The State Employment Agency (SEA), the public employment service of Latvia, plays a crucial role in using active labour market policies (ALMPs) to connect people with jobs. This includes meeting the needs of jobseekers, as well as persons at risk of employment. This chapter assesses how the SEA’s services and measures reach people in need of ALMPs and provide them with appropriate support. The chapter does this by analysing rich administrative microdata including detailed information on ALMP participation, unemployment histories, social insurance receipt, and earnings histories.
Modernising Latvia’s Public Employment Service through Digitalisation
Connecting People with Jobs
Abstract
This chapter uses administrative data to assess how well targeted the Latvian State Employment Agency’s (SEA) services and measures are to different jobseekers. The chapter uses detailed administrative data to diagnose the different obstacles that individuals might face to labour market integration and then analyses whether the services and measures provided help address the specific barriers and are targeted towards individuals according to their needs.
The SEA uses a profiling tool to group jobseekers into different categories of risk based on their perceived likelihood of finding employment. These groups are determined via a count of the number of specific risk factors an individual has. This tool is used to support the SEA counsellors to provide appropriate guidance to their clients and to make effective ALMP referral decisions. The chapter analyses whether the different groups of individuals that are defined by the profiling tool are associated with participation in the different ALMPs designated for these groups. It then looks at how the different profiling tool groups are associated with specific barriers to employment identified in the chapter.
This chapter stands as a natural complement to the prior chapters, which assessed how well the overall IT infrastructure and the jobseeker profiling, job matching and other digital tools available to the SEA counsellors, support them to help jobseekers, people at risk of job loss and employers.
The analysis relies on datasets of registered unemployed people and jobseekers not in registered unemployment, their background characteristics and their interactions with the SEA. The data come from two sources, the register data held by the SEA for its clients, and information on earnings and receipt of social insurance benefits provided by the Latvian State Social Insurance agency (SSIA).
Section 5.2 presents the data that are used in the analysis. Details are provided on the type of ALMPs that are undertaken by jobseekers and on what the different programmes comprise. To aid exposition, ALMPs are categorised into broad support types, using the methodology utilised by the OECD and the European Commission when reporting on ALMPs (OECD, 2022[1]). In Section 5.3, data are analysed to determine how attached different jobseekers are to the labour market, to identify those that may be in greater need of support from ALMPs. A number of different obstacles of employment are identified, to determine the precise needs of individuals with weaker labour market attachment. In Section, 5.4, the SEA’s jobseeker profiling tool is assessed against both its pre‑determined ALMP selection for employability categories and against how these employability groups enter into ALMPs addressing specific needs. Finally, in Section 5.5 a quantitative assessment is conducted on the specific labour market obstacles identified, to understand the match between participation in ALMPs and barriers to employment.
The following sections outline the data that are used in the chapter and then provide some summary statistics to set the scene for the more detailed quantitative analysis conducted later in the chapter.
The chapter makes use of individual level administrative data provided by two government agencies, SEA and the SSIA. The use of linked administrative data for policy analysis in Latvia is well-established (see for example (OECD, 2019[2]; 2015[3]), enabling Latvia to benefit from the evidence‑building and the rich policy insights that such data linked analysis permits (OECD, 2020[4]). The continued use of their administrative data for policy analysis allows Latvia to progressively build evidence on its policies, so that they can refine and improve them to better service citizens.
The data in this chapter are linked using anonymised individual-level keys based on an individual’s personal code. In this manner it is possible to link individuals between different datasets within the SEA administrative data and between the SEA and the SSIA data. This latter linking allows the chapter to utilise information on individuals’ past trajectories in the labour market, to help provide more insight on determining needs of participants. The data are provided for all individuals that were registered with the SEA between 2017 and 2022 (all new and existing registrations between these dates). In addition to this, data stretch back into the past to allow the chapter to analyse how previous labour market histories impact upon the likelihood of starting an ALMP. SSIA data go back to 2015, whilst the chapter has information on the SEA unemployment registrations back to 2012.
Detailed individual-level data allow the analysis to thoroughly disaggregate individuals by their personal characteristics (Table 5.1). Data from the SEA provide information on unemployment histories, education, nationality, age, gender, disability, return from parental leave, self-reported language and IT skills, and whether individuals have previously been incarcerated. SSIA data provide information on receipt of various different social insurance benefits and on monthly earnings of all the SEA clients in the sample.
Note: ALMPs: Active labour market policies.
Source: OECD summary of Latvia’s State Employment Agency (SEA) and Latvian State Social Insurance agency (SSIA) data utilised in the chapter.
Data on the number of individuals starting ALMPs by month show a gradual decrease over time, both in absolute terms and as a proportion of all registered unemployed (Figure 5.1). Across the observation window in this chapter there are three relatively distinct periods of varying participation levels.
Relatively high participation marked the period from the start of 2017 until June 2018, where there was on average 16 700 individuals (or 23% of all registered unemployed) starting ALMPs every month. This period also benefited from relatively high participation from ESF-funded projects as part of the 2014‑20 programme of works.
In the period from July 2018 to February 2020, participation began to fall and only an average of 12 000 individuals started an ALMP each month (19% of registered unemployed).
In the post-COVID‑19 period, removing the months whereby confinement stopped participation in many programmes, fewer individuals participated in ALMPs. From July 2020 to December 2022, on average 7 900 individuals started an ALMP each month, or 11.6% of registered unemployed.
In this context, where participation has decreased over time, targeting ALMPs towards those that need them most can ensure that scarce resources are deployed most effectively. Increasing the funding for ALMPs, to reverse the recent decline in participation, is also important to ensure that as many individuals as possible can benefit from ALMP support (including those with both high and low predicted employability and to ensure jobseekers are equipped to combat the megatrends such as digitalisation, AI and the green transition).
Number of individuals participating in active labour market policies (ALMPs) by month
Note: Data here represent individuals and not ALMPs, so that an individual with more than one recorded ALMP in a month is recorded only once.
Source: OECD calculations based on Latvia’s State Employment Agency (SEA) administrative data.
Labour market services offered by the SEA are the most frequent ALMPs offered to jobseekers (Table 5.2). The data comprise just over a million ALMP participations in total (between 2017 and 2022) and, of these, labour market services account for around 70% of the total participations. This is a usual feature of public employment services, as these are short interventions offered to jobseekers (often several times) in order to better orientate themselves in their job search, with career guidance and advice on job applications. The mean duration for all interventions in the services category is less than one day (0.2 of a day).
Training is also common for the SEA clients and covers a range of different courses, designed to provide upskilling and reskilling to participants. Training programmes include various additional vocational development programmes and modular training courses with a relative focus on IT learning. These cover a range of topics, including computer science (15% of all courses), language courses and driving qualifications. Other IT courses are also offered (for example, on computer-aided design (CAD), web development and digital marketing). Vocational training comprises around one‑third of all training and include a large range of professionally orientated training. The largest five vocational courses are for small business organisation (11%), project management (10%), arc welding (8%), customer service operator (7%) and clerk (7%). So, it is evident to see that the range of training is diverse, can cover many different topics and therefore can potentially help individuals to find employment across many different sectors and jobs. It can change with labour market demands as well, for example such that recent demands for carer, cooks and pastry assistants, can be catered for with well-tailored short vocational courses.
Active labour market policies (ALMPs) sorted by frequency (in terms of participations) from largest to smallest
Note: Duration represents mean duration of ALMP in days. Measures with start and end dates on the same day will appear as 0.
Source: OECD analysis of Latvia’s State Employment Agency (SEA) data.
Outside of labour market services and training, direct job creation schemes and employment incentives are also relatively frequent, covering around 15% of total ALMP participations. Start-up incentives and sheltered employment are very infrequent and comprise only of one programme in each of these two categories.
The data show that the ALMPs offered to clients by the SEA are varied and can potentially cater to many different needs and barriers. The chapter will now proceed to analyse how well these different ALMPs are provided to individuals when mapped against their individual needs.
This section uses administrative data to identify groups that would be likely to benefit from ALMPs. It follows previous OECD work (OECD, 2021[5]) which categorises individuals based on their previous labour market attachment and the different types of potential barriers they might have to entering the labour market.
One major difference between the previous work and the work in this chapter is that here the focus is only on the registered unemployed, because no data are available on the wider group of non-registered unemployed and inactive individuals. This limits the ability of the chapter to provide information on this wider group of individuals.
To determine which individuals may need further assistance to help them engage in the labour market, administrative data are used to determine the extent of an individual’s labour market engagement in the 12 months prior to becoming unemployed. Labour market weakness is identified by grouping three types of individuals:
No earners – those with no earnings in the previous year;
Low earners – those earning less than 50% of the median earnings of all individuals for that year;
Working with interruptions – those with positive earnings in the year, but with at least three months of unemployment.
By contrast, individuals who do not share these features are defined as having good prior labour market attachment. Reviewing both individuals with good and weak labour market attachment in the year prior to unemployment allows the analysis to determine whether groups with worse labour market attachment also face more obstacles to labour market integration. Identifying the precise nature of these barriers then allows to determine whether individuals are receiving the support they need from ALMPs.
Around one‑third of registered jobseekers have a weak attachment to the labour market and could benefit from ALMPs to help connect them to jobs (Figure 5.1). Just under 20% of individuals had no earnings in the 12 months prior to registered unemployment and a slightly smaller proportion of individuals had low earnings. A smaller proportion (around 5%) worked but with interruptions in the year.
Note: Any labour market weakness represents the presence of any of the three individual light blue characteristics for an individual.
Source: OECD calculations based on Latvia’s State Employment Agency (SEA) and Latvian State Social Insurance agency (SSIA) administrative microdata.
The analysis groups different supply-side obstacles to the labour market using a similar methodology to previous OECD work on this topic (2021[5]) into broad thematic areas based around skills and qualifications, geographic challenges, family-related obstacles, health-related obstacles, integration challenges and motivational challenges. The groupings in this chapter are based around these same broad areas, but the precise indicators included in each group are slightly different and based on the availability of the administrative data available.
Skills and qualifications – this grouping looks at people’s capabilities, recognising that low skills may limit the extent to which people can work in high-paying jobs. Having fewer skills can also lead to fewer job opportunities in the labour market, as it becomes more difficult to satisfy requirements for job applications. This challenge is captured via three variables: 1) poor qualifications, defined as ISCED 2011 categories 0‑2; 2) no professional education, defined as no vocational or higher education degree; 3) not having Latvian reading and writing skills; 4) Not having any IT skills recorded with the SEA. The drawback of the current measures on language and IT skills is that they are self-assessed and therefore subjective and prone to individual error. The SEA is currently exploring how it might capture this information using more objective measures.
Geographic challenges – is measured by whether an individual lives in an area where job opportunities are particularly scarce. This is defined using two variables. The first looks at the SEA branches with the highest share of jobseekers with weak attachment to the labour market (so that the top 10% of branches are identified). The second looks at whether the number of registered vacancies per registered unemployed person in the local branches of the SEA is in the bottom top 10% of branches.
Family-related obstacles – look at whether a person is available to take on job search and employment. Care obligations may limit the number of hours a person is available for work. This is measured by two variables. One which captures those individuals that have recently returned from a period of parental leave and looks at receipt of family-related benefits in the previous year. A second variable looks at whether individuals have care responsibilities for a child or an adult with disabilities in the previous year. The drawback of these data is that they will not show caring relationships where the carer is not a direct relative of the carer, so may understate the number of potential carers in the data.
Health challenges – receipt of disability benefits (having Latvian disability status) is used to indicate whether an individual has a health condition. Having a health condition may make it more difficult for individuals to undertake certain jobs and may limit the amount of work someone might be able to do. This will present further challenges to labour market integration relative to someone in good health. Disability in this context is defined as long-term or permanent limitation of functioning which affects the mental or physical capacity, capacity, self-care and integration of a person. The limitation of this definition of disability is that it is not possible to determine how much this affects the ability of an individual to work, meaning for some individuals it could over-state the impact on their labour-supply, whilst for other individuals it could understate it. Similarly, it will not capture other individuals with health challenges which may impact work, but who do not have Latvian disability status.
Integration challenges – circumstances which may mean that individuals have additional challenges integrating into society and may contribute to further difficulties with labour market integration: 1) Long periods without work, defined as three consecutive years without any earnings from employment and without registered unemployment; 2) Individuals without Latvian nationality, who may have additional problems integrating into society (either due to language or other networking and integration challenges); 3) Individuals who have been released from prison.
Alongside categorising individual needs by these broad categories, the chapter also groups ALMPs by whether they are general in nature (for example, general job-search support given by counsellors), or whether they cater to needs specific to the groups of obstacles outlined above (for example, financial support for re‑location which helps individuals with geographic obstacles). By doing this, the chapter can consider how well targeted ALMPs are to individuals’ needs.
People with several obstacles to employment and weak labour market attachment will have an increased need for ALMPs to help them overcome difficulties finding employment and high-quality jobs. Table 5.3 displays the correlation between weak labour market attachment and different obstacles to employment, as defined in the previous sections. It also shows how frequently individuals face more than one different type of barrier to employment. Alongside those individuals with weaker labour market attachment, statistics are presented for those with good labour market attachment, so that the relative incidence of obstacles in each group can be contrasted between those with better and worse previous labour market attachment.
Note: Percentages are defined as the proportion of individuals in each group (e.g. good labour market attachment) that have each of the attributes in the rows. The attribute categories (e.g. Integration barrier) displays the proportion of individuals with at least one of any of the underlying attributes in that category.
Source: OECD calculations based on Latvia’s State Employment Agency (SEA) and Latvian State Social Insurance agency (SSIA) administrative microdata.
Almost all unemployed people have some kind of barrier to employment, but multiple barriers are more common for those with weak labour market attachment. For those with weak labour market attachment, almost one‑third (29%) face three or more different types of obstacles to employment, whilst four‑fifths (79%) face at least two different obstacles. For those with good labour market attachment, multiple obstacles are still common, but are less prevalent. Only 12% of the group with good labour market attachment have three or more different barriers and only half (54%) have two or more.
Table 5.3 also demonstrates that ALMPs that promote upskilling are important in the Latvian context. Despite some differences between those with good and weak labour market attachment, skills obstacles are common across all groups. Ninety percent of those with good labour market attachment face some kind of skill barrier, whilst this figure rises to 95% for the group with weak labour market attachment.
A lack of high levels of education is common across all groups, but is worse for the groups with weaker labour attachment. Around one‑third (34%) of individuals with good labour market attachment have professional level qualifications, whilst only one in five (22%) of those with weak labour market do. Individuals with weak labour market attachment are around twice as likely to have low levels of education (29%) as those with good labour market (16%).
When looking at softer skills, around three‑quarters (76%) of individuals with weak past labour market attachment do not report the presence of any IT skills with the SEA. This is an imprecise measure, because it relies on individuals reporting their different IT skills to the SEA, but it hints at a possible need to ensure upskilling of individuals to participate in upskilling programmes that can build these skills to enable participation in the modern, digital economy. Here too these obstacles are more acute for those with weak labour market attachment, relative to their peers who are well attached to the labour market (only 71% without IT skills recorded with the SEA). Better recording of the specific skills that jobseekers possess by counsellors and how these relate to vacancies in desired occupations would facilitate more nuanced and complete discussions in counselling sessions.
Finally, almost one‑third (29%) of individuals do not report presence of Latvian reading or writing skills, above an elementary level (A2 in the European Framework of Reference for Languages). Not having the requisite language skills in Latvian will present a significant obstacle to employment and training and will preclude any job that involves use of the Latvian language to communicate, limiting the opportunity for access to good-quality jobs with significant progression opportunities.
Living in an area with limited employment opportunities is another obstacle faced by people with potential ALMP needs. Those with weak labour market attachment are almost two‑thirds more likely than those with good labour market attachment to seek work at a branch of the SEA which is associated with high levels of labour market detachment or where there are few vacancies per jobseeker. This means that for individuals searching for work in these areas, employment opportunities may be harder to come by and there may be greater needs to travel for work. ALMPs that seek to surmount these issues, such as those that help with travel and relocation costs, or that look for ways to broaden job search, can help individuals to alleviate obstacles which are dependent on their geographical location.
One‑third of individuals with potential ALMP needs have family commitments that might impinge upon their ability to look for employment in labour market. Having family needs may limit both the number of hours of work sought, and the type of jobs that individuals are able to apply for to accommodate their schedules of family care. Similarly for those individuals that are returning to the labour market after a period of absence, due to family-related care obligations, more support may be necessary to ensure that individuals are able to fully engage with the labour market opportunities that are present. Of the group that had not worked in the past year, 36% had some kind of family-related obstacle to employment. Ensuring ALMPs that are cognisant of these issues and find ways to nurture and encourage individuals to overcome these obstacles are important in this context. This must be done in across institutions as well, to ensure that separate social services are joined-up and reinforce one another.
Bad health, which limits the ability of an individual to search for and take‑up work is much more commonly associated with those with weaker past labour market attachment. One in ten (11%) individuals in this group are recorded as having some kind of severe health problem, and the individuals with weak labour market attachment are over three times as likely to have a disability as the group with good labour market attachment (4%). This proportion is highest for those who previously worked with interruptions (19%), highlighting the need for ALMPs which can unlock the labour market potential of individuals and facilitate individuals to engage with work that can be flexible to accommodate diverse health needs.
Using the European Union Labour Force Survey, it is possible to study those individuals of working-age not currently in the labour force due to not looking for work. This can provide insight on what policies may help to bring individuals back into the labour market, and provide clues as to their barriers to accessing the labour market. It complements the analysis of the registered unemployed population, which can only provide insight on those currently looking for work.
In 2020, 22% of the Latvia working-age population (aged 15‑64) were inactive. This compares favourably to the OECD average of 29%. In Latvia, of the group that are inactive, 18% say they would like a job (4% of the total working-age population). This is lower than other EU countries, for whom 21% would like a job (around 6% of the working-age population).
The proportion of prime‑aged individuals (those aged between 30‑54) who are inactive but would like to get a job is 40%. Illness and family or caring responsibilities are a large determinant of this group not actively seeking work. Twenty-seven percent do not seek employment due to illness and a further 34% do not work because of caring (17%) or other family responsibilities (17%).
For older individuals (aged 55‑64) bad health plays an even greater role for the inactive, with 37% of individuals reporting this as a reason for not seeking employment.
For those aged 15‑29, the overwhelming reason for not seeking work is due to being in education or training (83% of individuals), although 8% report family and caring responsibilities as the reason.
Policies to alleviate family-related and health obstacles to employment could help to activate some currently inactive individuals. For example, further help with childcare or caring costs may help individuals to combine some element of caring with labour market participation. The domain of these policies is not likely to sit entirely within the Ministry of Welfare or the SEA. Here the SEA can have an important role to co‑ordinate with the relevant social services providing this support. It is also important for the SEA to be aware of these obstacles more generally, because if individuals are successfully brought into the labour market, it is likely that many of these barriers would still exist in some form or other, meaning that having ALMPs to further support them would be important to help continue their journey back to the labour market.
Source: OECD calculations based on the European Union Labour Force Survey (EU-LFS) data.
The SEA introduced a new profiling tool in 2019, to separate jobseekers into three groups depending on their job readiness and their perceived likelihood of finding employment. This tool replaced an existing tool that was overly complicated and seen not to be serving the needs of counsellors and jobseekers. More detail outlining of the current and previous profiling tools can be found in Chapter 4. The 2019 “rules-based” tool relies on counsellors collecting information on risk factors at an interview with the jobseeker. These factors are then counted to divide jobseekers into three groups, dependent on the number of risk factors an individual has.
Fifteen different risk factors are gathered by counsellors at the interview to categorise jobseekers into the three different groups. The factors cover attributes like age, disability, language, mobility for work, education, parental responsibility, work experience and skills. Some risk factors are counted with a factor of 2, to give them more weight in the grouping calculation. Those in the group with the highest chances of finding employment are expected to do so within 6 months, the medium risk group in between 6 and 12 months, and the lowest in over 12 months.
Based on an individuals’ risk group, the profiling tool should direct counsellors to refer individuals to different types of ALMPs:
ALMPs for all risk groups;
ALMPs for those with medium and low employability;
ALMPs for those in the low employability group.
These risk groups only provide recommendations for the type of ALMPs that individuals go into and counsellors have the discretion to deviate from recommendations where they feel it necessary.
ALMPs deemed suitable for all risk groups cover general career advice and guidance, more general and basic training (for example obtaining a driving license), support for regional mobility and support for setting up a business. ALMPs which are suitable only for those with medium and low risk introduce further professional training and skills development ALMPs. Those ALMPs reserved for the low employability group include public works programmes, ALMPs that cover motivation and social mentoring for the long-term unemployed with disabilities, counselling with a psychologist and subsidised employment programmes.
The profiling tool has a good association with the likelihood of individuals participating in ALMPs categorised for those with employability challenges. For ALMPs designed for those with low employability, there is a much higher likelihood of individuals from this group participating in them. Individuals in the low employability group are almost five times as likely as those from the high employability group to participate in this type of ALMPs (Table 5.4). They are also around 2.5 times more likely to participate in this kind of ALMPs than those from the medium employability group.
In addition to low employability individuals participating in ALMPs intended for the low employability group, fewer individuals from the good employability group participate in ALMPs for the low employability group, relative to those from the medium employability group. So there is a logically consistent relative participation likelihood of each of the groups. The low employability group have the highest likelihood of participating in ALMPs for the low employability group, followed by those with medium employability, followed by those with high employability.
Relative probability of entry into ALMPs targeting different employability groups
Note: Odds ratios reported, where any value greater than one implies higher probability and any value less than one lower probability. For example, 1.2 would indicate a 20% greater chance, 0.8 would indicate a 20% smaller chance. Results based on a logistic regression of ALMP entry types using risk profiling groups as explanatory variables Stars denote statistical significance‑ * significant at 5%, ** significant at 1%, ***significant at 0.1%. Individual employability groups and ALMP types as designated by the SEA profiling tool.
Source: OECD calculations based on Latvia’s State Employment Agency (SEA) and Latvian State Social Insurance agency (SSIA) administrative data.
High-employability people are more likely to participate in ALMPs that are meant to reach more those with medium and low employability, according to the client profiling. Those individuals from the medium employability risk group are 26% less likely than those with high employability to participate in ALMPs designated for those with medium or low employability. Those from the low employability group are almost two‑thirds less likely to participate (64%) than those with high employability. More efforts to ensure that participation from these two lower employability groups are needed, to redress the imbalance that results in more individuals from the high employability group undertaking these ALMPs. For example, counsellors can invest more time in explaining the benefits of these ALMPs to individuals in the low and medium employability groups to encourage participation from them. Similarly, ensuring there is sufficient ALMP provision for those designated with high employability, such as shorter specific programmes focused on skills gaps, may help to reduce demand for ALMPs intended for individuals with greater difficulties finding work.
Participation in ALMPs that are suitable for all individuals, is slightly more weighted towards those individuals with more risks to finding employment. Whilst the profiling tool guidance designates that these ALMPs are available for all types of individuals, the fact that those with more challenges to finding employment are associated with increased participation means that more support is being given to individuals that have may have potentially more difficulty in securing employment. Both the medium and low employability groups are around 20% more likely to participate in this set of ALMPs, relative to those with good employment prospects (although some caution should be attributed to the estimate for the low employability group, as the result is not statistically significant).
It is also possible to review the association of the profiling tool’s risk groups with ALMPs catering for the different types of employment obstacles identified in section 5.3.2. Here more could be done to differentiate the low and medium employability groups. When looking across ALMPs that serve general and geographic needs, there is not much differentiation between those in the low and medium employability groups (Table 5.5). Better differentiation would allow ALMPs to be better targeted towards those with the lowest chance of finding employment, so that prospective labour market disadvantage could be alleviated through greater support.
Association between profiling risk group and ALMPs addressing specific labour market needs
ALMPs: Active labour market policies.
Note: Odds ratios reported, where any value greater than one implies higher probability and any value less than one lower probability. For example, 1.2 would indicate a 20% greater chance, 0.8 would indicate a 20% smaller chance. Results based on a logistic regression of ALMP entry using risk profiling groups as explanatory variables. Individual employability groups as designated by the SEA profiling tool. ALMPs grouped by type of obstacle they primarily address, categorised using a qualitative description of programme content by the SEA and programme documentation. Stars denote statistical significance‑ * 5% confidence level, ** 1% confidence level, *** 0.1% confidence level.
Source: OECD calculations based on Latvia’s State Employment Agency (SEA) and Latvian State Social Insurance agency (SSIA) administrative data.
ALMPs designed to address skills gaps (for example training programmes provided by employers) need particular attention so that individuals with worse employability participate more. Here individuals with high employability are the most likely to participate in ALMPs addressing skill needs. Those with medium employability are also more likely than those with low employability to participate. The pattern of participation in these ALMPs reinforces existing labour market disparity between groups and those with the most need for further upskilling are the least likely to participate in these ALMPs. Redressing this imbalance will help to mitigate disparities in labour market outcomes and will improve the lot of those with the greater difficulty of labour market integration.
For ALMPs that address health and integration needs, there is a higher likelihood that those with the greater needs participate. Individuals in the low employability category are over five times more likely than those with good employment prospects to participate, and this likelihood is also higher than those with medium employability. People in the medium employability group are also more likely to participate than those with good labour market prospects. Similarly, individuals in the low employability group are the most likely to participate in ALMPs intended to address health barriers.
Data on profiling show that profiling information is missing for a significant minority of individuals and therefore these individuals do not have an associated employment risk grouping. After an initial roll-out of the profiling tool in 2019, the proportion of monthly cohorts of registered unemployed individuals with profiling information dropped and never fully recovered (Figure 5.3). The timing of the decline coincides with the onset of the COVID‑19 pandemic and the related health measures. All in-person visits were cancelled and re‑scheduled for a future date and communication was carried out using telephone and internet. But since the easing of restrictions and return to normality, the proportion of unemployed individuals with profiling scores has not recovered to its previous levels. By March 2022, all visits had been re‑instated in person, rather than using telephone appointments.
Proportion of individuals with profiling information, by month of unemployment registration
Source: OECD calculations based on Latvia’s State Employment Agency (SEA) and Latvian State Social Insurance agency (SSIA) administrative data.
This missing information for some jobseekers means that counsellors have to make decisions on ALMP referral without recourse to the same information for all. This may affect their provision of information to individuals on which ALMPs may be beneficial to them and also on the likelihood of any referral to different ALMPs.
The existing SEA profiling tool was introduced without a formal counterfactual evaluation in its implementation period, which makes an analysis of its impact problematic. A qualitative evaluation was completed after the initial pilot, using data collected from regional office employees, that facilitated the further development of the profiling tool content. This chapter examines whether individuals from different employability risk groups are more or less likely to enter ALMPs that have been designated for them (and that the profiling tool should direct them to). However, this is not the same thing as knowing whether the profiling tool is causing participation in these different ALMPs. It may simply be that counsellors and individuals were already making decisions into which ALMPs to enter, regardless of whether the profiling tool was utilised or not.
As Figure 5.3 demonstrated, the current profiling tool was simultaneously introduced nationwide, which limits the opportunity to look at individuals that are subject to profiling and those that are not in the different regions. It is only possible to look at individuals right before and right after the introduction of the profiling tool in 2019, to determine whether those that were subject to profiling enter into different types of ALMPs than those that were not. However, this analysis is not feasible due to the temporal factors that see ALMP participation reducing over time. It is also important to note that a previous profiling tool did exist, so that individuals who entered before the introduction of the new profiling tool were profiled using the old tool.
A falsification test that compares “before/after” cohorts in the same months, the year prior to the introduction of profiling (where there is no before and after and no individuals are profiled using the new tool), shows differences in the likelihood of entering different ALMPs (Table 5.6). This is likely related to that temporal decline in participation and means that any similar analysis of “before/after” cohorts around the time of the introduction of the profiling tool will also be subject to such confounding factors.
Impact of active labour market policy (ALMP) participation of June‑August cohorts compared to March-May cohorts
Note: Compares the relative likelihood of the June‑August registered unemployed cohort entering different types of ALMP, relative to the individuals that start in the months March-May. For example, a value of 1.2 would indicate that the June‑August cohort was 20% more likely than the March-May cohort to enter a particular type of ALMP, whereas a value of 0.8 would indicate it would be 20% less likely. Stars denote statistical significance‑ * 5% confidence level, ** 1% confidence level, *** 0.1% confidence level.
Source: OECD calculations based on Latvia’s State Employment Agency (SEA) and Latvian State Social Insurance agency (SSIA) administrative data.
It is not possible to use cross-sectional variation in profiling application to test its impact either. Figure 5.3 also shows that, after April 2020, profiling information is recorded for some but not all registered unemployed. If the selection of jobseekers who are profiled was random, one could examine the impact of profiling on ALMP referrals. Figure 5.4 shows that jobseekers without profiling information leave the unemployment register more quickly than any of the profiled groups, suggesting that there are non-random differences between groups. For example, those individuals with better recent labour market experience or better education might be better able to find a job quickly, therefore there may be more of these individuals in the group without profiling information. If the group without profiling information were similar to the groups with profiling information, the expectation would be that their expected length of time in registered unemployment was similar to a weighted average of those groups (for example, somewhere around the line for the medium risk group in Figure 5.4).
The lack of information on the date of the profiling scores makes it challenging to conduct further analysis, because it is unknown whether individuals are profiled right after they register with the SEA or some other point in time (in theory, all profiling should be completed within 31 days of registration with the SEA. But profiling is completed at the first meeting with the jobseeker, so it may occur after this point if the meeting is not held within that timeframe).
Looking forward, it will be important to have a pre‑defined evaluation plan when any new profiling tool is implemented in the future. This will make identification of its effect on ALMP participation much easier to determine. For example, by rolling out any tool over time, and randomly determining which counsellors or branches will implement the tool first, it would be possible to generate variation which would allow the effects of the tool on individuals’ ALMP participation to be identified. Having such information is essential in determining how the profiling tool influences the behaviour of individuals (counsellors and clients) in making their ALMP choices.
Proportion of cohorts left registered unemployment by risk profiling group, days of unemployment
Note: Plots Kaplan-Meier survival curves which are right censored for ongoing registered unemployment spells. This estimates the number of individuals left in registered unemployment at any point in time, but takes account of the fact that some unemployment spells are still ongoing.
Source: OECD calculations based on Latvia’s State Employment Agency (SEA) administrative data.
This section reviews how well the obstacles that have been identified as potentially hindering individuals’ progress in the labour market are matched against participation in different ALMPs that cater to these particular needs. It extends the previous section by looking further into the particular types of barriers and how these are associated with participation in different ALMPs. This analysis can provide more insight into how the profiling tool might be amended to incorporate different information to direct individuals to the correct ALMPs. For example, rather than simply count risk factors as is done now, different weights could be given to different risk factors depending on the strength of their association with obstacles to employment.
The analysis which looks at how participation is related to specific barriers is restricted to only those individuals identified as having greater need of assistance to find employment in the labour market (those with weak labour market attachment).
The package of ALMPs offered by the SEA means that individuals are most likely to participate in ALMPs that serve general needs (Figure 5.5). General ALMPs comprise almost two‑thirds (62%) of the different types of ALMPs that individuals receive, leaving one one‑third which are targeted towards specific obstacles that individuals might face. The targeted ALMPs are overwhelmingly comprised of ALMPs which are designed to ameliorate skills needs (this group comprises 76% of the targeted total). Increasing the proportion of ALMPs which target specific barriers would help to ensure that individuals are able to address the individual specific obstacles to employment that they face.
Number of individuals having undertaken different ALMPs, grouped by type of labour market need ALMP addresses
Note: Groups count whether an individual has ever received the specific active labour market policy (ALMP) type in the first nine months of their registered unemployment, such that multiple occurrences of the same type of ALMP would only be counted once.
Source: OECD calculations based on Latvia’s State Employment Agency (SEA) administrative data.
Individuals with skills needs are most likely to have their particular needs directly addressed by an ALMP (Figure 5.6). Despite skills barriers being the most prevalent across registered unemployed individuals, the large number of ALMPs addressing skills needs mean that individuals with these barriers are still more likely to have their needs addressed, relative to the other types of barriers face. By contrast, the absence of any ALMPs directly addressing family-related obstacles means that individuals with these barriers do not get them catered for at all. This may be due to institutional design, as the SEA is not responsible for providing care services, but in this instance it means that links with the appropriate institutions are essential to ensure policies between institutions are holistic and linked-up. Individuals with geographic, health or integration barriers have a similar likelihood of having their specific obstacles addressed via an ALMP. This likelihood is fairly low across all of these groups (for example 10% of individuals with a health obstacle participate in an ALMP addressing health needs, whilst the similar figure for geographic and integration needs is 8%). The same caveat on agencies joining together to design policies holistically also applies across other areas, such as health and integration.
Proportion of individuals participating in ALMPs grouped by labour market obstacle
Note: A specific active labour market policy (ALMP) is one where the primary obstacle addressed by the ALMP matches the type of obstacle that an individual has.
Source: OECD calculations based on Latvia’s State Employment Agency (SEA) and Latvian State Social Insurance agency (SSIA) administrative microdata.
Participation in ALMPs that address general labour market needs is higher overall for those individuals presenting the different labour market barriers identified. For example, this means that across the barriers identified as potentially limiting labour market integration, individuals having these barriers are more likely to participate in ALMPs. Individuals with long periods of inactivity are 21% more likely to participate in general ALMPs than those without such inactivity and ex-prisoners are 41% more likely to participate than individuals without previous prison sentences (Table 5.7).
However, there are specific obstacles that are not associated with greater participation in general ALMPs. For example, individuals with only primary level education, without Latvian writing or reading skills, or without Latvian nationality are less likely to participate in general ALMPs. Increasing the weight of these characteristics in the profiling tool could help counsellors give stronger recommendations and referrals to these individuals in ALMPs that could further help them to find employment.
Odds ratios of individual characteristics on active labour market policy (ALMP) participation
Note: Odds ratios reported, where any value greater than one implies higher probability and any value less than one lower probability. For example, 1.2 would indicate a 20% greater chance. Stars denote statistical significance‑ * 5% confidence level, ** 1% confidence level, *** 0.1% confidence level.
Source: OECD calculations based on Latvia’s State Employment Agency (SEA) and Latvian State Social Insurance agency (SSIA) administrative data.
Redefining the profiling tool so that it better maps individuals with specific obstacles to ALMPs that address these obstacles would be particularly beneficial to individuals with skills obstacles. Despite those with skills needs being most likely to have them addressed by an ALMP, still less than 20% of individuals with skills needs have these addressed by an ALMP targeting skills gaps.
The association between participation in ALMPs targeting skills and previous educational attainment runs in exactly the opposite direction. Those with the lowest education are the least likely to participate in skills-enhancing ALMPs and those with higher education levels are most likely to participate. For example, those individuals with primary education or less have only one‑third of the likelihood as those with tertiary level education to particate in ALMPs addressing skills, and even those with post-secondary education are 20% less likely to participate than those with tertiary education. This would mean that if each education group started with the same number of jobseekers, for every 10 tertiary-educated participants in skills-based ALMPs, there would be only 3‑4 primary-educated individuals and 8 post-secondary-educated individuals.
Individuals with primary level education or less are only one‑third as likely as those with tertiary education to participate in such ALMPs. A profiling tool that gave more weight to those with lower education would help to ensure that individuals with the least pre‑existing human capital were able to participate in policies which help them to gain the skills needed to gain good jobs with good prospects. Similarly, ensuring that programmes are available that cater for the specific skills needs of individuals is important so that counsellors have an appropriate set of ALMPs to refer individuals to.
Not having any computer skills recorded is also associated with a lower participation likelihood in ALMPs offering skills improvement. Offering more courses which enable individuals to build up basic and intermediate computer skills could be beneficial in unlocking jobs for individuals that they are currently out of the labour market because of limited computer skills. This is important in modern digital economies where the proliferation of jobs with some digital content means that more opportunities are available for those people with decent IT skills.
ALMPs that help with barriers to integration (such as long periods out of the labour market or having been in prison) less likely to have non-Latvians participate in them, relative to Latvians. There is also no positive impact of being an ex-prisoner on the likelihood of participation in ALMPs aiding integration, relative to other individuals.
ALMPs that help with integration needs, for example measures that are targeted to the long-term unemployed and temporary paid public works scheme, aim to provide individuals with additional support to overcome barriers that are more difficult to address via shorter interventions (such as general counselling support) or are more difficult to address through other ALMPs (like re‑training programmes). ALMP support has to accommodate the diverse and challenging needs of these individuals such to provide them with the vital guidance and support they need to re‑integrate into society and the labour market.
Non-Latvian nationals are 11% less likely than Latvians to participate in integration measures. The need to ensure good government policy to help with the integration of migrants is something which is common across all countries and in all government policy domains. ALMPs are an important part of the support for these general integration objectives. Access to information which appropriately signposts counsellors to the needs of this group will help to ensure that individuals who have moved to Latvia and are meaningfully participating in the labour market can access ALMPs that can unlock the barriers to employment that they may have.
Although the group of ex-prisoners is relatively small in number, they face a range of different barriers to employment. Encouraging the use of ALMPs targeting integration for this group would permit more individuals to benefit from their support.
Currently there is no good evidence to suggest that individuals from areas with weak labour market attachment are more likely to participate in ALMPs that try to overcome geographical barriers, via support for mobility to expand geographic job search horizons. Neither being in an area with few vacancies per jobseeker, nor being in one with many jobseekers with limited labour market attachment is associated with increased entry into ALMPs addressing geographical needs.
The current profiling tool only includes transport as a barrier in its risk categorisation and the signposting to different ALMPs. Revising the guidance and support given to counsellors, to take account of broader regional dynamics (such as the number of vacancies and strength of the local labour market) could help to ensure that regions in Latvia are not left behind and that every individual is able to fully participate in the Latvian economy, regardless of their location.
The likely of participation in an ALMP addressing a health need is 17 times as high for those individuals with an identified health barrier. This is also likely to be related to specific eligibility conditions for such ALMPs, that preclude participation unless specific conditions are met. For example, the Minnesota programme that addresses addiction problems that hinder labour market participation is conditional on first having a registered addiction diagnosis.
This chapter has shown that jobseekers in Latvia have a range of obstacles that prevent them from achieving their full potential in the labour market. These obstacles encompass a wide range of needs, across skills, health conditions, geographic challenges, family-related factors and integration barriers. Registered jobseekers with weaker past labour market attachment are also more likely to have these barriers, often several of those at the same time. A suitable suite of ALMPs available to address these needs is an important part of the SEA’s toolkit. Previous counterfactual impact evaluation of Latvia’s ALMPs has shown that they are successful in increasing the employment of participants, underlining the importance of ensuring they are sufficiently available to help jobseekers find work (OECD, 2019[2]).
The SEA makes use of a profiling tool, that groups individuals into three broad risk groups and recommends a range of different ALMPs for these groups. For those with low employability, the tool does a good job at directing individuals to participate into the ALMPs that are designated for them. However, those ALMPs designated for those with medium or low employment prospects are more frequently undertaken by those with high employment prospects. On top of this, the current ALMP participation patterns of individuals are not always that well aligned with the specific obstacles that these individuals face. Re‑calibrating any profiling tool, so that it more accurately maps individual needs to ALMPs that address these needs, will help counsellors to refer their clients to the ALMPs that address their specific needs well. This is particularly important for ALMPs addressing skills barriers, given they are the most prevalent specific ALMP offered, but lower educated individuals are less likely to participate. Similarly, care should be taken when designing eligibility conditions for ALMPs, so that they are available to individuals which have need for them.
Restructuring the basket of ALMPs offered, so that more ALMPs which address specific needs are undertaken, will also help to ensure that more individuals are given well-targeted support. For example, this could be to ensure that any new funding to expand the reach of ALMPs and to reverse the recent relative decline in participation is focussed on ALMPs which address specific barriers, rather than those which serve general needs.
Increasing co‑ordination with other agencies which provide other social services could help to make sure ALMPs are able to address specific needs. For example, better join-up between the SEA and agencies which provide services on family and caring responsibilities could help to ensure that any individual who is offered work, or has time commitments, as a result of ALMP participation has the appropriate wrap-around care directly available to enable their participation.
References
[1] OECD (2022), Coverage and classification of OECD data for public expenditure and participants in labour market programmes, https://web-archive.oecd.org/2022-09-16/120573-Coverage-and-classification-of-OECD-data.pdf.
[5] OECD (2021), Improving the Provision of Active Labour Market Policies in Estonia, Connecting People with Jobs, OECD Publishing, Paris, https://doi.org/10.1787/31f72c5b-en.
[4] OECD (2020), “Impact evaluation of labour market policies through the use of linked administrative data. Final report”, https://www.oecd.org/els/emp/Impact_evaluation_of_LMP.pdf (accessed on 5 November 2021).
[2] OECD (2019), Evaluating Latvia’s Active Labour Market Policies, OECD Publishing, Paris,, https://doi.org/10.1787/6037200a-en.
[3] OECD (2015), Investing in Youth: Latvia, Investing in Youth, OECD Publishing, Paris, https://doi.org/10.1787/9789264240407-en.