there are still a number of data-related issues that need to be discussed before turning to the analysis of the effectiveness of employment subsidies. First, a careful examination of the subsidy duration in 2013 and 2014 (Table 5.5) and the trends in participation in the different programmes presented in Table 5.4 may indicate that some participants in the youth programme in 2013 and 2014, just before and during the first year of the Youth Guarantee (fully implemented in 2015), may have been coded under the programme for vulnerable groups. The introduction and phasing in of the Youth Guarantee measures would explain the drop in the number of participants in the youth programme from 523 in 2012 to 152 in 2013 and 283 persons in 2014, before increasing again to about 500 persons per year in 2015-2017. At the same time, a temporary increase in the number of participants in the programme for vulnerable groups is observed in 2013 and 2014.
Second, matching subsidy spells with jobs spells is challenging. For the vast majority of participants (95%), the start date of the subsidy coincides with the start date of their job (for all their employment subsidy spells).9 However, for 336 participants who represent 5% of all participants, at least one employment spell starts before the subsidy spell starts. The median difference between the start of the employment spell and the start of the employment subsidy spell is 12 months (for the first spell of employment subsidies) so it is difficult to attribute this to administrative delays. These observations have been dropped from the analysis that follows.
Moreover, prior unemployment status and unemployment duration are not accurately recorded for all registered unemployed. A careful examination of persons who receive a wage subsidy shows that about 20% of them were not recorded as unemployed in the period(s) prior to participation or were just recorded as unemployed only in the month before participation started. Some of these cases, had an unemployment spell within a year prior to participation in the subsidy programme, which was followed by a period when they were neither unemployed nor employed.
Who are the programme participants?
An analysis of the characteristics of the participants at the time they took up subsidised employment closely reflects the targeting of the programme. Figure 5.16 (Panel A), which draws the age distribution of participants at the start of their subsidised job, shows a clear spike for youth (around 24-25 years) and a second – less clear – spike at 55-56 years.
Note: UN: unemployment duration. These figure include participants in the programmes for vulnerable groups and youth from January 2012 to October 2017. Individual characteristics are measured at the start of the subsidised job. In Panel B, the shares of persons aged 55 and over and youth exclude persons with disabilities who are taken into account in the “persons with disabilities” category.
Source: Latvian Office of Citizenship and Migration Affairs, Latvian Social Insurance Agency, Latvian State Employment Agency and OECD estimates.
Persons with disabilities, one of the target groups of the programme for vulnerable groups, represent close to one third of all participants (Figure 5.16, Panel B). Among the remaining participants, 27% are young persons aged up to 29 years, 11% are individuals aged 55 and over, 6% are persons who had been unemployed for 1-2 years prior to participation and 1.2% are long-term unemployed (for two or more years). The remaining 23% comprise other eligible individuals such as refugees, those with alternative status, and possibly other cases, on which it may be difficult to obtain accurate information. The low share of LTU among participants and the high share of all other remaining groups likely reflects the data limitation in the coding of prior unemployment duration described above.
Programme participation is equally split between men and women. More than half of all subsidy beneficiaries have secondary education (56%) and 21% have primary education only. Among young participants, three quarters have at most secondary education, while individuals with higher education represent just 13% of all participants. Beneficiaries of Slavic origin represent 36% of participants in the programme for vulnerable groups and 26% of participants in the youth programme. Only one quarter of persons in subsidised jobs live in households with children.
The distribution of participants across regions is driven by the distribution of subsidies across SEA local offices, described in earlier sections of the Chapter and tied to local labour market conditions and the size of the target groups in the area. Latgale represents 45% of all participants in the programme, followed by Kurzeme (15%) and Zemgale (12%). In contrast, only 17% of participants live in Riga or the Pieriga region.
Disaggregating the data according to the time spent in unemployment prior to participation, more than half of all employment subsidy recipients had been unemployed for 6 months or less (Figure 5.17). Persons with disabilities represent 28% of this group of short unemployment duration and youth (without disabilities) represent 32%. About 20% of all programme participants had been unemployed for up to 12 months prior to participation. Participants who had been unemployed for 13 to 24 months represented 16% of the total.
Note: Includes participants in the programmes for vulnerable groups and youth from January 2012 to October 2017.
Source: Latvian Social Insurance Agency, Latvian State Employment Agency and OECD estimates.
Programme duration varies by target group
The majority (84%) of all of the subsidy spells described above, for which the start date is observed, also have an end date within the observation period. The remaining 1 228 spells are right censored and are excluded from the analysis that follows. An analysis of the non-censored participation spells reveals two clear spikes in their duration at 11-12 months and 23-24 months. These correspond well to the expected duration of employment subsidies as described in Table 5.3. More than one third (38%) of all completed spells last for 6 months or less (the typical duration for subsidies to young persons) and an additional 44% last between 7 and 12 months. Another 17% of these spells go up to two years, which is the maximum duration for persons with disabilities, the LTU and some specific categories of the unemployed (Table 5.5).
More specifically, persons with disabilities stay on the subsidy for longer periods than all the other target groups. Their average subsidy spell is 14 months, versus 8.7 months for youth, 11 months for persons 55 and over, and 9.8 months for the entire group of programme participants.
The maximum duration of the subsidy for persons with disabilities increased from 24 to 36 months in 2014, but was cut again to 24 months in 2015. This change is somewhat reflected in the administrative data. Although the actual duration of subsidies for persons with disabilities does not exceed 24 months, an increase in the duration can be observed in 2014. In 2013, 34% of subsidy recipients with disabilities participated in the programme for 12-24 months, whereas in 2014, this share rose to 73%, to decline again in 2015 to 54% (Figure 5.21, Panel A). The reason for reverting to a shorter duration of the subsidy was based on financial considerations and the need to achieve a good predictability of financial resources necessary for this programme and reflect the actual duration of the subsidy for persons with disabilities. Moreover, as also indicated in the administrative data discussed in the previous paragraph, most employers used the subsidy for less than 36 months, hence its duration was de facto below the maximum duration stated in the legislation.
Subsidised jobs differ substantially from non-subsidised ones
It does not come as a surprise that subsidised jobs are in many ways different from non-subsidised ones and are concentrated in specific sectors. Using the Statistical Classification of Economic Activities in the European Community (Nomenclature statistique des activités économiques dans la Communauté européenne, NACE) codes associated with each job, close to 17% of all subsidised jobs10 are in agriculture, whereas the analogous share is only 4% for non-subsidised jobs. “Other services” also account for 10% of all subsidised jobs but only 2% of non-subsidised jobs. In contrast, public administration and transportation and storage, which accounts for 8% of non-subsidised jobs each, are much less prevalent among subsidised jobs (0.5% and 2% respectively).
Vitally, average earnings are lower for subsidised jobs. The average earnings for subsidised jobs are EUR 100 lower than those for no-subsidised jobs. However, this result is mainly driven by the upper tail of non-subsidised jobs. The median of the two sets of jobs is quite similar (about EUR 350), but the 75th percentile is EUR 479 for subsidised jobs, versus EUR 608 for non-subsidised jobs.
Combining different employment-support programmes
Many employment subsidy beneficiaries also benefit from other employment support programmes before, during or after the end of their participation in the subsidy programmes (Figure 5.18). Close to one-third of participants in the subsidy programme for vulnerable groups also participate in public works during the observation period. Two-thirds of them participated first in public works and then received the employment subsidy. There is also an important overlap between the main subsidy programme and participation in the student summer programme (24% of all subsidy beneficiaries). Two-thirds of these persons participated in the student summer programme first and subsequently went on to the employment subsidy programme. For youth, 19% of those participating in the youth subsidy had also participated in the student summer programme first. Around 13% of participants in the youth employment subsidy programme had first participated in some other employment programmes offered under the Youth Guarantee (such as volunteering work in NGOs, or first work experience subsidy). The combination of two types of employment subsidies (youth and vulnerable groups) is rather rare. Only 1.4% of participants in the main employment subsidy programme (for all groups besides youth) had also participated in the equivalent programme offered under the Youth Guarantee. Similarly, only 4% of those who participated in the youth employment subsidy programme had also participated in the main employment subsidy programme.
Note: ALMPs: Active labour market policies. PWP: Public Works Programme. YG: Youth Guarantee. These figures include information on participation in employment-related measures between January 2012 and October 2017. Some individuals may have participated in more than two measures.
Source: Latvian State Employment Agency and OECD estimates.
Labour market outcomes of programme participants
A simple analysis of the post-participation outcomes of subsidy beneficiaries reveals that 46% of them are employed one month after the end of the subsidy (Figure 5.19). This share declines to 38% three months after the end of the programme and stabilises around that level up until one year after programme participation. In the first three months after the end of the subsidy, the vast majority of former beneficiaries are employed on the same job as before (87% one month after the end of the subsidy and 70% two months later). However, their share declines over time. One year after the end of their subsidy, 57% of all employed former subsidy beneficiaries are on a new job. These descriptive statistics show that for more than half of those subsidy beneficiaries who remain in or re-enter employment, the programme operates as a stepping-stone for a non-subsidised job. However, this analysis cannot make causal inferences about the possible effect of the programme on post-participation outcomes. For that, it is important to have an appropriately defined control group that is similar to the group of participants. This makes it possible to estimate what would have happened to participants had they not received employment subsidies and, in turn, identify the true effects of the programme on participants’ labour market outcomes.
Note: This figure refers to the period between January 2012 and October 2017 and reports the labour market outcomes of former subsidy beneficiaries one, three, 6 and 12 months after the end of the employment subsidy. It separates out persons who are employed in a new job from those who continue to be employed on the job for which they received the subsidy.
Source: Latvian Social Insurance Agency, Latvian State Employment Agency and OECD estimates.
Impact evaluation of employment subsidies
A wealth of academic papers examine the effects of various ALMPs on participants’ labour market outcomes in many OECD and developing countries and in a variety of settings and labour market conditions (see Card, Kluve and Weber (2018[13]) for a meta--analysis of the results of these papers). Overall, the literature suggests that ALMPs that promote work as early as possible during the unemployment spell such as job search assistance, or incentives to enter work quickly tend to have more positive effects in the short-term (which are also stable over time) than ALMPs that involve investment in human capital (such as training) which can have negative short-term effects but large and significant in the longer term. Moreover, the first group of programmes are found to be more successful for disadvantaged groups of participants who would have a very low probability of entering the labour market without such support, whereas the latter group of ALMPs is more successful for the LTU.
Stylised facts and trade-offs in the existing literature
The majority of papers that evaluate the impact of employment subsidies programmes focus on Austria, Germany, Switzerland and the Nordic countries. These papers demonstrate that the effectiveness of private-sector hiring subsidies depends on their design and targeting and on the size of their indirect effects, such as lock-in effects, deadweight losses and displacement effects (Card, Kluve and Weber, 2018[13]). The empirical literature highlights some important trade-offs policy makers face when implementing employment subsidies. These trade-offs typically depend on labour market conditions and the needs of the target groups.
Just as in the training programmes discussed in Chapter 3 of this Review, programmes providing support for private sector employment may have negative labour market effects in the short-term due to so-called lock-in effects. During their participation in the programme, the unemployed tend to limit their job search activities and hence may be less likely than non-participants or participants in other ALMPs to find a non-subsidised job (van Ours, 2004[19]; Fremigacci and Terracol, 2013[20]; Wunsch, 2016[21]). Lock-in effects may be related not only to the limited time that participants have to devote to job search but also to their actual status vis-a-vis the PES. In many cases, participants are considered as employed and are outside the radars of caseworkers. As a result, they receive limited or no assistance with job search. In addition, job search activities are less well monitored by the PES during their participation in subsidized employment programmes.
The size of lock-in effects depends on the duration of the programme, the state of the labour market (and hence the probability of finding employment quickly for non-participants with similar characteristics), and the timing of participation within the unemployment spell (See Wunsch (2016[21]) for a summary of the factors determining lock-in effects of activation measures). For instance, if participation in the programme occurs during the period when exit from unemployment to employment is highest (e.g. between four and six months of unemployment) the lock-in effects are likely to be higher.
The quality of targeting of employment subsidies is a key determinant of both their effectiveness and their indirect effects. The earlier employment subsidies are provided during the unemployment spell, the sooner they are likely to produce positive effects. Nevertheless, dispersing employment subsidies early may raise the likelihood of participants foregoing other employment opportunities as well as increasing the potential deadweight losses which appear because some participants might have found a job anyway, even without the help of a subsidy (Wunsch, 2016[21]; Boockmann, 2015[15]). Ideally, these programmes should thus be targeted to the unemployed who have already been unemployed for a certain time and need more intensive support to acquire work experience. This is the case for Latvia’s programme for vulnerable groups. Nonetheless, there is a clear trade-off between this effort to minimise indirect effects and the need for early intervention to avoid long-term unemployment (OECD, 2015[22]).
Latvia’s employment subsidies programmes that provide support to persons who have been unemployed for at least 12 months are in line with the evidence from other countries targeting the LTU (Brown, 2015[23]; Brown and Koettl, 2015[24]; Wunsch, 2016[21]). For example, strong effects of employment subsidies have been found for the LTU in Sweden (Sianesi, 2008[25]) and Switzerland (Gerfin, Lechner and Steiger, 2005[26]).
Nevertheless, although tight targeting is necessary to ensure the effectiveness of employment subsidies while minimising their possible negative indirect effects, it can also lead to stigmatisation of participants, who are perceived as low-productivity workers (Brown, 2015[23]).
Methodological choices and issues for discussion
The analysis in this Chapter compares the outcomes of programme participants (treated group) with those of similar individuals who do not (have not) participate(d) in the programme (control group). The motivation to compare individuals who have “similar” observed characteristics comes from the fact that programme participation is unlikely to be random. Only a sub-set of eligible individuals are selected for participation and this selection is made either by employers – who primarily select on the basis of perceived productivity of workers – or by caseworkers – whose choices are assisted but not fully determined by the profiling tool introduced in 2013.
The probability to be treated for eligible unemployed persons is very low. Among persons who have been unemployed for 12 months or longer, the probability to be selected for participation in subsidised employment is 3%. For unemployed youth and persons aged 55 and over, the probability is even lower, 2.1% and 1.5% respectively. Nevertheless, among persons with disabilities, the probability to be treated is slightly higher, at 6.6%.
Quite surprisingly, more than one third (38%) of participants appear not to fulfil the eligibility criteria at the time of the start of their participation in the subsidy programme. However, there is not sufficient information in the data to identify refugees and those with alternative status who would be eligible for participation. Moreover, for youth who represent 40% of the ineligible persons, the information on past history is not sufficiently detailed to characterise perfectly their eligibility status which may explain some of these cases of ineligible youth participating in the subsidy programme.
The time that the “clock starts” (i.e. the moment when an unemployed person becomes eligible for support) both for the treatment and control groups is defined in two ways: i) at 6 months of unemployment; and ii) at 12 months. The first group includes all those persons who are eligible to participate in the subsidy programme soon after registering with the SEA (they are required to be unemployed for 6 months only). This includes unemployed persons who are eligible for participation without any requirement related to the duration of their unemployment spell, e.g. persons with disabilities, persons aged 55 and over, refugees and those with alternative status and certain categories of youth. Moreover, those selected for participation after only 6 months from registration are those treated as a priority by the SEA caseworkers. 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. Participants are compared with other unemployed persons who have been in unemployment for at least 6 months and at least 12 months respectively (for whom the “clock is set to start” at the same time). The two groups (treated and control) are in theory eligible for participation, but some persons are selected for participation whereas others are not.
The treated group includes persons who received the employment subsidy within 6 months of the moment they became eligible (when the clock starts), while individuals who began employment subsidies after that time are dropped from the analysis. Initially, analyses were also performed defining the treated group as those persons who were treated at any time after becoming eligible for employment subsidies, but interrupted unemployment spells and people coming back to participate later on in employment measures complicated the interpretation of the results. All remaining persons who spent either 6 months or 12 months in unemployment (and hence became eligible) are included in the control group.
In order to make the comparisons between the treatment and control groups more reliable, the econometric analysis controls for individual characteristics (age, gender, education, ethnicity, disabilities) household characteristics (any child in the household, urban residence) and location (dummies for regions) all measured at the start of the clock (i.e. when a person becomes eligible for support). Profiling outcomes (at the moment of registration) are also taken into account. All regressions also include month of registration dummy variables (to account for seasonal effects), as well as SEA branch fixed effects. The standard errors are clustered at the SEA branch level. Moreover, to account for other activities unemployed persons may be doing in the first months of their unemployment spell (6 or 12 months) and before they become eligible for participation in subsidised employment, the analysis controls for participation in formal training, non-formal training and other employment measures.
The main results presented in this chapter are estimated using Ordinary Least Squares (OLS) regressions, because – as discussed in Box 3.2 in Chapter 3 – this presents three key advantages over using matching or other more complicated econometric techniques. Firstly, it is easier to compare multiple treatment groups and look at interactions between different treatments using OLS, whereas matching models are better equipped to estimate differences between a single treatment and a single control group. Secondly, matching estimators may be susceptible to the Incidental Parameters Problem: matching is typically done using a propensity score, which itself relies on a probit or logit model to estimate individuals’ likelihood of being treated, yet it is helpful to include fixed effects (for example at the level of the SEA branch) to make the estimated treatment effects more reliable. Thirdly, using OLS substantially speeds up computation.
In order to further eliminate potential sources of bias, the OLS regressions are estimated using only observations that lie within the region of “common support”.11 Broadly, these are individuals that, given their observable characteristics, had at least some chance of being in either the treatment or the control group. To do this, the propensity score – a variable capturing individuals’ likelihood of participating in employment subsidies – was calculated by running a probit model including all of the regressors used in the main OLS regressions.12 It emerges that the main results are virtually unchanged whether the sample is restricted to the area of common support or not.
The main results persist if the same treatment effects are estimated using propensity score matching (the main results on employment can be found in Annex Figure 5.A.4) instead of OLS regression (the main results are presented in Figure 5.20), further suggesting that the findings are robust to tweaking the analytical approach. In particular, the results were re-estimated by taking the same propensity scores used to restrict the sample to the area of common support, and treatment and control observations were matched using the nearest neighbour technique. That the results remained largely unchanged suggests that assuming a linear relationship between the outcome variable and the control variables (as well as the employment subsidies treatment), as in the OLS regressions is a tenable model for these data. The similarity between the results emanating from OLS regressions and propensity score matching or other more complex techniques has also been seen in several other similar studies that evaluate the effects of ALMPs (see Box 3.2 in Chapter 3).
Nevertheless, it is important to recognise that OLS, propensity score matching, and other similar techniques can only estimate treatment effects conditional on observable differences between the treatment and control group. Unobservable differences – for example, in terms of motivation, latent ability, or idiosyncratic preferences – may still bias the results and this should be kept in mind when interpreting the results in this section.
Selecting outcome indicators
The main labour market outcome considered in this analysis is the likelihood of employment. Separate analyses are conducted excluding employment in subsidised jobs from the outcomes and comparisons are drawn between these results and those on the overall likelihood of employment. The objective of this programme is to help participants find a non-subsidised job after the end of the programme, either with the same or a different employer. However, for some groups with very low perceived productivity and limited chances to find a non-subsidised job, the time they spend in subsidised employment can be considered a positive outcome through its effects on income, skills development and social inclusion, especially for the harder to reach groups of unemployed.
Ideally, additional employment indicators should be used to estimate the effects of the subsidy. For instance, in the evaluation of a subsidy programme for older, full-time, low-wage workers introduced in Finland in 2006, Huttunen, Pirttilä and Uusitalo (2013[27]) find a programme effect on the intensive margin (hours worked) but no effect on the extensive margin (the likelihood of employment). Unfortunately, it is not possible to calculate with precision the hours worked in the administrative that have been collected for this Review.
In order to provide some analysis on the quality of the job found after the end of the programme, the analysis also considers earnings as an outcome indicator. Chapter 3 of this Review includes a discussion about the possible bias introduced by the issue of selection into employment when earnings indicators are examined as possible outcomes. Employment and earnings outcomes are estimated at 6, 12, 18, 24, 30, 36, 42 and 48 months after the unemployed become eligible for support (at 6 or 12 months, depending on the model estimated).
A positive and persistent effect is found on the likelihood of employment
This section presents the effect of programme participation on the probability of employment up to four years after an unemployed person becomes eligible for participation in the subsidised employment programme, which can correspond to up to 4.5 or 5 years since the unemployment start. These are fairly long-term effects of employment subsidies, in comparison with the existing literature on such programmes. Indeed, very few studies examine the long-term effects of employment subsidies in other countries and no such study exists in Latvia. In one example, Sianesi (2008[25]) estimates long-term effects of an employment programme in Sweden and finds a higher probability of employment for participants of about 40 percentage points just after the end of the programme, and 10 percentage points five years later. These effects are usually stronger for the LTU.
The results presented in Figure 5.20 suggest that the estimated effect of the programme is strong and positive no matter when persons become eligible for participation (at 6 or at 12 months). Participation in the programme increases the probability of employment by 43 percentage points 12 months after the clock starts (Figure 5.20, Panel A). The size of the estimated coefficient is large but plausible, as most programme participants are still in their subsidised job 12 months after they become eligible for participation, which implies at most 11 months after they started their participation in subsidised employment. The effect declines to reach 14 percentage points at 36 months and then increases again to 18 percentage points four years after the clock start (for the persons who have been unemployed for six months, (Figure 5.20, Panel A). The corresponding effects estimated with propensity score matching can be found in Annex Figure 5.A.4.
The effects are even larger for persons who have been unemployed for at least 12 months (Figure 5.20, Panel B) up to two years after they become eligible (at least in terms of the point estimates). A number of reasons may explain this finding. First, this group comprises the LTU for whom the maximum programme duration can last longer, going up to 24 months. Second, the LTU, are – in most cases – more difficult to place, and may thus benefit more from programme participation. Moreover, those persons who have stayed longer in unemployment before participating in the programme could have participated in other activation measures in between.
The size of the estimated effects (Figure 5.20) is close to those found in other papers in the related literature. The majority of papers estimating the effects of subsidies on labour market outcomes compare participants with other unemployed persons with similar characteristics using a propensity score matching model (Carling and Richardson, 2004[28]; Sianesi, 2008[25]; Jaenichen and Stephan, 2011[29]; Bernhard, Gartner and Stephan, 2008[30]; Neubäumer, 2010[31]). The estimated effect is large in most of these papers. For example, Sianesi (2008[25]) and Bernhard, Gartner and Stephan (2008[30]) find increases in the employment rate of 20-35 percentage points and 40 percentage points respectively.
The SEA monitors the outcomes of the employment subsidies programme using data from the State Revenue Service and its own system to follow participants’ labour market outcomes after their participation in the measure, but these statistics do not distinguish between employment in the employer offering the subsidised job and employment in a new workplace. This is an important distinction for evaluating the effectiveness of the programme. Figure 5.20 (Panel B) presents the results of the estimations excluding persons who are still on subsidised jobs at up to four years after they became eligible for participation (i.e. after the clock start). This is an attempt to assess the capacity of the programme to help the unemployed transition into non-subsidised employment.
Overall, the impact of the programme on employment is lower when subsidised jobs are excluded, at least up to 24 months after the clock start (the respective results from propensity score matching are reported in Annex Figure 5.A.4). This result is well justified by the typical duration of the subsidies which can go up to 24 months for some groups of unemployed. The difference between the two lines in the two panels of Figure 5.20 gets smaller at 30 months after the clock start.
For persons who were unemployed for at least six months, programme participation is associated with a ten percentage points (11 percentage points) higher probability to be employed in a non-subsidised job 24 (36) months after they became eligible, and climbs to 15 percentage points at four years (Figure 5.20, Panel A). For those who had been unemployed for at least 12 months, the effect is seven percentage points at 24 months and six to eight percentage points thereafter (Figure 5.20, Panel B).
The programme is less effective for some groups of unemployed
To ascertain whether the effects of employment subsidies differ for certain sub-groups of the population, the analysis described above can be repeated with the sample restricted to i) individuals with disabilities; ii) individuals aged 55 and above; and iii) young people (20-29 years). Individuals are classified into these three groups according to their disability status and age at moment they register with the SEA. Rather than anchoring the analysis at six or 12 months after the month of registration as in the previous sub-section, instead the treatment group now comprises those who receive employment subsidies within six months of registering while the control group comprises those who receive no employment subsidies in that period. This is because different sub-groups – and even different individuals within the sub-groups – become eligible at different times. For example, disabled people and those aged 55 or more are eligible for at least some types of employment subsidies immediately after registering with the SEA, while young people may need to wait six months in unemployment to become eligible unless they have low educational attainment, limited work experience, or refugee status (see Table 5.3). The controls used are the same as in the analysis described above and the same specification tests have been performed.
Note: The clock is set to start at 6 months of unemployment for Panel A and at 12 months for Panel B and reflects the moment at which different groups of unemployed become eligible for participation in the programme. Therefore, the analysis includes all persons who have been unemployed for at least 6 and 12 months respectively. The reported coefficients represent the effect of the programme on the probability of employment in percentage points. Treated persons are those who participate in the subsidy programme within 6 months from the time the clock starts. The dependent variable is a dummy variable equal to one if the person is employed and zero otherwise. Every point in the figures indicate a coefficient on programme participation from a linear probability model. They are derived from a separate regression which includes controls for age, gender, education, ethnicity, disability status, household characteristics (any child in the household, urban residence), and regional dummies for regions all measured at the start of the clock. Profiling outcomes are also taken into account. Month of registration dummy variables and SEA branch fixed effects are also included in the regressions. Controls are included for participation prior to the clock start in formal training, non-formal training and other employment measures. The standard errors are clustered at the SEA branch level. The analysis is restricted to the region of common support. Missing dots in the figures indicate coefficients which are not significant at the 5% level.
Source: Latvian Office of Citizenship and Migration Affairs, Latvian Social Insurance Agency, Latvian State Employment Agency and OECD estimates.
The results reported in Table 5.6 show strong positive effects for all three groups when all jobs (both subsidised and non-subsidised ones) are considered. The results phase out over time, reflecting precisely the fact that, in the first one to two years, many of these jobs are subsidised ones, which do not necessarily lead to non-subsidised employment. The decrease in the magnitude of the estimated coefficients kicks in earlier for youth than for persons with disabilities, reflecting again the longer duration of the subsidy for persons with disabilities than for many of the other beneficiaries.
When subsidised jobs are excluded from the analysis, the results are very different from those discussed in the previous paragraph. For persons with disabilities, the programme does not seem to have any significant effect on non-subsidised employment. In contrast, for youth, the estimated coefficients are initially smaller in size than the ones when all jobs are considered, but the results for all jobs and non-subsidised jobs become quite similar after 30 months. Young unemployed persons who participated in the programme have a seven (eight) percentage points higher probability of employment three (four) years after first registering with the SEA than similar unemployed youth who have not benefited from the programme. This suggests that there is a true effect of programme participation on non-subsidised employment for youth. For older persons, the results are quite mixed, but there is a positive and statistically significant effect three and four years after SEA registration.
A positive effect on earnings is found three years after an unemployed person becomes eligible for participation
The analysis of the impact of employment subsidies on earnings finds a positive effect both when the clock starts at 6 months and when the clock starts at 12 months. Programme participation is associated with a 23% earnings premium at 12 months, 6% at 24 months and 9% at 36 months in the for persons who become eligible for participation after six months in unemployment. For those who need to be unemployed for at least 12 months in order to become eligible for programme participation (i.e. the clock starts at 12 months), the estimated coefficients are higher: 28% at 12 months and 17% at 36 months. When only non-subsidised jobs are included in the analysis, programme participation does not seem to significantly affect earnings. There is only a positive and statistically significant result of 6% at 36 months for persons who had been unemployed for 6 months before the clock starts and 12% for those who had been unemployed for 12 months.
Assisting unemployed persons with disabilities
The results presented in the previous section suggest that the employment subsidies are less effective in boosting employment among persons with disabilities. There may be for a number of reasons behind this, including a relatively weak link between employment and social services and the need to increase the incentives for employers to hire persons with disabilities and support them to adapt to the workplace. Discriminatory practices may also explain why participation in subsidised employment does not lead to employment for persons with disabilities after the end of the programme.
This negative finding should, however, be considered more broadly. Although boosting work outcomes is the main objective for activation measures targeting the unemployed, additional outcomes might be considered for persons with disabilities. For example, social integration and well-being outcomes may be equally important, or even more important, despite being much more difficult to measure.
The temporary change in the maximum duration of the subsidy for persons with disabilities from 24 to 36 months in 2014 seemed to have an impact on the actual duration of subsidised employment for persons with disabilities. Although realised total duration never exceeded 24 months in total, there is evidence of an increase in the average duration, in particular for this group of unemployed persons in 2014 (Figure 5.21, Panel A). There was a simultaneous increase in subsidies’ average durations for other target groups as well but this remains in 2015, whereas the increase observed for persons with disabilities can only be observed in 2014.
The number of participants with disabilities around the date when the maximum duration changed is too small to conduct an econometric analysis that would allow examining in a rigorous way whether this change was associated with better outcomes for unemployed persons with disabilities. Instead, it is possible to draw a simple comparison of the labour market outcomes of programme participants with disabilities before and after 2014. Figure 5.21 (Panel B) presents the labour market outcomes of former subsidy beneficiaries with disabilities six months after the end of the subsidy. There is no significant difference in the likelihood of employment between those who started their participation in the programme in 2014 and those who started before or after. Although these results do not represent any causal link between programme duration and labour market outcomes, they still hint towards the fact that extending the programme duration may do little for the group of unemployed persons with disabilities.
In many countries, subsidies offered to persons with disabilities – unlike subsidies for all vulnerable groups – are paid for long time periods, can extended several times, and can even go up to covering the entire duration of a job. This can happen, for instance, when there are no improvements in the work capacity of beneficiaries. Latvia’s scheme is fairly flexible as it is available to a wide range of employers and is also fairly long in terms of maximum duration. However, renewal is not permitted and repeated participation is only allowed at least one year after the previous one.
Empirical evidence on the effectiveness of subsidies for persons with disabilities is rare. The few evaluations of the Flexjobs scheme in Denmark (Datta Gupta, Larsen and Thomsen, 2015[32]) and the wage subsidy scheme in Finland (Kangasharju, 2007[33]) find positive effects of the schemes (larger effects in the case of the Finnish scheme) but also evidence of deadweight loss (in the case of the Danish programme). Flexjobs concerns mainly part-time jobs and those where working conditions that can be adapted to the needs of the workers. Datta Gupta, Larsen and Thomsen (2015[32]) show that the scheme was effective among persons with less severe health conditions who were able to work. An evaluation of Sweden’s subsidised employment scheme for persons with disabilities conducted in the early 2000s finds a small positive effect on employment, but also evidence of displacement effects, casting doubts about the overall efficiency of the programme (Calmfors, Forslund and Hemström, 2002[34]).
Note: In Panel A, the categories of youth/ persons aged 55 years and over and persons with disabilities are not mutually exclusive as for example youth (or older persons) with disabilities are counted under both the youth (older persons) group and that of participants with disabilities. Panel B reports the labour market outcomes of former subsidy beneficiaries six months after the end of the employment subsidy. The year reported in the Figure is the year when the subsidised job started. The analysis separates out persons who are employed in a new job from those who continue to be employed on the job for which they received the subsidy.
Source: Latvian Office of Citizenship and Migration Affairs, Latvian Social Insurance Agency, Latvian State Employment Agency and OECD estimates.
Previous OECD work (OECD, 2010[35]) highlights the need for subsidies for persons with disabilities to be adaptable to changing circumstances on the side of the employer and the worker, notably to reflect changes in the work capacity of the persons with disabilities and the changing job tasks and requirements. In the case of Latvia, the subsidy is fixed at 50% of the wage paid (with a cap of 1.5 the minimum wage for persons with disabilities). As such, the subsidy does not depend on assessed remaining work capacity or job-specific capacity. Moreover, there is no reassessment of the need for the subsidy at regular intervals nor is the subsidy amount and duration dependent on the person’s work capacity or degree of disability as defined by the three disability groups according to the severity of their conditions. One option for consideration would be to differentiate the conditions of the subsidy (both its amount and duration) according to the person’s specific needs and in accordance with the type of work to be performed. In addition, changes in the subsidy could be considered to match changes in the work capacity of the worker as well as changes in the job content. For persons with severe disabilities for whom this subsidy is absolutely necessary for them to have a job, the subsidy could take the form of a more permanent work-support measure.
Over the past few years, Latvia has introduced a number of new measures to further support the integration of persons with disabilities into the labour market. For example, mentoring support for unemployed persons with disabilities was introduced in 2018. For the moment, mentors are contracted by the SEA with public procurement procedures. Going forward, NGOs could be involved at the local level, given their deep knowledge of the needs and barriers that persons with disabilities face, which makes them well placed to assist the SEA in this role. Prior to participation in the six-month mentoring programme, clients follow a 20-day training course on CV writing, preparation for job search, etc. Moreover, all persons in disability groups I and II13 (not only the unemployed ones) enjoy free use of public transport and some tax deductions.
Employers have a key role to play in hiring and retaining persons with disabilities and need to be trained to work with this group and be accompanied in this process. The SEA provides training and consultations to employers on a regular basis. A campaign to promote diversity among employers has been conducted by the Ministry of Welfare, but no further action or information campaign has been conducted by the SEA to promote the hiring of persons with disabilities among employers. It has been shown in other countries that small employers are likely to benefit more from wage subsidies for persons with disabilities, but such employers typically have limited information about the subsidies that are available to them.
The public works programme is useful for some groups of unemployed
Many OECD and developing countries have in place direct job creation schemes, which provide people with a source of income. Such schemes can be particularly useful during crisis periods where jobs are not available as they serve as emergency social safety nets. Moreover, they can help local authorities to maintain a decent level of public infrastructure (roads, schools, municipal buildings, etc.) in periods where public funds are limited. Nonetheless, critics suggest such schemes are imperfectly targeted with not-so-needy households and individuals participating in the scheme, as well as foregone employment opportunities.
Direct job creation programmes, which include public works, represent 10% of Latvia’s expenditure on ALMPs today, down from 34% in 2011. The “LVL 100” stipend programme (initially “Workplaces with stipend”, replaced by the “Temporary Public Works Programme” in 2012), Latvia’s main Public Work Programme (PWP), was introduced in September 2009 as a means to support household income and mitigate the effects of job and income losses, given that unemployment insurance and social assistance reached only a small share of persons who were affected by the economic crisis. Moreover, the short duration of unemployment insurance left many eligible unemployed persons without income support.
Latvia’s PWP mainly plays an activation and anti-poverty role, offering unemployed persons the possibility to acquire some work experience and earn a basic income (a monthly stipend of EUR 150 as well as social insurance contributions), while offering services to municipalities or non-profit organisations. Eligibility conditions include registration with the SEA for at least six months and non-receipt of unemployment benefits or old-age pensions. Persons who have been registered unemployed for less than six months but have been without employment in the past year are also eligible to participate in PWP. The tasks usually performed include services in schools and social care as well as infrastructure maintenance and repairing work. Using PWP to substitute existing employees is prohibited. Hires should be in newly created jobs or jobs that have been vacant for at least four months. During the programme, participants can devote two days a month to active job search under the guidance of the SEA or participate in short courses offered by the SEA.
The programme offers employment for up to four months (continuously or with interruptions) in a given year and participants cannot return before one whole year has elapsed.14 Between January 2012 and October 2017, the period covered by the administrative data available for this Review, close to two thirds of participants benefited from the programme more than once and more than one quarter of all the participants participated in the programme four times or more Figure 5.22. Given that this is an underestimation of the total numbers of participations (they could well expand beyond that if data were available past October 2017), it becomes clear that repeated participation in the PWP is a life strategy for some individuals and households, especially in Latvia’s remote areas.
Note: Refers to participations in the public works programme in the period between January 2012 and October 2017.
Source: Latvian State Employment Agency.
Latvia’s PWP has been examined and evaluated from different perspectives by a number of studies. The evaluation by Azam, Ferré and Ajwad (2013[36]) uses propensity score matching and finds that the programme has been successful in raising participating households’ incomes by 37% relative to similar households which did not participate in the PWP. Moreover, it may be well targeted to those persons most in need mainly because of the low amount of the benefit it offers and the work requirements that accompany the benefit. Furthermore, this evaluation finds only small effects in terms of forgone income. Public works plays the role of social protection during periods where jobs are scarce as was the case during the economic crisis and is often the case in remote areas, depending on seasonal labour demand.
This Chapter does not conduct an impact evaluation of the PWP on post-participation outcomes. Instead, Figure 5.23 reports the labour market outcomes of participants, one, three, 6 and 12 months after they left the programme. A number of interesting observations can be made. The share of former participants who find a job increases from 13% one month after the end of the programme to 41% one year later. This is mainly driven by the substantial share of participants who return to public works one year after their previous participation. Indeed, one in five former PWP participants return to the scheme one year later. Although it is unlikely that the programme creates employment beyond that offered under the scheme, it can have important effects on skill development and, in particular, on social inclusion for participants. The programme is used in a recurrent way by a number of unemployed or inactive persons as a safety net when there are limited alternative work opportunities and for this reason, it should be maintained. It is possible to quickly scale down (or up) such a scheme, depending on the prevailing economic conditions and on the prevalence of other ALMPs, as was done during and after the economic crisis.
Note: PWP: Public Works Programme. This figure reports the labour market outcomes of former participants in public works, 3, 6 and 12 months after the end of the employment subsidy. For employed persons, it distinguishes between those who find employment in public works and elsewhere.
Source: Latvian Social Insurance Agency, Latvian State Employment Agency and OECD estimates.
Using linked administrative data to evaluate the effectiveness 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.
Nonetheless, a number of limitations with the linked administrative data were identified in this Review. Some of them are driven by changes in the IT system used by the State Social Insurance Agency as well as Latvia’s SEA. Other limitations concern missing – or difficult to acquire – parts of the data, such as full employment histories and certain training voucher information prior to November 2015.
It is important to maintain a well-developed system of detailed and linkable administrative data in order to facilitate the regular monitoring and evaluation of the effectiveness of activation measures. This data collection can also serve to answer other policy relevant questions, which are well beyond the field of activation policies. It is of primary importance to ensure comparability or some kind of continuity over time in the data produced in the system. This is a precondition for the assessment of policy changes and their impact on the outcomes of their target groups.
Latvia’s efforts to build this rich data system requires further investment in human resources to build the necessary technical skills. This could be easily achieved in Latvia where investment in IT skills has been high. Lessons from other OECD countries (e.g. Estonia, the Netherlands, Norway, Germany, Flanders in Belgium, etc.) could be used to further boost Latvia’s capacity in this field.
References
[36] Azam, M., C. Ferré and M. Ajwad (2013), “Can public works programs mitigate the impact of crises in Europe? The case of Latvia”, IZA Journal of European Labor Studies, https://doi.org/10.1186/2193-9012-2-10.
[30] Bernhard, S., H. Gartner and G. Stephan (2008), “Wage Subsidies for Needy Job-Seekers and Their Effect on Individual Labour Market Outcomes after the German Reforms”, SSRN, https://doi.org/10.1111/j.0042-7092.2007.00700.x.
[15] Boockmann, B. (2015), “The effects of wage subsidies for older workers”, IZA World of Labor, https://doi.org/10.15185/izawol.189.
[23] Brown, A. (2015), “Can hiring subsidies benefit the unemployed?”, IZA World of Labor 2015, p. 160, https://doi.org/10.15185/izawol.163.
[24] Brown, A. and J. Koettl (2015), “Active labor market programs - employment gain or fiscal drain?”, IZA Journal of Labor Economics, https://doi.org/10.1186/s40172-015-0025-5.
[18] Caliendo, M., S. Künn and R. Schmidl (2011), “Fighting Youth Unemployment : The Effects of Active Labor Market Policies”, IZA Journal of Labor Policy.
[34] Calmfors, L., A. Forslund and M. Hemström (2002), “Does active labour market policy work? Lessons from the Swedish experiences”, IFAU - Institute for Labour Market Policy Evaluation.
[13] Card, D., J. Kluve and A. Weber (2018), “What works? A meta analysis of recent active labor market program evaluations”, Journal of the European Economic Association, https://doi.org/10.1093/jeea/jvx028.
[28] Carling, K. and K. Richardson (2004), “The relative efficiency of labor market programs: Swedish experience from the 1990s”, Labour Economics, Vol. 11/3, pp. 335-354, https://doi.org/10.1016/J.LABECO.2003.09.002.
[32] Datta Gupta, N., M. Larsen and L. Thomsen (2015), “Do wage subsidies for disabled workers reduce their non-employment? - evidence from the Danish Flexjob scheme”, IZA Journal of Labor Policy, https://doi.org/10.1186/s40173-015-0036-7.
[6] Employers’ Confederation of Latvia (2013), Work Conditions and Risks in Latvia, 2012–2013, http://www.lddk.lv/wp-content/uploads/2014/01/Working-conditions-and-risks-in-Latvia-2012-2013.pdf.
[12] European Union (2016), Council recommendation of 15 February 2016 on the integration of the long-term unemployed into the labour market.
[11] European Union (2013), Council Recommendation of 22 April 2013 on establishing a Youth Guarantee.
[7] Eurostat (2013), Self-reported work-related health problems and risk factors - key statistics, http://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=hsw_pb4&lang=en.
[10] Ferré, C., H. Immervoll and A. Sinnott (2013), Profiling of People with No or Limited Labor-market Attachment, World Bank, Washington, D.C., http://www.lm.gov.lv/upload/aktualitates/lv_profiling_270513.pdf.
[20] Fremigacci, F. and A. Terracol (2013), “Subsidized temporary jobs: lock-in and stepping stone effects”, Applied Economics, https://doi.org/10.1080/00036846.2013.797644.
[26] Gerfin, M., M. Lechner and H. Steiger (2005), “Does subsidised temporary employment get the unemployed back to work? An econometric analysis of two different schemes”, Labour Economics, https://doi.org/10.1016/j.labeco.2004.04.002.
[9] Gotcheva, B. (2013), Review of Key Design Parameters and Legislation for Social Assistance Programs in Latvia, in Latvia: Who is Unemployed, Inactive of Needy? Assessing Post-crisis Policy Options, World Bank, Washington, DC.
[27] Huttunen, K., J. Pirttilä and R. Uusitalo (2013), “The employment effects of low-wage subsidies”, Journal of Public Economics, https://doi.org/10.1016/j.jpubeco.2012.09.007.
[29] Jaenichen, U. and G. Stephan (2011), “The effectiveness of targeted wage subsidies for hard-to-place workers”, Applied Economics, https://doi.org/10.1080/00036840802600426.
[33] Kangasharju, A. (2007), “Do wage subsidies increase employment in subsidized firms?”, Economica, https://doi.org/10.1111/j.1468-0335.2006.00525.x.
[8] Karanikolos, M. et al. (2016), Effects of the global financial crisis on health in high-income OECD countries: A narrative review, https://doi.org/10.1177/0020731416637160.
[31] Neubäumer, R. (2010), “Can Training Programs or Rather Wage Subsidies Can Training Programs or Rather Wage Subsidies Bring the Unemployed Back to Work ? A Theoretical and Empirical Investigation for Germany”, IZA Discussion Papers Series.
[1] OECD (2016), OECD Reviews of Labour Market and Social Policies: Latvia 2016, OECD Reviews of Labour Market and Social Policies, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264250505-en.
[14] OECD (2015), Investing in Youth: Latvia, Investing in Youth, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264240407-en.
[22] OECD (2015), OECD Employment Outlook 2015, https://doi.org/10.1787/19991266.
[35] OECD (2010), Sickness, Disability and Work: Breaking the Barriers: A Synthesis of Findings across OECD Countries, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264088856-en.
[17] Roger, M. and P. Zamora (2011), “Hiring young, unskilled workers on subsidized open-ended contracts: A good integration programme?”, Oxford Review of Economic Policy, https://doi.org/10.1093/oxrep/grr016.
[25] Sianesi, B. (2008), “Differential effects of active labour market programs for the unemployed”, Labour Economics, https://doi.org/10.1016/j.labeco.2007.04.004.
[3] State Labour Inspection (2018), Annual Report for 2017, http://www.vdi.gov.lv/files/sdo_2017_en_gala.pdf.
[4] State Labour Inspection (2015), Annual Report for 2014, http://www.vdi.gov.lv/files/sdo_2014_en_gala_versija.pdf.
[5] State Labour Inspection (2010), Annual Report for 2010, http://www.vdi.gov.lv/files/sdo_2010_lv_en.pdf.
[2] The World Bank (2015), The Active Aging Challenge: For Longer Working Lives in Latvia.
[19] van Ours, J. (2004), “The locking-in effect of subsidized jobs”, Journal of Comparative Economics, https://doi.org/10.1016/j.jce.2003.10.002.
[16] Van Reenen, J. (2004), Active Labor Market Policies and the British New Deal for the Young Unemployed in Context, https://doi.org/10.3386/w9576.
[21] Wunsch, C. (2016), “How to minimize lock-in effects of programs for unemployed workers”, IZA World of Labor 2016, p. 288, https://doi.org/10.15185/izawol.288.
Database references
Latvia’s Municipal Information System Database.
OECD/Eurostat Labour Market Programme Database, https://doi.org/10.1787/data-00312-en.
Note: The sample includes beneficiaries of unemployment benefits aged 15-64, during the period January 2012 to October 2017.
Source: Latvian Social Insurance Agency and OECD estimates.
Note: Only entrants (i.e. new disability benefit recipients) are included in the analysis. The figure shows the number of persons whose first observed disability benefit started in the reported observation month. Repeated disability benefit spells are not included in this graph.
Source: Latvian Social Insurance Agency and OECD estimates.
Note: GMI: Guaranteed Minimum Income. This figure includes persons aged 15-64 and covers the period January 2016 to October 2017. Education is based on the SEA data and supplemented using SOPA data. Household size is derived from the SOPA data. Age and other personal characteristics indicated on this figure refer to January 2016.
Source: Latvian Office of Citizenship and Migration Affairs, Latvian Social Insurance Agency, Latvian State Employment Agency, Latvia’s Municipal Information System Database and OECD estimates.
Note: The clock is set to start at 6 months for Panel A and at 12 months for Panel B. The reported coefficients represent the effect of the programme on the probability of employment in percentage points. The analysis includes all persons who have been unemployed for at least 6 and 12 months respectively. Treated persons are those who participate in the subsidy programme within 6 months from the time the clock starts. The dependent variable is a dummy variable equal to one if the person is employed and zero otherwise. Every point in the figures indicate a coefficient on programme participation from a nearest neighbour matching estimator. Age, gender, education, ethnicity, disability status, household characteristics (any child in the household, urban residence), and regional dummies for regions all measured at the start of the clock are used for the matching. Profiling outcomes are also taken into account.
Source: Latvian Office of Citizenship and Migration Affairs, Latvian Social Insurance Agency, Latvian State Employment Agency and OECD estimates.
Notes
← 1. The information on entitled persons is not available to the OECD team.
← 2. In Latvia, after reaching retirement age a person with disabilities can either receive disability pension or old-age pension, and most persons with disabilities of retirement age switch to old-age pension. Still, given their disability status, they can be entitled for the transport compensation for persons with disabilities.
← 3. It should be noted that the amount of the disability pension does not change after reassessment.
← 4. From the 1st of July 2009 to December 31 2011, benefit entitlement required contributions for nine months out of the previous 12 months whereas before that it was 12 months within the previous 18 months.
← 5. Full unemployment benefits were paid for the entire period of nine months if the contribution history was at 20 years and above; if the contribution history was 10‐19 years, the full benefits were paid for 6 months; if the contribution history was less than ten years, full benefits were paid for four months only. For those months where the recipients were not entitled to full defined unemployment benefits, they received 45 LVL (EUR 64) per month.
← 6. These groups include recipients of disability pension, old-age pension or state social security benefit, women on maternity leave, one of a child's parents or other person during a child care period, one of a disabled child's parents, if the child does not receive appropriate care services, as well as persons from 15 years of age who are acquiring full-time education in basic education, general secondary or vocational secondary education institution or either full-time students in higher education institutions.
← 7. 22 months in the case of 2016-17.
← 8. The relevant information is provided on the SEA website: http://www.nva.gov.lv/docs/32_5b8e575f211834.07019741.doc.
← 9. This information corresponds to the participants for whom the start date of their employment subsidies is observed.
← 10. Only persons with one job are included in the calculation of these statistics.
← 11. Restricting the analysis to the area of common support ensures that the relationship that is estimated between the outcome variable and the control variables is not distorted by observations in the control group that were very unlikely to ever access employment subsidies and would therefore serve as poor comparators for observations in the treatment group. In other words, the extent to which the OLS regressions extrapolate the relationship between the outcome variable and the control variables from control observations to treatment observations (and, indeed, vice versa) is reduced.
← 12. Any treatment group observations that had a propensity score greater than the maximum propensity score for the control group or less than the minimum propensity score for the control group were classified as off common support and were dropped. Equally, any control group observations that had a propensity score greater than the maximum propensity score for the treatment group or less than the minimum propensity score for the treatment group were classified as off common support and were dropped.
← 13. Disability groups I and II include persons with severe disabilities, whereas disability group III refers to persons with milder disabilities.
← 14. Only exception is for persons who did their training in local governments and return to work there as PWP participants.