This chapter provides new insights into the identification and characterisation of economically inactive people in Poland. It starts by outlining the need to increase labour force participation in Poland. Using a novel machine learning approach, the chapter then identifies three key economically inactive groups that the Polish Public Employment Service can effectively reach, engage, and activate: people who receive disability benefits, individuals near the statutory retirement age who receive old-age benefits, and women without caregiving responsibilities for young children. The analysis explores the reasons for remaining inactive in each group and describes regional and local variations in the composition of the economically inactive population. Finally, the chapter assesses each group’s average willingness to work using three distinct measures: financial incentives, estimated through potential earnings in the labour market; a survey-based measure of willingness to work; and insights from focus group interviews with the different target groups.
Developing Public Employment Services for Economically Inactive People in Poland
Abstract
Identifying and characterising groups of economically inactive people in Poland
A shrinking pool of jobseekers has shifted Poland’s focus towards activating economically inactive individuals to help mitigate labour shortages and support long-term economic growth. In 2023, Poland’s unemployment rate for 15–64-year-old individuals stood at just 3%, well below the OECD average of 5%. The low number of jobseekers is gradually contributing to labour demand pressures, as employers face increasing difficulty in filling vacancies. Between 2013 and 2023, the job vacancy rate, defined as the proportion of vacant positions relative to the total number of occupied and unoccupied jobs, more than doubled, rising from 0.4% to 0.9%. While it remains significantly lower than the EU-27 average of 2.8% in 2023, Poland’s rapidly ageing population is likely to exacerbate labour shortages in the coming years. Some sectors, such as Information and Communication Technology (2.0%) and Construction (1.6%), are already experiencing vacancy rates well above the national average.
Poland’s economic inactivity rate is on par with both the OECD and EU-27 averages but varies significantly across regions and local areas. The economic inactivity rate measures the share of the working-age population that is neither employed nor unemployed, meaning they are not actively seeking work. In 2023, Poland’s economic inactivity rate stood at 25% of the working-age population, on a par with the EU-27 average and slightly below the OECD average of 26%. The difference between Poland’s region with the highest and lowest rates of economic inactivity for people aged 15-64 years-old stood at 14 percentage points, above the OECD average of 10 percentage points. At the county (powiat) level, and excluding students, geographic disparities are pronounced: in 2021, Wodzislawski and Rybnicki in Silesia recorded economic inactivity rates of 21%, while Wroclaw in Lower Silesia had the lowest rate at 7%.
Effective outreach and activation by Poland’s Public Employment Services (PES) require a clear understanding of the employment barriers faced by different groups within the economically inactive population. As a new target group for the PES, the economically inactive have not yet been systematically analysed to inform policy interventions. Economic inactivity may be voluntary, as in the case of students or early retirees, or involuntary, due to individual or structural constraints such as disability or caregiving responsibilities. These barriers are often linked to socio-economic characteristics and place of residence and vary considerably across sub-groups. Outreach and activation measures must address the specific barriers of different sub-groups, while ensuring that the identified target groups are sufficiently large for an efficient use of public resources.
A novel machine learning algorithm that segments the economically inactive population by their main employment barriers identifies three primary target groups: individuals receiving disability benefits, those near the statutory retirement age who receive old-age benefits, and women without caregiving responsibilities for young children. A decision tree model is used to segment the economically inactive population in Poland into distinct, policy-relevant subgroups based on socio-economic characteristics and employment barriers, using data from the EU-SILC survey. Three main target groups emerge:
Target group I: People receiving disability benefits. In 2022, approximately 770 000 individuals of working age in Poland received disability benefits. The labour force participation rate of 15–64-year-old people with a disability stands at just 34%, significantly below the EU-27 average of 55%. Around 94% of disability benefits recipients have a chronic illness or condition. Additional employment barriers such as low education levels and limited work experience further decrease their employment prospects. In 2021, 26% of disability benefits recipients had lower secondary education or less, compared to 6% of the general population. Around 29% of disability benefits recipients had no work experience at all.
Target group II: Economically inactive who receive old-age benefits. In 2022, there were over 310 000 inactive old-age benefit recipients of working age who do not receive disability benefits. Economic inactivity is also high among those aged above the statutory retirement age. In 2022, the labour force participation rate for men up to five years above the statutory retirement age was just 18%, well below the OECD average of 36%. For women the rate was 24%, compared to the OECD average of 47%. In Lesser Poland and Silesia, where early retirement is common among those employed in the mining industry, the share of inactive individuals in the total working-age population receiving old-age benefits was the highest at 4% in 2021, compared to just 1% in other macro-regions.
Target group III: Women without caregiving responsibilities towards children below the age of six. In 2022, there were around 1 000 000 economically inactive 16–59-year-old women without caregiving responsibilities for children under the age of six in Poland who were neither disability nor old-age benefits recipients. In 2023, around 69% of women aged 15-64 in Poland participated in the labour market, a rate lower than in 28 OECD countries and below the EU-27 average of 70%. On average, in 2021, 16% of 16–59-year-old women without childcare responsibilities for children under the age of six were willing to work. Both the willingness to work and income potential is significantly lower among the 87% who do not hold a university degree. In 2022, 58% of 16–59-year-old women without children under the age of 6 who did not hold a university degree and who had no previous work experience lived in rural areas, further limiting employment options.
Focus group interviews (FGIs) reveal that the willingness to work in all groups is high if the right conditions are put in place. Economically inactive FGI participants largely remain without employment against their will, except for those citing health issues or financial security after retirement. Many are willing to accept less favourable job conditions or upskill but express that Public Employment Services offer insufficient support. According to FGI participants, inactivity is often linked to a combination of perceived discrimination, health constraints, caregiving duties, and a lack of opportunities in the local labour market. Financial concerns drive many to seek work, especially retirees and people with disabilities.
The current lack of flexible work arrangements is likely to contribute to economic inactivity in all target groups. In 2023, part-time employment as a percentage of total employment in Poland was relatively low at 5%, compared to the OECD average of 15%. Similarly, only 5% of working-age employees work from home at least half of their working days, well below the EU-27 average of 9%.
Increasing labour force participation has become a priority across OECD countries for social, demographic, economic, and fiscal reasons. With historically low unemployment rates in many economies, the pool of available jobseekers has decreased, making it increasingly difficult for employers to hire the right workers. At the same time, ageing populations are reducing the number of people of working age, further intensifying the challenge. In the medium to long term, growing labour shortages risk constraining economic growth. Reducing economic inactivity can partially counter this development, as inactive individuals represent untapped potential for the labour market. Unlocking their potential not only helps mitigate labour shortages but also alleviates fiscal pressures, as low labour force participation increases reliance on public benefits. Many economically inactive individuals further belong to disadvantaged groups, such as people with disabilities or those with care responsibilities at home, and face systemic barriers to employment. Engaging these groups can therefore reduce inequalities and support broader social objectives.
In Poland, high economic growth rates and rising wages have led to a drop in economic inactivity over the past decade. Between 2013 and 2023, Poland's GDP per capita grew at an average annual rate of 3.7% in real terms, compared to 1.5% for the OECD as a whole.1 The rise in the productivity per capita level was partly a result of a rise in labour force participation itself. However, the opportunity to earn higher income also incentivised employment (OECD, 2021[1]). As a result, economic inactivity of the working-age population fell significantly from 35% in 2013 to 25% in 2023, compared to a much smaller decline from 29% to 26% on average across the OECD during the same period.
However, some population segments continue to face difficulties in joining the labour force, and geographic differences in economic opportunities persist. For example, in 2022, 66% of working-aged people with a disability were inactive in Poland, significantly above the average of 45% in the European Union (EU-27). Similarly, 33% of women of working age in the country were economically inactive, compared to 31% of women on average in the OECD. Economic inactivity rates also continue to vary significantly across Poland, with the eastern regions of Poland struggling the most for reasons related to a historic dependency on the shrinking agricultural sector and a lack of market access to its EU neighbours (OECD, 2021[1]). For instance, in 2023, Podkarpacia recorded an inactivity rate of 31%, compared to just 18% in the city of Warsaw.
Across the OECD, Public Employment Services (PES) have traditionally focused on registered unemployed individuals, but some countries are expanding their efforts to include the economically inactive. By 2023, around half of PES in Europe reported conducting outreach to inactive individuals not registered with employment services (Jakubowska et al., 2024[2]). These efforts mainly target inactive youth and involve collaboration with external partners and the promotion of coaching and training services. A smaller group of PES, including France and Ireland, adopt a broader approach that targets disadvantaged neighbourhoods, long-term inactive groups and social assistance recipients.
A recent major labour market reform has expanded the responsibilities of Poland’s PES to include contacting, engaging, and activating economically inactive individuals. Economically inactive people will be added to the existing PES client base, which currently includes registered unemployed persons and jobseekers, as outlined in the amended 2004 Law on the Employment Promotion and Labour Market Institutions (Chancellery of the Prime Minister, 2024[3]). Under the new 2025 Act on Labour Market and Employment Services, Regional Labour Offices (Wojewódzkie Urzędy Pracy - WUP) and Powiat Labour Offices (Powiatowe Urzędy Pracy - PUP) will be tasked with implementing measures to identify, reach, and activate professionally inactive individuals (Ministry of Family, Labour and Social Policy, 2024[4]). Among OECD countries, Poland’s reform represents one of the most significant changes in the provision of active labour market policies in recent years (OECD, 2025[5]).
To respond to this legislative change, the Polish PES must develop an understanding of the employment barriers faced by different groups within the economically inactive population. As a new target group for the PES, the economically inactive have not yet been systematically analysed for policy interventions. Economic inactivity may be voluntary, for example among students or early retirees, or involuntary, due to individual or structural constraints such as disability or caregiving responsibilities. The barriers faced by those who are involuntarily inactive are often linked to their socio-economic characteristics and place of residence and vary significantly across sub-groups. Identifying segments of the economically inactive population that share similar characteristics and barriers to labour market participation support policymakers in ensuring that new client groups are large enough to make policy interventions efficient. Target group definitions further enable the PES to prioritise outreach efforts and develop customised activation measures that address the distinct challenges faced by each group in the economically inactive population. Well-defined target groups also facilitate monitoring and evaluation of policies, allowing PES to pilot different policy instruments that can then be ranked by their effectiveness.
This chapter provides an overview of the Polish labour market and introduces a novel methodology for creating categories of economically inactive individuals and applies it to the Polish context. It starts by describing the labour market situation in Poland, highlighting the need to increase labour force participation. The chapter then details the mixed-methods approach used to define target groups within the economically inactive population. Next, it examines each target group in turn, providing a characterisation with a focus on their willingness to work, assessed through various metrics. Finally, the chapter analyses regional and local differences in the prevalence of these target groups, emphasising that counties (powiats) will need to prioritise specific groups based on local population characteristics.
Poland’s unemployment rate has declined significantly over the past decade and is now below the OECD average across all of its regions. Between 2013 and 2023, Poland's unemployment rate dropped from 10.7% to 2.9% in the 15-64 age group, significantly below the OECD average of 5% in 2023 (Figure 2.1, Panel A). This represents one of the largest declines in unemployment rates among OECD countries, with only Greece (-17 percentage points), Spain (-14pp), Portugal (-11pp), Ireland (-10pp), and the Slovak Republic (-8pp) experiencing an equal or greater drop over the period. Regional differences in the unemployment rate are small. In 2023, the unemployment rate ranged from a low of 1.8% in Warsaw to a high of 4.3% in Podkarpacia (Figure 2.1, Panel B). By 2023, the unemployment rate was therefore below the OECD average in all of Poland’s voivodeships.
Unemployment rate in the OECD in 2013 and 2023 (Panel A) and unemployment rate in Polish regions in 2023 (Panel B), percentage of 15–64year-old labour force
Note: The data correspond to the TL2 regions. The Lubusz region is omitted due to a lack of available data (Panel B).
Source: (Panel A) OECD Statistics, Infra-annual labour statistics, (Panel B) OECD Regional Statistics, Labour statistics – Regions.
Long-term unemployment in Poland has also decreased by more than half over the past decade, now showing little variation across Polish regions. In 2023, Poland’s long-term unemployment rate, defined as the share of the labour force that has been actively seeking for a job for more than 12 months, stood at 0.8% of the 15–64-year-old population in the labour force and was among the lowest in the EU-27, where the average was 5.6% (Figure 2.2, Panel A). The long-term unemployment rate has decreased by 3.6 percentage points since 2013 when it stood at 4.8%, compared to a 3.6 percentage point decrease in the EU-27 over the same period. Long-term unemployment varies only slightly across regions, with Podkarpacia reporting the highest rate at 1.6%, and Warsaw the lowest at 0.6% (Figure 2.2, Panel B).
Long-term unemployment in the EU-27 in 2013 and 2023 (Panel A) and long-term unemployment in the Polish regions, percentage of 15–64-year-old labour force
Note: The data in Panel B correspond to the TL2 regions. No data available for long-term unemployment in Lubusz, Opole region, Podlaskie, Pomerania, Swietokrzyskie, Warmian-Masuria and W. Pomerania. The long-term unemployment for Lesser Poland, Podkarpacia and Warsaw only available until 2022.The long-term unemployment is defined as unemployment lasting 12 months and more.
Source: OECD elaboration based on Eurostat table [lfst_r_lfu2ltu].
While Poland's low unemployment and long-term unemployment rates reflect the additional opportunities created by a fast-growing economy, the scarcity of jobseekers may hinder business expansion in the future. Low unemployment, including long-term unemployment, means fewer individuals are actively searching for work. In a rapidly growing economy, these labour supply indicators should be analysed alongside labour demand measures, such as the job vacancy rate, defined as the proportion of vacant positions relative to the total number of occupied and unoccupied jobs. The job vacancy rate provides insight into how much employers are struggling to fill positions — a combination of low unemployment and a rising job vacancy rate signals potential labour shortages.
The job vacancy rate in Poland has more than doubled over the past decade, with some variation across regions and sectors of economic activity. In 2023, the job vacancy rate stood at 0.9%, a significant increase from 0.4% in 2013 (Figure 2.3). Vacancy rates varied across voivodeships, with the highest in West Pomerania (1.2%) and the lowest in Lublin (0.4%). Among economic sectors, Information and Communication had the highest vacancy rate (2.0%), followed by Construction (1.6%), Other Service Activities (1.3%), and Transportation and Storage (1.3%). The lowest rates were recorded in Education (0.4%) and Financial and Insurance Activities (0.5%) (Statistics Poland, 2024[6]). While Poland’s job vacancy rate remains well below the EU-27 average of 2.8% in 2023, the growing share of unfilled positions suggests that employers are increasingly struggling to recruit suitable workers. With a limited pool of jobseekers, labour demand may continue to outpace supply, potentially constraining business growth and, ultimately, economic growth.
Job vacancy rate in the EU-27 in 2023 and 2013
Note: Job vacancies refer to newly created, unoccupied or about to become vacant paid posts, and the job vacancy rate is defined as the proportion of vacant positions relative to the total number of occupied and unoccupied jobs.
Source: OECD elaboration based on Eurostat table [jvs_a_rate_r2].
Increasing labour force participation can help mitigate rising labour shortages while advancing broader fiscal and social objectives. In a context of low unemployment, indicating limited labour supply, and rising job vacancies, reflecting strong labour demand, expanding the workforce can improve labour market functioning by tapping into the potential of economically inactive individuals. Unlocking this potential not only eases labour shortages but also reduces fiscal pressures, as low labour force participation increases reliance on public benefits. Many economically inactive individuals belong to disadvantaged groups, such as people with disabilities or those with caregiving responsibilities, who often face systemic barriers to employment. Engaging these groups can therefore help reduce inequalities and support broader social goals. In the long term, integrating individuals with low labour market attachment will become even more important in addressing labour shortages and fiscal pressures driven by population ageing. For example, Poland’s shrinking working-age population is projected to lead to a decline in the number of employed of more than 12% by 2035 (Kukołowicz, Leszczyński and Lubasiński, 2024[7]).2
Poland’s economic inactivity rate is on par with the OECD and the EU-27 average. The economic inactivity rate measures the share of individuals of working age who are neither employed nor unemployed, meaning they are not actively seeking work. Economic inactivity can be voluntary, as in the case of early retirees or students, or involuntary, due to individual and systemic constraints such as disabilities or caregiving responsibilities. In 2023, Poland’s economic inactivity rate stood at 25% of the working-age population, on a par with the EU-27 average of 25% and the OECD average of 26% (Figure 2.4, Panel A). This marks a significant improvement from a decade ago when Poland’s inactivity rate was 35%, significantly above the OECD average of 29%. However, despite this decline, a substantial share of Poland’s population remains economically inactive, representing a potential labour resource.
Poland’s regional disparities in economic inactivity are larger than in most other OECD countries, reflecting historical and geographical differences in the labour market. In 2023, the difference between Poland’s region with the highest and lowest rates of economic inactivity for people aged 15-64 years-old stood at 14 percentage points, above the OECD average of 10 percentage points. This TL2-level regional disparity was greater than in 25 of 33 OECD countries, including neighbouring countries such as the Slovak Republic (9 pp), Hungary (6 pp), and Czechia (5 pp).3 Economic inactivity was highest in Podkarpacia at 31%, while Warsaw had a much lower rate of 18% (Figure 2.4, Panel B). Poland’s regional disparities in economic inactivity are closely linked to differences in regional GDP, which can be attributed to the decline of agricultural activity and spatial differences in market access that affect the ability to attract foreign investment. The eastern regions of Poland, including Podkarpacia, Podlaskie, Lublin, and Swietokrzyskie, historically had larger agricultural sectors that have shrunk rapidly over the past decades, contributing to higher inactivity rates. The geographical position of these eastern regions has further placed them at a disadvantage in attracting foreign investment. As a result, the growth of foreign capital stock has lagged behind other parts of Poland, which have benefited from greater access to the EU market (OECD, 2021[1]).
Inactivity rate in the OECD in 2013 and 2023 (Panel A) and inactivity rate in Polish regions in 2013 and 2023 (Panel B) percentage of working age population, 15–64-year-olds
Note: The data in Panel B correspond to the TL2 regions.
Source: (Panel A) OECD Statistics (Panel B) OECD elaboration based on Eurostat table [lfst_r_lfsd2pwn].
The economic inactivity rate in Poland remains on a par with the EU-27 when students are excluded and the working age is adjusted to match Poland’s statutory retirement age. Students are typically inactive by choice as they pursue education that prepares them for entering the labour market at a later stage. In Poland, the share of 15–24-year-old young people in education was 70% in 2022, slightly above the EU-27 average of 66% (Eurostat, 2024[8]). Excluding students allows for a better focus on economically inactive individuals that may face involuntary barriers to employment. Similarly, Poland’s relatively low statutory retirement of 59 for women means that internationally comparable labour force participation statistics that cover 15–64- year-old men and women include some women who are above the statutory retirement age. Box 2.1 discusses Poland’s statutory retirement age in more detail. Figure 2.5 shows that, after adjusting the economic inactivity rate to exclude students and limiting the working age to 16 to 59 for women and 16 to 64 for men, Poland’s rate at 13% is significantly lower compared to the conventional economic inactivity rate shown in Panel A of Figure 2.4. However, even after these statistical adjustments, Poland’s economic inactivity rate remains in line with the EU-27 average (13%). Between 2013 and 2023, Poland recorded the second largest decline in the adjusted economic inactivity rate among EU-27 countries (-4.1 percentage points), equal to Portugal and only behind Hungary (–6.7 pp), and well above the EU-27 average reduction (–2.2 pp).
The statutory retirement age in Poland is 65 for men and 60 for women. The statutory retirement age in Poland was set in the early 1990s and remained unchanged until 2012, when a law was passed to gradually equalise it for both genders. However, this decision was reversed in 2017, reinstating the original 60/65 age limits (OECD, 2021[1]). Only a few other OECD countries, including Colombia, Costa Rica, Hungary, Israel, and Turkey, still have different statutory retirement ages for men and women.
Poland’s statutory retirement age is low for women compared to the OECD and its regional peers, and no plans to raise it exist. The statutory retirement age for women of 60 years in Poland is below the OECD average of 63.6 years. Poland current statutory retirement age for women is also low compared to Central and Eastern European OECD member states (see Table 2.1). Looking ahead, many of these countries plan to raise their statutory retirement ages. Notably, Estonia plans to increase its statutory retirement age from 64.3 to 71 for both genders, while the Slovak Republic aims to raise it from 62.8 to 69. In contrast, Poland has not announced any plans to increase its statutory retirement age, further widening the gap with other countries in the future.
Previous research indicates that the statutory retirement age is one of the key factors influencing retirement decisions in Poland. While workers are not legally required to stop working upon reaching the statutory retirement age, studies show that it significantly affects labour supply among older individuals (Ruzik-Sierdzińska, 2018[9]). Additionally, most individuals claim their old-age pensions either at the statutory retirement age or within two years after reaching it (Chłoń-Domińczak, 2019[10]). In fact, the 2017 reversal of the retirement age law in Poland led to an immediate decrease in the retirement age for women (OECD, 2021[1]).
Note: The current statutory retirement age refers to a person with a full career from age 22 who retired in 2022. Future statutory retirement age describes a person with a full career from age 22 who entered the labour market in 2022.
Source: OECD Statistics (2024[11]), Pensions at glance, OECD (2021[1]) Regional Economic Inactivity Trends in Poland, Chłoń-Domińczak (2019[10]), Impact of retirement age changes on the old-age pension take up in Poland after 1990, Ruzik-Sierdzińska, A (2018[9]) “An Attempt to Identify Factors Influencing Retirement Decisions in Poland.”
Economic inactivity rate (excluding students) of 16–59-year-old women and 16–64-year-old men, 2013 and 2023
Note: The age categories reflect the statutory retirement age in Poland, where women retire at 60 years old and men at 65 years old.
Source: OECD elaboration based on EU LFS 2013 and 2023 data. Data missing for Germany and Norway in 2013.
Economic inactivity rates not only vary between Polish regions (voivodeships), but also between counties (powiats) within regions. The highest shares of economically inactive were recorded in rural powiats in northern and south-eastern parts of the country (Figure 2.6). Excluding students, powiats with the lowest inactivity rates among the working population in 2021 were recorded in major urban areas, including Wroclaw (7.5%), Warsaw Metropolitan Area (7.8%), Poznan (8.1%), Opole (8.2%), and Krakow (8.5%). In contrast, the ten powiats with the highest inactivity rates were concentrated in regions like Silesia, West Pomerania, Lublin, Kuyavian-Pomerania, and Warmian-Masuria. These include Wodzislawski (20.7%), Rybnicki (20.6%), Swidwinski (20.2%), Wlodawski (20.1%), Lipnowski (19.9%), Bartoszycki (19.5%), Ketrzynski (19.2%), Jastrzebie-Zdroj (19.2%), Chelmski (18.9%) and Choszczenski (18.9%).
Economic inactivity rate (excluding students) across Polish powiats among 18-59/64-year-olds
Note: Data analysed at the powiat level. Borders shown on the regional (voivodeship) level, distinguishing the Warsaw capital region.
Source: OECD calculations based on Population Census Data, 2021.
One potential reason for the persistence of economic inactivity among some population segments in Poland is the lack of flexible work arrangements. Flexible opportunities to work, including working part-time, working from home, and working flexible hours enable some people to participate in the labour market who would not be able to so on a full-time contract. In Poland, only a small share of women has the flexibility to determine their workday start and finish times, which would help them balance work and childcare obligations (Magda, 2020[12]). Similarly, research suggests that flexible working conditions are an important motivator for older workers wishing to remain employed beyond retirement age (Sewdas et al., 2017[13]). Opportunities for older workers to reduce their hours before full retirement and engage in part-time employment can encourage those still willing to work but seeking more family time, or those unable to commit to full-time employment due to health reasons, to delay full retirement.
Poland has a low-part time employment rate, a potential contributor to economic inactivity in particular among women. In Poland, in 2023, part-time employment, defined as working fewer than 30 hours per week in one’s main job and measured as a percentage of total employment in the 15-64 age group, is relatively low at 4%, compared to the OECD average of 15% (Figure 2.7). The Polish part-time employment rate is comparable to other Central and Eastern European countries such as Slovenia (6%), Latvia (5%), Czechia (5%), Lithuania (4%), Hungary (3%), and the Slovak Republic (3%), a region characterised by lower part-time employment rates than in other OECD countries. In contrast, the Netherlands stands out as a country where more than one in three people (34%) work part-time. Poland’s gap in part-time work is particularly pronounced among women. In 2023, only 7% of those aged 15-64 were in part-time employment (men: 2%), compared to the average of 23% in the OECD (men: 8%).
Part-time employment in the OECD as a percentage of total employment of the 15–64-year-olds, in 2023 by gender
Note: Part-time employment is defined as working less than 30 hours per week in one’s main job (whether as an employee or in self-employment).
Source: OECD Statistics, Full-time and part-time employment.
Work from home arrangements can also play a role in reducing economic inactivity. For instance, remote work is beneficial for individuals facing mobility restrictions. Past research has shown that remote work boosts job satisfaction and reduces absenteeism for workers with a disability and caregivers across Europe (Giovanis and Ozdamar, 2019[14]). It also gives people with disabilities more control over when and how they work, improving health management (Taylor et al., 2022[15]). The COVID-19 pandemic has increased employer acceptance of such arrangements in some countries such as the United States, although, the potential remains constrained by occupational distribution (Ameri et al., 2022[16]). Older employees across the OECD are also more likely to work remotely, often seeking more flexible work options as they approach retirement (Özgüzel, Luca and Wei, 2023[17]).
A low share of workers in Poland benefits from teleworking opportunities compared to the European average. Remote work, defined as the share of 15–64-year-old individuals working from home at least half of the days worked, is still less common in Poland than in many other European countries, with only 5% of employed individuals working from home, well below the EU-27 average of 9% (Figure 2.8). Poland’s teleworking rate is comparable to other Central and Eastern European countries, such as Slovenia (6%), Czechia (6%), Lithuania (6%), Slovakia (5%), and Hungary (4%) and significantly below Estonia’s (12%). The gap between Poland’s work from home rate and the EU-27 average has widened over the past decade, from 1 percentage point in 2013 to 4 percentage points in 2023. Poland’s teleworking rate thus remains low despite the COVID-19 pandemic that temporarily encouraged many firms in Poland to adopt remote work options (Radziukiewicz, 2021[18]). While cross-country differences in remote work adoption can partially be attributed to cross-country variations in industrial structures and rural-urban differences (see Adams-Prassl et al. (2022[19]) and Özgüzel, Luca and Wei (2023[17])), the lack of remote work options poses an employment barrier for individuals who cannot work in traditional office settings, such as people with disabilities and caregivers.
Work from home employment in the EU-27 in 2023 and 2013, percentage of total employment of the 15–64-year-olds
Note: Working from home is defined as having worked from home at least half of the days worked in a reference period of four weeks.
Source: OECD elaboration based on Eurostat table [lfsa_ehomp].
Defining target groups within the economically inactive population for policy interventions involves a trade-off between tailoring policies to individuals and ensuring the efficient use of public resources. In theory, PES aiming to activate individuals distant from the labour market often encounter highly idiosyncratic situations, where each person faces a unique combination of employment barriers. In the most individualised approach, each person’s circumstances would require a bespoke policy tool. Conversely, a cost-efficient approach might favour a one-size-fits-all solution. Balancing this trade-off requires defining broad categories of economically inactive people based on shared characteristics and barriers, allowing for tailored and resource-efficient policy responses.
A mixed methods approach is used to categorise groups of economically inactive in Poland and to understand the reasons behind their inactivity. The mixed methods approach first deploys a novel machine learning method to partition Poland’s economically inactive population into groups that can be targeted by the Polish PES. Following the quantitative exercise, focus group interviews were conducted in Poland to deepen the understanding of barriers faced by the different groups of economically inactive people, their motivation to work, and the policy tools that economically inactive people consider particularly useful to overcome employment barriers.
In a first step, a decision tree approach is applied to identify target groups among the economically inactive population. The decision tree is a simple machine learning method that systematically divides the economically inactive population into subgroups based on shared socio-economic characteristics and employment barriers, ensuring the categories are both distinct and actionable for policymakers. Using data from the European Union Statistics on Income and Living Conditions Survey (EU-SILC) for Poland, the algorithm identifies the most significant factors influencing inactivity and organises them into easily interpretable subgroups. The process balances statistical rigor with policy relevance by limiting the number of subgroups to ten. This ensures practicality for Public Employment Services while maintaining meaningful differentiation between groups. Box 2.2 provides an intuitive explanation of the decision tree approach, while Annex 2.A provides a technical explanation of the methodology.
Compared to simple cross-tabulations or more complex statistical techniques, a decision tree approach offers an efficient and systematic way to identify key subgroups among the economically inactive population that remains easy to interpret. Rather than manually testing different combinations of socio-economic characteristics and employment barriers, the algorithm can process hundreds of potential predictors simultaneously, such as age, education, benefit status or household structure, and automatically selects and ranks the most relevant variables based on their explanatory power. This allows for an interpretable segmentation of the population into distinct groups, ordered by the strength of the predictors, that can be easily reproduced. In contrast, more complex methods such as latent class analysis may yield groupings that are more statistically precise, but often produce clusters that are difficult to interpret or use in policy.
In a second step, additional insights and geographically detailed results are obtained by mapping the identified target groups onto other data sources. While EU-SILC data provides detailed information on income, socio-economic characteristics, and household composition, it lacks some variables relevant to the analysis, such as the self-reported willingness to work and reasons for economic inactivity. These variables are sourced by linking the identified groups to Poland’s Labour Force Survey (LFS). Similarly, since EU-SILC data is only available at the NUTS-1 level, which corresponds to seven macroregions, additional geographic detail at the powiat (district) level, relevant for labour market policies carried out by Polish Powiat Labour Offices, is derived from Population Census data. Due to differences in variables across the datasets, some approximations were necessary. For instance, one important group identified through the decision tree using EU-SILC data includes individuals receiving disability benefits. Since neither the LFS nor the Population Census directly captures information on benefits, individuals self-reporting economic inactivity due to health problems were used as a proxy for this group.
In a third step, focus group interviews (FGIs) were carried out with the main identified target groups to understand their barriers to employment, the interaction of employment-preventing barriers and distinct support needs. The groups included pre-retirement individuals, those above statutory retirement age, people with disabilities, and women with childcare responsibilities for children over the age of six. Participants represented diverse socio-demographic characteristics, with a balance between urban and rural residents. The interviews complement the quantitative analyses by exploring participants' employment histories, reasons for inactivity, willingness to work, and their support needs for reintegration into the labour market. Particular emphasis was placed on the interaction between different employment barriers and their connection with participants’ socio-economic characteristics. All FGIs were designed as informal conversations and tailored to each of the groups. Box 2.3 provides more details on the FGIs. Annex 2.B describes the participants and the methodology.
Decision trees offer a data-driven method for defining target groups among the economically inactive population that is easy to interpret for policymakers. The approach works by iteratively splitting the data based on observable characteristics, selecting at each step the feature that best separates active and inactive individuals. In the context of categorising economically inactive people, these features reflect socio-economic characteristics or predefined employment barriers. The splitting process continues until the data is fully partitioned or a stopping criterion is applied to prevent overfitting (Myles et al., 2004[20]). The resulting groups are as distinct as possible in terms of inactivity, enabling policymakers to identify shared characteristics and barriers among target groups. These characteristics can then be used to define the groups in an intuitive way. Moreover, decision trees naturally select the largest subgroups within the economically inactive population, and thus meeting the efficiency requirement of employment activation policies. Figure 2.9 illustrates the procedure.
Note: The split variables are either socio-economic characteristics or a priori defined employment barriers. The illustration is an example of a “perfect” split; in real world applications, all split variables are likely to produce impure subgroups such that an algorithm is needed to produce partitions that are as pure as possible.
The primary dataset used for the analysis is the European Union Statistics on Income and Living Conditions Survey (EU-SILC) for Poland. The working sample covers all active and inactive individuals in Poland aged 16-64 years for males and 16-59 years for females, which corresponds to the working age for these groups in the country. To ensure the dataset only reflects the economically active population of interest to the PES, students have been excluded, as their investment in education is expected to enhance their future employability. The socio-economic variables and employment barriers identified used as inputs in the decision tree are shown in Table 2.2 and described in more detail in Annex 2.A.
Source: Myles, Anthony J.; Feudale, Robert N.; Liu, Yang; Woody, Nathaniel A.; Brown, Steven (2004[20]). An introduction to decision tree modeling.
To complement the quantitative grouping of economically inactive individuals, focus group interviews (FGIs) were conducted for each identified target group, with a total of 35 economically inactive individuals participating. These groups included:
individuals approaching statutory retirement age in Poland, i.e. women aged 50 to 59 and men aged 55 to 64 years;
people above the statutory retirement age, >59 years for women and >64 years for men;
people with a disability;
women of working age with childcare responsibilities for a child aged above six years.
Participants varied in socio-demographic characteristics such as gender, education, and professional experience, as well as their place of residence, for which a balance between participants from urban and rural areas was achieved. The interviews were held in the four Polish cities of Gdańsk, Kraków, Olsztyn, and Opole. All participants were recruited through collaboration between Regional Labour Offices and local social economy organisations. Participation was voluntary, and all respondents completed the interviews which lasted about two hours each. The interviews were structured around tailored scripts designed to reflect each group’s distinct circumstances and objectives. Sessions were informal to encourage open conversations.
The FGIs explored participants’ past experience in seeking employment, the reasons for their current economic inactivity, and their willingness to work in the future. Discussions also addressed the subjective support needs from employers, Public Employment Services, and other institutions to facilitate their labour market reintegration.
Categories of economically inactive people
This decision tree approach yields ten target groups of the economically inactive people, based on joint socio-economic characteristics and employment barriers:
1. Recipients of disability benefits;
2. Recipients of old-age benefits, who don’t receive disability benefits;
3. Women without kids under 6, older than 26, without university education or experience, who don’t receive disability benefits or old-age benefits;
4. Women with kids under 6 that do not receive childcare, without university education, who don’t receive disability benefits or old-age benefits;
5. Women with kids under 6 that do not receive childcare, with university education, who don’t receive disability benefits or old-age benefits;
6. Women without kids under 6, older than 26, without university education, with some experience, who don’t receive disability benefits or old-age benefits;
7. Women with kids under 6 that receive childcare, who don’t receive disability benefits or old-age benefits;
8. Women without kids under 6, with university education, who don’t receive disability benefits or old-age benefits;
9. Males not included in other groups, who don’t receive disability benefits or old-age benefits;
10. Women without kids under 6 that receive childcare, without university education, 26 years old or younger, who don’t receive disability benefits or old-age benefits.
A distinct feature of the decision tree approach is its hierarchical structuring of groups based on predictive importance, such that all group names represent the primary reason for the groups’ economic inactivity. For example, recipients of disability benefits (1) represent the first split in the tree, as they are the strongest predictor of economic inactivity in Poland. This group may, by design, include individuals who also belong to other categories. The second split consists of old-age benefit recipients, a distinct group that does not include those in the first category. Subsequent splits, such as women, exclude both disability and old-age benefit recipients, and so on. While some overlap exists in reality, the hierarchical structure of the decision tree ensures that individuals are categorised based on their primary reason for inactivity, as determined by the strongest statistical predictors.
After identifying ten groups of economically inactive individuals, these were narrowed down to three target categories: people with a disability, people around retirement age, and women without caring responsibilities for young children. To refine the target groups into actionable categories, a two-step approach was used:
Selection based on group size: The largest groups of economically inactive individuals were identified. This process resulted in three main categories, with the presented numbers not being mutually exclusive such that some people may fall into more than one category:4
People receiving disability benefits. This group includes approximately 770 000 individuals.
Recipients of old-age benefits who are of working age. This group consists of approximately 310 000 individuals who do not receive disability benefits
Women with and without children under the age of six. This group consists of approximately 1 400 000 individuals who receive neither disability benefits nor old-age benefits.
Consultation with the Polish Public Employment Services (PES): Feedback was gathered from four Regional Labour Offices in Gdansk, Krakow, Olsztyn, and Opole, leading to two adjustments:
The second group, old-age benefits recipients, was expanded to include individuals above the statutory retirement age who are willing to work. This change was based on the low labour force participation rate among those just above retirement age (Figure 2.15) and evidence of PES success in engaging retired individuals at job fairs and other recruitment events.
The third group, women, was refined to exclude those with caregiving responsibilities for children under the age of six. This decision reflects that childcare availability and cultural factors are the main drivers of economic inactivity for women with young children. This choice is in line with previous OECD research that identified limited childcare options, particularly in rural areas, as a key factor contributing to women’s economic inactivity (OECD, 2021[1]). It reduced the total size of the target group to approximately 1 000 000 women.
The activation potential of each group depends on their willingness to work. While each of these target groups represent a potential source of labour, it is important to assess their actual interest in seeking employment. Some groups may have a high likelihood of inactivity but little willingness to re-enter the workforce. By understanding the willingness to work between target groups, the PES can establish clear benchmarks and targets while conducting internal capacity assessments, such as evaluating staff needs and the necessary scope of activation tools.
The willingness to work of each target group is assessed using three approaches: financial incentives, survey responses, and in-depth discussions with focus group participants. The willingness to work of the different target groups was assessed in three different ways:
Financial incentives to work: The first assessment estimates the average potential income for each group. Using EU-SILC data for Poland, potential gross income is imputed for economically inactive individuals based on employed individuals with similar work experience and educational attainment. Means-tested benefits are deducted and the individual-level data is then averaged for each group. Box 2.4 provides a detailed explanation of the methodology.
Survey responses: The second assessment relies on the self-reported willingness to work, captured through survey data. Specifically, LFS data for Poland includes a binary indicator that measures whether individuals express a willingness to work. The individual-level data is then averaged for each group.
Focus group discussions: The third assessment draws on in-depth discussions with focus group participants. This qualitative approach provides a more nuanced understanding of the needs and motivations of different target groups, complementing the survey data.
To assess the financial incentive to work for each target group, the analysis uses EU-SILC data for Poland and follows a three-step approach, ensuring that potential income estimates are available for all individuals aged 15 to 64.
In the first step, the potential gross monthly income of individuals who are not currently employed is imputed based on their most recent occupation, previous sector of economic activity, level of education, and degree of urbanisation of their current place of residence. The analysis also accounts for the interaction between sector of economic activity and education level. The gross income of employed individuals serves as the input data for the imputation, using their current occupation, sector, education level, and degree of urbanisation. To improve accuracy, a multiple imputation process is conducted 50 times, generating a range of plausible values for missing income data rather than relying on a single estimate. Each iteration introduces random variation based on the observed data distribution, drawing imputations from a linear regression model that accounts for uncertainty. The final potential income estimate for each individual is taken as the median of these 50 imputed values. The imputation process also ensures that imputed incomes remain within realistic bounds by truncating values between zero and the highest observed income in the employed population. For the small share of inactive individuals without any work experience, potential income is inferred from the average income of individuals with the same education level and one year of work experience.
In the second step, all means-tested benefits received by economically inactive individuals which they would forgo when taking up employment are deducted from their imputed income. The difference between the imputed gross income and lost benefits provides an estimate of the financial incentive to work.
In the third step, these individual-level estimates are averaged across the ten identified target groups.
One limitation of the described approach is that it does not account for income tax. The estimated financial incentive to work represents an upper bound and overestimates the difference between the potential (net) income and means-tested benefits for most individuals. The results nevertheless allow for a comparison of the financial incentive to work between the different groups of economically inactive individuals.
Analysing the relationship between the economic inactivity rate and the potential gross monthly income of different groups of economically inactive people reveals stark differences in income potential, and thus the financial incentive to seek employment. Figure 2.10 shows that the highest earning potential is observed among recipients of old-age benefits (group 2; 80% inactivity rate), women without children under 6 who have a university degree (group 8; 4% inactivity rate), and women with kids under 6 that do not receive childcare, with university education (group 5; 25% inactivity rate). The potential gross monthly income net of means-tested benefits lies around PLN 12 000 for all these groups. Conversely, groups with generally lower education levels exhibit the lowest earning potential. These include recipients of disability benefits (group 1; 82% inactivity rate), women without children under the age of six who are themselves over 26 years old, without a university degree but with some work experience (group 7; 14% inactivity rate), women with kids under the age of six that do not receive childcare, without university education (group 4; 52% inactivity rate), and women without children under the age of six who are themselves over 26 years old, with neither a university degree nor work experience (group 3; 77% inactivity rate). The potential gross monthly income net of means-tested benefits lies between PLN 6 700 and PLN 8 000. The lowest earning potential of around PLN 6 500 is observed among women without children under the age of six that receive childcare for their older children, who do not have university education and who themselves are 26 years old or younger (group 10; 1% inactivity rate).
There is no correlation between potential gross income and the economic inactivity rate across groups. Intuitively, one might expect a negative correlation between the financial incentive to work and the economic inactivity rate: The higher the earning potential, the more attractive work becomes, leading to a lower inactivity rate. However, Figure 2.10 shows no such relationship. One explanation is that factors such as work experience, which strongly influences salary expectations, may also make it more difficult for certain groups to find employment. For instance, among old-age benefit recipients, the high income potential could reflect a relatively high reservation wage. Higher salary expectations may reduce the likelihood of accepting available job opportunities and may lead fewer employers to hire older workers, contributing to continued inactivity among this group. A second explanation is that elements other than financial motivation affect the willingness to work differently across groups. For instance, some groups among the economically inactive, such as highly educated women with children, face challenges that reduce their ability to work, regardless of their relatively high income potential.
Potential gross monthly income minus means-tested benefits in PLN (y-axis) and the average economic inactivity rate in each group (x-axis)
Note: Economically inactive individuals are placed into groups based on their primary reason for economic inactivity determined by a decision tree with the hierarchical order “Recipients of disability benefits” → “Recipients of old-age benefits” → “Men/Women”. Thus, all groups labelled (2)-(10) do not include recipients of disability benefits and all groups labelled (3)-(10) further do not include recipients of old-age benefits. The size of the bubbles reflects the number of the economically inactive in each group, with larger bubbles indicating relatively larger groups. The groups are the following: (1) Recipients of disability benefits, (2) recipients of old-age benefits, (3) women without kids under 6, older than 26, without university education or experience, (4) women with kids under 6 that do not receive childcare, without university education, (5) women with kids under 6 that do not receive childcare, with university education, (6) women without kids under 6, older than 26, without university education, with some experience, (7) women with kids under 6 that receive childcare, (8) women without kids under 6, with university education, (9) males not included in other groups, (10) women without kids under 6 that receive childcare, without university education, 26 years old or younger. No childcare refers to at least one child under 13 years old who receives less than 30 hours of non-parental care per week.
Source: OECD calculations based on EU-SILC data, 2022.
Survey-based self-reported willingness to work indicators show significant variation across groups of economically inactive people. Survey responses based on LFS questions constitute a direct measure of the willingness to work in the different groups of the economically inactive. The indicator is based on individuals stating that they are either searching for or generally wanting a job. Box 2.5 shows the correspondence between the groups in EU-SILC and LFS data that serves as the basis for Figure 2.11. The figure shows that the self-reported willingness to work differs significantly across the groups of economically inactive. The highest willingness to work is observed among
Males who do not receive disability or old-age benefits (group I, 30% willingness to work),
Women without children under the age of six with no university degree aged 26 years or younger who do not receive disability or old-age benefits (group E, 28% willingness to work),
Women without children under the age of six, with a university degree who do not receive disability or old-age benefits (group F, 27% willingness to work), and
Women without children under the age of six, aged 26 years and above, without a university degree but with some experience, who do not receive disability or old-age benefits (group D, 21% willingness to work).
The lowest willingness to work is found among
Those of working-age who are inactive due to retirement, excluding those inactive due to disability (group B, 3% willingness to work), and
Those inactive due to health problems (group A, 10% willingness to work).
Despite a low willingness among some groups, their large absolute numbers can make them important target groups. For instance, people receiving disability benefits willing to work make up around 77 000 people.
While the groups defined based on EU-SILC data cannot be mapped perfectly onto LFS data, reasonable approximations allow to identify the same groups in both data sources. Table 2.3 shows the correspondence.
Source: OECD elaboration.
The main limitation in the above correspondence table is the match between recipients of disability benefits (EU-SILC) and those who report being inactive due to health problems (LFS), as well as the match between recipients of old-age benefits (EU-SILC) and those who report being inactive due to retirement (LFS). The limitation occurs due to a lack of information on benefits in the LFS data. Thus, while there is still large overlap between these groups, the LFS-based economic inactivity rates within the groups as shown, for example in Figure 2.11, are at 100% by design and are only included for completeness.
The share of each group indicating the willingness to take up employment (y-axis) and the average economic inactivity rate in each group (x-axis)
Note: The size of the bubbles reflects the number of the economically inactive in the group. Groups indicated as follows: (A) inactive due to a health problem, (B) inactive due to retirement, (C) women, no children under 6, no university, older than 26, no experience, (D) women, no children under 6, older than 26, no university, some experience, (E) women, no children under 6, no university, 26 or younger, (F) women, no children under 6, university, (G) women, child under 6, no university, (H) women, child under 6, university, (I) males who are not part of group A or B.
Source: OECD elaboration based on EU-LFS data, 2021.
Survey-based measures that assess individuals’ willingness to work through a binary question may lack highly relevant nuance. One major drawback of using a binary question to assess individuals’ willingness to work is that it lacks nuance. Respondents may interpret the question on wanting a job in relation to their current life situation rather than their general willingness to work if employment barriers were removed. This distinction is particularly relevant for groups facing barriers beyond their individual control, such as caregiving responsibilities when childcare or eldercare facilities are unavailable, discrimination in the hiring process or a (perceived) lack of appropriate local work opportunities. FGIs provide a more nuanced understanding of willingness to work, help identify employment barriers and allow for discussions of “what-if” scenarios. Box 2.3 describes the FGIs conducted to explore these issues in more detail.
Economically inactive FGI participants indeed state that they mostly remain without employment against their will and intentions. Among FGI participants, economic inactivity was seen as a voluntary choice only in cases involving people mentioning health-related reasons for their economic inactivity and people above the statutory retirement age without the financial need to work. Many expressed a willingness to endure disadvantages, such as accepting short employment contracts, shorter working hours, or even working without a formal contract. Most participants were also open to enhancing their skills but felt that better support from Public Employment Services would be needed in this area, for example, in training specifically for in-demand qualifications in the labour market.
Economic inactivity often results from a combination of unfavourable factors. Among FGI participants, the causes of inactivity were strongly linked to demographic characteristics such as age and gender, which in turn were associated with perceived discrimination in the labour market, health constraints, or caregiving responsibilities. Additionally, many FGI participants reported that their economic inactivity stems from a lack of job opportunities in local labour markets, a mismatch between individual professional competences and employer needs, or misaligned salary expectations between jobseekers and employers. Many FGI participants reported that these factors often coincide and, in some cases, reinforce one another.
FGI participants reported financial concerns as an important motivation to seek employment. For example, retirees and people with disabilities often find that their social benefits are insufficient to meet basic living expenses. As a result, their willingness to seek employment is driven more by necessity than by a desire for professional growth or self-fulfilment. Additionally, many economically inactive individuals also remain financially dependent on their partners or parents. For instance, many women in the surveyed group indicated a desire to pursue work for both financial reasons and personal development. The primary barrier to employment for these women is the unequal distribution of household responsibilities, which often disproportionately falls on them.
Prolonged negative experiences in the labour market and societal exclusion significantly affect the psychological well-being of FGI participants, often discouraging them from continuing their job search. Many economically inactive individuals reported a decline in self-confidence due to extended periods of unemployment and repeated, often unsuccessful, job applications. For some, this loss of confidence in securing stable, quality employment led to reduced job-seeking efforts. Factors such as age discrimination and unfair treatment by employers further diminished their motivation to pursue work. Additionally, many participants highlighted the lack of social interaction resulting from not having a job as another factor negatively affecting their mental well-being.
Target group I: Disability benefit recipients
Recipients of disability benefits are the largest group of the inactive in Poland, totalling around 770 000 people who are not working. EU-SILC data shows that almost all in this group suffer from a form of chronic illness or condition (94%), however, only about half (45%) face severe difficulties in daily activities. Thus, focusing on disability benefits recipients who can and are willing to work presents an opportunity for reducing economic inactivity. Moreover, activating this group can have significant positive spillover effects. About 40% of the inactive individuals with poor health live in households where no one is employed, despite only 9% living alone, suggesting that their caretakers are inactive as well.
In 2022, the inactivity rate of people with disabilities, aged 15-64, in Poland was notably higher than that of their counterparts on the EU level, highlighting a potential area for increased workforce participation. In 2022, 45% of 15–64-year-old people with disabilities in the EU-27 were inactive, compared to 23% of those without a disability (Figure 2.12 and Box 2.6). In contrast, the economic inactivity rate among 15–64-year-old individuals with disabilities in Poland was significantly higher at 66% in 2022, compared to 23% among those without a disability. Thus, the gap in the economic inactivity rate between people without a disability and people with a disability is significantly larger in Poland than in the EU-27.
In Poland, three degrees of disability are defined in the Act of 27 August 1997 on Vocational and Social Rehabilitation and Employment of Persons with Disabilities (Government of Poland, 1997[21]). These are determined based on the assessed degree of limitation in bodily functions, ability to work, and capacity to function in daily life:
Light degree of disability - includes people with partial inability to work. People in this category can typically perform work under appropriately adapted conditions and do not require permanent or long-term care.
Moderate degree of disability - includes people with total or partial inability to work who require temporary or partial assistance in daily functioning.
Severe degree of disability - includes people who are totally incapacitated and require permanent or long-term care, and assistance in daily activities.
The degree of disability is determined by government evaluation committees which consider medical records and the impact of the disability on the person's professional and social life.
Since the degree of disability is defined differently across OECD countries, internationally comparable statistics shown in this report are based on EU-SILC survey data, which includes a question on “limitations in activities due to health problems”. Survey respondents can choose between “severely limited”, “limited but not severely” and “not limited at all”. All respondents who report to be either severely or somewhat limited in their activities are used to calculate and compare the economic inactivity rate of people with a disability internationally.
Economic inactivity rate of people with some or severe level of disability, 15–64-year-olds
Note: All respondents who report to be either severely or somewhat limited in their activities are categorised as having some or severe level of disability, see Box 2.6 for details.
Source: OECD calculations based on EU SILC 2022.
Individuals with poor health, a proxy for those on disability benefits, who are inactive in Poland face multiple barriers to employment, including a lack of work experience and relatively low education levels. A third of this group has no work experience at all, which provides a partial explanation for the relatively low potential income and high likelihood of inactivity compared to other groups (Figure 2.13, Panel A). On average, individuals with disabilities have only worked for 40% of the time they could have potentially been employed. Similarly, the level of education is low among people with a disability compared to the general population in Poland. In 2022, about 26% of the 16–64-year-old males and 16–59-year-old females with a disability completed lower secondary education or less as their highest educational attainment, compared to 13% of the 15–64-year-old general population in Poland. In the same year, 59% of people with a disability held a vocational degree (general population: 46%), and only 7% had higher education (general population: 11%).
Age also plays a role in shaping employment barriers, as the share of disability benefit recipients increases with age. The largest proportion of economically inactive disability recipients is found in the 55-64 age group (Figure 2.13, Panel B). With age, individuals are more likely to develop chronic illnesses, sustain injuries, or face other conditions that limit their ability to work. This intersection of age and disability means many disability benefit recipients also encounter challenges typical for those nearing retirement, such as outdated skills due to prolonged inactivity or age-related discrimination in the labour market.
Potential gross monthly income minus means-tested benefits in PLN (y-axis) and the average economic inactivity rate in each group (x-axis) aged 15-64 (Panel A) and recipients of disability benefits by age (Panel B)
Note: (Panel A) Economically inactive individuals are placed into groups based on their primary reason for economic inactivity determined by a decision tree with the hierarchical order “Recipients of disability benefits” → “Recipients of old-age benefits” → “Men/Women”. Thus, all groups labelled (2)-(10) do not include recipients of disability benefits and all groups labelled (3)-(10) further do not include recipients of old-age benefits. Size of the bubbles reflects the number of economically inactive in the group. Groups indicated as follows: (1) recipients of disability benefits, (2) recipients of old-age benefits, (3) women without kids under 6, older than 26, without university education or experience, (4) women with kids under 6 that do not receive childcare, without university education, (5) women with kids under 6 that do not receive childcare, with university education, (6) women without kids under 6, older than 26, without university education, with some experience, (7) women with kids under 6 that receive childcare, (8) women without kids under 6, with university education, (9) males not included in other groups, (10) women without kids under 6 that receive childcare, without university education, 26 years old or younger. No childcare refers to at least one child under 13 years old who receives less than 30 hours of non-parental care per week.
Source: OECD elaboration based on EU SILC 2022 and EU LFS 2021.
FGI participants with disabilities report that employers are reluctant to hire them partly due to their disability. While some individuals with disabilities are entirely unable to work due to health conditions, the majority can still perform professional duties. For instance, FGI participants with a mild or moderate disability stated that their disability is generally compatible with most types of work. Participants shared various work experiences in Poland and abroad, highlighting that disabilities like deafness do not significantly limit their ability to perform tasks, particularly with the aid of modern communication technologies.
From the perspective of people with disabilities, employers often have misinformed attitudes about their capacity to work, leading to unnecessary exclusion from the labour market. According to FGI participants, there is a widespread negative perception of people with disabilities as job applicants, even when their disability does not prevent them from fulfilling job requirements. FGI participants with a disability believe that a lack of education about different types of disability is at least partially responsible for this. Educating employers and changing their perceptions could help improve employment opportunities for people with disabilities.
Target group II: People around retirement age
Polish society is aging rapidly, increasing the need to boost employment among older workers. The median age in Poland is projected to increase from 39 years in 2015 to 45 in 2030 and 50 in 2050 (Bień, 2024[22]). By 2035, the working-age population will decrease by 2.1 million (Kukołowicz, Leszczyński and Lubasiński, 2024[7]). This demographic shift will require higher labour market participation from older workers, as fewer young workers will be available to fill job vacancies. Economically inactive people aged slightly below the statutory retirement age of 60 for women and 65 for men who receive old-age benefits currently make up more than 300 000 people across Poland. Although the number of early retirees in Poland has decreased in recent years, they still constitute the third largest economically inactive group in Poland (Figure 2.14, Panel A). Among these, 89% are men who most commonly worked in sectors such as mining, metal production, or industrial processing (The Social Insurance Institution, 2023[23]). Their age distribution indicates that, while most early retirees are found in the 55-64 age bracket, more than 20% are below the age of 55 (Figure 2.14, Panel B).
Potential gross monthly income minus means-tested benefits in PLN (y-axis) and the average economic inactivity rate in each group (x-axis) aged 15-64 (Panel A) and recipients of old-age benefits before retirement age, by age (Panel B).
Note: (Panel A) Economically inactive individuals are placed into groups based on their primary reason for economic inactivity determined by a decision tree with the hierarchical order “Recipients of disability benefits” → “Recipients of old-age benefits” → “Men/Women”. Thus, all groups labelled (2)-(10) do not include recipients of disability benefits and all groups labelled (3)-(10) further do not include recipients of old-age benefits. Size of the bubbles reflects the number of economically inactive in the group. Groups indicated as follows: (1) recipients of disability benefits, (2) recipients of old-age benefits, (3) women without kids under 6, older than 26, without university education or experience, (4) women with kids under 6 that do not receive childcare, without university education, (5) women with kids under 6 that do not receive childcare, with university education, (6) women without kids under 6, older than 26, without university education, with some experience, (7) women with kids under 6 that receive childcare, (8) women without kids under 6, with university education, (9) males not included in other groups, (10) women without kids under 6 that receive childcare, without university education, 26 years old or younger. No childcare refers to at least one child under 13 years old who receives less than 30 hours of non-parental care per week.
Source: OECD elaboration based on EU-SILC data (2022, Panel A) and LFS data (2021, Panel B).
Early retirement schemes such as bridging pensions and teacher compensation benefits incentivise early retirement. In 2024, 40 200 people received bridging pensions, a temporary benefit granted to individuals working in special conditions or performing jobs of a special nature until they reach the statutory retirement age, with 91% being men.5 This is a result of big share of male employment in the sectors like heavy work, such as loading operations, mining, or metallurgy which fall under having worked in special conditions or of a specific nature (The Social Insurance Institution, 2024[24]). Teacher compensation benefits apply to teachers with at least 20 years of service and a minimum of half-time teaching hours. In total, 12 500 teachers received this benefit for 2024, 73.2% of whom were women, allowing for early retirement (The Social Insurance Institution, 2024[24]). These schemes, while providing financial security, also incentivise early retirement.
Labour force participation rates of people aged up to five years above the statutory retirement age are also low in OECD comparison. The relatively low labour force participation of those nearing the statutory retirement age is mirrored by a low labour force participation rate of those above the age of statutory retirement in Poland. Figure 2.15, Panel A shows that, in 2022, men up to five years above the statutory retirement age was low at 18%, compared to the OECD average of 36%. Similarly, in the same year, the labour force participation rate of women up to five years above the Polish statutory retirement age was at 24%, significantly below the OECD average of 47% (Panel B).
Labour force participation of men aged 65-69 in OECD countries in 2022 (Panel A) and labour force participation of women aged 60-64 in OECD countries in 2022 (Panel B)
Source: OECD Statistics, 2022, Employment and unemployment by five-year age group and sex – indicators.
Economically inactive FGI participants around the statutory retirement frequently mentioned employers’ negative perception of their advanced age as a key barrier to employment. Both those nearing retirement and those above retirement age reported that employers tend to view their advanced age unfavourably, often preferring younger candidates. Many participants believe that employers associate younger employees with greater flexibility and adaptability. Many in the group also expressed the belief that their skills and competencies are undervalued by employers. The overall perception among FGI participants is that hiring decisions are often motivated by age, which they perceive as a form of discrimination. They acknowledge older adults may face greater geographical mobility restrictions, which can further reduce their chances of being hired.
Academic literature supports the existence of age discrimination in the Polish labour market. Studies have not identified a clear rationale for the relatively lower hiring rates of older workers, suggesting that some hiring decisions may be shaped by stereotypes. Research indicates that more than two-thirds of individuals aged 50-59 would prefer to hire a 30-year-old man over a similarly qualified 60-year-old candidate (Baszczak et al., 2021[25]).
Once retired, a low willingness to work may reduce the effectiveness of activation efforts, highlighting the need for preventive measures targeting individuals nearing retirement. Figure 2.11 shows that only 3% of those inactive due to retirement are willing to work. EU-SILC data shows that, on average, in 2022, this group reported relatively high non-work income of around PLN 6 500. Still, around 12% of individuals in this group experienced difficulty making ends meet. Evidence suggests that a preference for spending time with family over working further contribute to the groups’ low willingness to re-enter the labour market (Interreg Danube, 2024[26]). Health-related factors also contribute, as individuals in poor health are less likely to stay employed (Reine and Rajevska, 2024[27]). Taken together, the findings suggest that preventive measures targeting older workers who face the decision to retire could prove more efficient. For instance, flexible work arrangements could incentivise older workers to reduce their working hours instead of retiring (see also Figure 2.7 and Figure 2.8).
Target group III: Women without young children
The labour force participation rate of women in Poland is below that of most OECD countries. In 2023, 68.9% of women of working age in Poland were active on the labour market, compared to 70.3% of women on average in the EU-27 (Figure 2.16). Although Poland’s female labour force participation rate is above the OECD average of 66.6%, only eight OECD countries report a lower participation rate. Female labour force participation in Poland also falls behind regional neighbours such as Czechia (70.4%), the Slovak Republic (72.8%) and Hungary (73.6%). Across all age groups, women in Poland have lower labour force participation rates than men, indicating persistent gender-specific barriers to employment.
Labour force participation of women aged 15-64 in OECD countries
Note: The figure excludes Chile due to a lack of available data.
Source: OECD Statistics, Infra-annual labour statistics.
Women who do not receive disability or old-age benefits make up nearly half (46%) of the inactive population in Poland and represent a diverse group. The seven subgroups of economically inactive women who do not receive old-age or disability benefits vary in their socio-economic characteristics. The main distinctions between groups include:
Holding a university degree;
having children under the age of 6;
and having work experience.
These characteristics also determine differences in the likelihood of economic inactivity and the willingness to work across the different groups of inactive women (Figure 2.11).
Women without a university degree have a relatively low income potential and are more likely to be economically inactive if they have young children to care for or lack work experience. Figure 2.17 shows that women who do not receive disability or old-age benefits and do not hold a university degree experience relatively high levels of economic inactivity. This is evident in groups 3, 4, 6, and 7, represented by darkblue bubbles, compared to women with university degrees (groups 5 and 8, shown in light blue). Among women without university degrees, the main barriers to labour market integration are the absence of work experience (group 3) and caring responsibilities for children under six (group 4). Women without university education who have some work experience and who do not have children under the age of six to care for (group 6) as well as young women without children (group 10) show significantly lower economic inactivity rates. Across all groups of women without university degrees, the difference between the potential income and means-tested benefits is significantly below that of women with university degrees. Thus, the higher economic inactivity rates among women without university degrees partially reflects their lower opportunity costs.
Groups of economically inactive in Poland by their likelihood of economic inactivity (x-axis) and their expected earnings on the labour market (y-axis).
Note: Economically inactive individuals are placed into groups based on their primary reason for economic inactivity determined by a decision tree with the hierarchical order “Recipients of disability benefits” → “Recipients of old-age benefits” → “Men/Women”. Thus, all groups labelled (2)-(10) do not include recipients of disability benefits and all groups labelled (3)-(10) further do not include recipients of old-age benefits. The size of the bubbles reflects the number of economically inactive in the group. Groups indicated as follows: (1) recipients of disability benefits, (2) recipients of old-age benefits, (3) women without kids under 6, older than 26, without university education or experience, (4) women with kids under 6 that do not receive childcare, without university education, (5) women with kids under 6 that do not receive childcare, with university education, (6) women without kids under 6, older than 26, without university education, with some experience, (7) women with kids under 6 that receive childcare, (8) women without kids under 6, with university education, (9) males not included in other groups, (10) women without kids under 6, without university education, 26 years old or younger. No childcare refers to at least one child under 13 years old who receives less than 30 hours of non-parental care per week.
Source: OECD elaboration based on EU-SILC data, 2022.
A strong predictor of economic inactivity among women who do not receive old-age or disability benefits is the responsibility of caring for a young child at home, reflecting a disproportionate burden of care responsibilities. Figure 2.16 shows that women with children at home who lack access to external childcare (groups 4 and 5) are more likely to be inactive than those with children under six who do have external childcare available (group 7). This disparity is in large parts due to the unequal distribution of childcare and household responsibilities, exacerbated by the limited availability of flexible work arrangements (Magda, 2020[12]).
Access to childcare services for very young children is still insufficient in Poland. In 2019, only 12% of children under the age of three were enrolled in childcare. Large differences in childcare coverage also exist across voivodeships, a 19% coverage in Lower Silesia, compared to a coverage of only 9% in Swietokrzyskie and Warmian-Masuria (OECD, 2021[1]). Figure 2.18 shows that, as a result, women’s economic inactivity rate in Poland rises sharply after childbirth and only begins to decline when children reach the age of three and start preschool. However, even as childcare responsibilities lessen, women’s labour force participation does not return to pre-childbirth levels. For women without university degrees, the economic inactivity rate remains around 15 percentage points higher than that of childless women under 30, even when their oldest child is 12 years old or older.
Economic inactivity rate among women (y-axis) and the age of the oldest child in the household (y-axis).
Note: At three years old, children in Poland enter pre-schools. While attendance is non-compulsory, universal pre-school education is available in Poland. At six years old, children enter compulsory schooling. The graphs are the result of a linear cross-sectional regression of a binary economic inactivity indicator on a categorical variable that indicates the age bracket of the youngest child in the household (x-axis). This categorical variable is further interacted with a binary indicator that captures whether the woman holds a university degree. For households in which no children are present, only women aged 30 and below where considered.
Source: OECD elaboration based on EU-SILC data, 2022.
The provision of childcare is outside the purview of the Public Employment Services in Poland and the responsibility of gminas, the local government that corresponds to municipalities. Childcare in Poland operates under two systems: the childcare system, supervised by the Ministry of Family, Labour and Social Policy, for children aged between zero and three years, offering care in nurseries (żłobki), kids’ clubs (kluby dziecięce), and through day-care providers (opiekun dzienny) or nannies (niania); and the education system, overseen by the Ministry of National Education, for children aged three to six/seven years old, providing preschool education in nursery schools (przedszkole), preschool classes (oddziały przedszkolne) in primary schools, and other preschool settings like preschool education units (zespół wychowania przedszkolnego) and centres (punkt przedszkolny) (Eurydice, European Commission, 2024[28]). The gmina, the local government that corresponds to a municipality, is responsible for the delivery of both systems. Key programs include Active Toddler (Aktywny Maluch), which subsidises childcare for children up to the age of three, and the Aktywny Rodzic program, launched in 2024, which offers financial aid for working parents and childcare services. Access to childcare has improved significantly over the past decade, with high enrolment rates for preschool education of 115% in urban areas (indicating some attendance from children residing in rural areas) and 73% in rural areas for 2023/2024 (Statistics Poland, 2024[29]).
Women without children under the age of six who hold university degrees often choose to remain inactive, and many have substantial non-work income. Analyses based on EU-SILC data show that, although a third of inactive women with no care responsibilities and university degrees have chronic illnesses, they are generally at low risk of poverty, often due to other household members working. Many (73%) previously fulfilled childcare responsibilities and still have a child under the age of 18 in the household. Women with higher education and no childcare responsibilities have the highest income potential among all target groups and predominantly live in urban areas (80%).
Women without children under the age of six, without a university degree and with no childcare responsibilities struggle with returning to the labour market. The willingness of women without higher education to work depends on their past experiences in the labour market. Those who have worked prior to childbirth show a relatively high willingness to work (21%). The majority of women with some work experience (51%) hold vocational degrees, but many were previously employed in low-income occupations.
Women without children under the age of 6 without a university degree and who have no previous work experience face the largest challenge in finding employment. These women face compound barriers to employment, due to their low education levels and non-child-related care responsibilities. A quarter has only primary education, and thus the income potential is relatively low. Furthermore, 20% live in a household that includes someone with a disability or poor health which may hinder their access to job opportunities. A majority (58%) live in rural areas, further limiting employment options. These factors influence the moderate willingness to work (17%) in this group.
Beyond childcare, focus group discussions revealed broader gendered barriers to labour market participation. Women reported facing persistent social expectations that reinforce their role as primary caregivers, often leading to financial dependency on their partners. Many also cited a lack of partner support as a reason for abandoning job searches, particularly when their partners earned higher wages. Older women, especially those nearing retirement, frequently mentioned age discrimination and the perception that their skills are less valuable as barriers to finding employment. In some regions, a mismatch between women’s skills and the physical demands of available jobs further restricts their employment opportunities, in particular when employers in male-dominant sectors are unwilling to consider female applicants. The absence of prior work experience compounds these barriers, making re-entry into the labour market even more difficult after periods of inactivity.
Regional differences in the population characteristics of economically inactive people
The economic inactivity rate and the composition of target groups within the share of the inactive differs between macro- regions in Poland. Figure 2.19 shows that, once students are excluded from the underlying population and the sample is restricted to those of working age in Poland, the regional differences shown in Figure 2.4 (Panel B) persist. Lesser Poland and Silesia record the highest population share of economically inactive receiving old-age benefits (4%) in 2021, whereas this share stood at only 1% of the working-age population in all other macro-regions Lesser Poland and Silesia are the centre of the Polish mining sector, where workers are often entitled to early retirement benefits (Frankowski and Mazurkiewicz, 2020[30]). The share of inactive women ranges from 7% in Greater Poland, West Pomerania, and Lubusz to 4% in Warsaw and Mazowiecki region. Disability benefit recipients are most prevalent in Greater Poland, West Pomerania, Lubusz, Kuyavian-Pomerania, Warmian-Masuria, Pomerania, Lublin, Podkarpacia, and Podlaskie, where the share reaches 5%. Men who do not receive disability or old-age benefits show a consistent 1% inactivity rate across all regions.
Share of the economically inactive in the working-age population by macro-regions and sub-groups, 2021
Note: Working-age population is defined as those aged between 18 and 64 (men) and 18 and 59 (women) years old, excluding students. The labels “women” and “men” refer to women and men who do not receive disability or old-age benefits.
Source: OECD elaboration based on EU-LFS data, 2021.
Causes of economic inactivity also vary by powiat. In Poland’s 2021 population census, the primary reasons for inactivity were divided into four categories: health problems, retirement, childcare responsibilities, and discouragement from working. In most powiats, health issues are the leading cause of inactivity but the reasons for economic inactivity vary significantly across powiats (Figure 2.20). For instance, the share of early retirees in the total working-age population (excluding students) differs between 1% in the City of Wroclaw and 8% in Rybnicki, Silesia. Local differences are apparent in Silesia and can be linked to early retirement among mining workers in some powiats. Similarly, the share of individuals who state to be economically inactive due to being in poor health, which includes those with disabilities, varies between 2% in the City of Wroclaw and 6% in Chelmski, Lublin. The share of discouraged workers ranges from close to 0% in Bierunsko-Ledzinski, Silesia to 5% in Przysuski, Mazowiecki region, whereas those citing childcare responsibilities as their main reason for inactivity represent 1% (City of Opole) to 4% (Rybnicki, Silesia) of the working-age population (excluding students). Discouragement as a reason for economic inactivity is relatively more common in rural powiats. The 20 powiats with the highest share of discouraged workers have an average population of 49 763, ranging from 20 601 to 82 151 residents. Conversely, the 20 powiats with the lowest share of discouraged individuals had an average population of 263 755, ranging from 51 560 to 1 861 599. Seven of the 20 powiats with the lowest share are cities with powiat status.
Note: Data analysed at the powiat level. Borders are also shown at the regional (voivodeship) level, including for the Warsaw capital region.
Source: OECD calculations based on Population Census Data, 2021.
According to FGI participants, the stark contrast in employment opportunities between rural and urban areas makes it more challenging to find work outside large urban labour markets. FGI participants across all target groups highlighted several reasons why discouragement is a more common driver of economic inactivity in rural areas compared to cities. Businesses in rural Poland tend to be less productive and operate with smaller profit margins, resulting in more frequent layoffs, fewer job opportunities, and less stable employment (OECD, 2021[1]). The underdeveloped and infrequent public transportation network further restricts access to jobs beyond an individual's place of residence. This issue is particularly severe for individuals nearing retirement age, many of whom do not own cars and are less willing or able to relocate for work.
Women are more likely than men to cite childcare responsibilities and discouragement as the primary reason for their inactivity, regardless of whether they live in urban or rural powiats. In contrast to men, childcare responsibilities are the primary cause of inactivity for women across most rural and urban powiats, with an average of 4.2% compared to just 0.4% for men (Figure 2.21). On average, men are more likely to cite health issues (4.7%) as the primary reason for their inactivity than women (3.5%) but health problems are also frequently cited as the most common reason for economic inactivity in the eastern regions of Poland among women. Across all powiats, a smaller share of working-age women report retirement as the cause of their inactivity (2.1%) compared to men (4.0%). Women (2.5%), on the other hand, are more likely to report discouragement as a reason for their inactivity compared to men (1.3%).
Note: Data analysed on the powiat level. Borders shown on the regional (voivodeship) level, distinguishing the Warsaw capital region.
Source: OECD calculations based on Population Census data, 2021.
The three identified target groups among the economically inactive, people receiving disability benefits, people around retirement age who receive old-age benefits, and women without young children at home, represent a significant untapped labour force. In Poland, there are nearly 770 000 people on disability benefits of working-age, 310 000 working-age individuals receiving old-age benefits who do not receive disability benefits and around 1 000 000 working-aged women without caregiving responsibilities towards children under the age of six who do not receive disability benefits or old-age benefits. Taken together, these target groups make up a significant share of the 5.2 million working-age economically inactive people. Even the most conservative survey-data based estimates show that around 350 000 among these economically inactive people are willing to work. In-depth focus group interviews further reveal that those unwilling to work are often unavailable only under their current life circumstances. If the barriers to employment were removed, a much larger share among the economically inactive could be integrated into the labour market.
To effectively integrate economically inactive people into the labour market, it is essential to address the numerous, often interrelated barriers they face. These challenges stem from both individual circumstances and local labour market conditions. Common obstacles include low educational attainment, lack of recent work experience, limited job opportunities, skill mismatches, employer discrimination, and a shortage of flexible work options. Individual characteristics such as disability, gender, and age can further exacerbate these barriers. For example, people with disabilities may struggle to find employers willing to provide necessary workplace accommodations. Early retirement schemes and low statutory retirement ages encourage older workers to leave the labour force earlier than in other OECD countries. Meanwhile, women without young children often remain constrained by traditional gender roles, facing a disproportionate share of household responsibilities with little support from their partners in pursuing employment.
The salience as well as the reasons for economic inactivity differ significantly by region and within regions, implying that different will have to prioritise different target groups. Economic inactivity rates range from just 17% in parts of the Mazowieckie region, including Warsaw and its surrounding powiats, to over 35% in rural or economically lagging areas such as Swidwinski in West Pomerania, Walbrzych and Walbrzyski in Lower Silesia, and Sztumski in Pomerania. Urban centres like Krakow, Wroclaw and Poznan also show relatively low inactivity rates. The composition of the economically inactive population similarly varies geographically. For example, early retirees account for only 1% of the working-age population in Wroclaw, compared to 8% in Rybnicki, a city at the core of Silesia’s Rybnik Coal Area. Similarly, the share of individuals who state to be economically inactive due to being in poor health, which includes those with disabilities, varies between 2% in Wroclaw and 6% in Chelmski, Lublin. These geographic disparities in the main reason for economic inactivity mean that the PES in each voivodeship and powiat will have to adjust outreach and activation measures according to regional and local needs.
Various technical approaches can be employed to segment economically inactive populations into target groups for PES. These range from traditional clustering methods, such as k-means analysis, to probabilistic clustering techniques like Latent Class Analysis (LCA). They also include machine learning algorithms, from simple decision trees to advanced ensemble methods like random forests or gradient-boosted trees.
Each approach to categorising economically inactive individuals has its advantages, often balancing policy relevance and simplicity against statistical precision. For example, the most widely used method, LCA, identifies subgroups within populations by analysing observable characteristics (OECD, 2022[31]). LCA uncovers latent, or unobservable, subgroups within the economically inactive population and assigns individuals to groups based on probabilities. This method can identify patterns and correlations of different employment barriers. However, LCA has two limitations for policymakers. First, it focuses solely on the inactive population, excluding information about active individuals who could provide valuable insights as a comparison group. This exclusion may lead to less robust groupings since it ignores the continuum and differences between economic inactivity and activity. Second, while LCA produces statistically precise classifications, it often results in numerous groups characterised by complex and overlapping traits. These outputs, while insightful for researchers, may be less intuitive for policymakers seeking clear and actionable categories. Simple machine learning methods offer higher interpretability and, when complemented by additional quantitative and qualitative analyses, can address some limitations of statistical precision while providing actionable insights for policymakers.
Decision trees offer a data-driven method for defining target groups among the economically inactive population that is easy to interpret for policymakers. The approach works by iteratively splitting the data based on observable characteristics, selecting at each step the feature that best separates active and inactive individual. In the context of categorising economically inactive people, these features typically reflect socio-economic characteristics or predefined employment barriers. The splitting process continues until the data is fully partitioned or a stopping criterion is applied to prevent overfitting (Myles et al., 2004[20]). The resulting groups are as distinct as possible in terms of inactivity, enabling policymakers to identify shared characteristics and barriers among target groups. These characteristics can then be used to define the groups in an intuitive way. Moreover, decision trees naturally select the largest subgroups within the economically inactive population, and thus meeting the efficiency requirement of employment activation policies.
The splitting process in decision trees relies on a technique called greedy recursive binary splitting. At each step, the predictor space, i.e. the underlying dataset, is divided into two groups based on a single observable characteristic, a process called binary splitting (Myles et al., 2004[20]). The algorithm then evaluates all possible splits at a given step and selects the one that is optimal for that specific step, without considering its impact on future splits, a process referred to as greedy splitting. To determine the best split, the algorithm minimises the likelihood of misclassifying an instance. This minimisation process is measured using the Gini Index,
Gini ,
where represents the number of the node in the tree and is the number of inactive individuals within that specific node, and N is the number of active individuals. The Gini Index is thus a metric of impurity within a group: It takes values between 0, which represents a pure group, where all instances belong to the same class of active or inactive individuals and 0.5, which represents a completely random assignment between classes. For a given split, the Gini Index is calculated as a weighted average of the Gini values for the two resulting subgroups. A lower Gini Index indicates a better split. The algorithm applies this procedure recursively, splitting the data until all groups are "pure", i.e. they only contain active or inactive individuals, or until a stopping criterion is met to prevent overfitting. For the purposes of creating subgroups of economically inactive people, the stopping criteria was set to 10 subgroups for two reasons: First, to keep the number of subgroups manageable and actionable for policymakers. Second, to avoid small subgroups below a predefined sample size that would make the group definition unreliable and less likely to replicate when using a different dataset.
The primary dataset used for the analysis is the European Union Statistics on Income and Living Conditions Survey (EU-SILC) for Poland. The working sample covers all active and inactive individuals in Poland aged 16-64 years for males and 16-59 years for females, which corresponds to the working age for these groups in the country. To ensure the dataset only reflects the economically active population of interest to the PES, students have been excluded, as their investment in education is expected to enhance their future employability. The analysis includes socio-economic variables, such as age, gender, and household composition, alongside a range of barriers that may affect employability. These barriers are categorised into experience and labour demand barriers (including educational attainment and lack of experience), health barriers (limited activities due to illness or chronic illness), and family barriers (presence of children, elderly, or sick household members). Additionally, factors related to motivation to work are captured, including non-work income and the employment status of others in the household. Other variables include migration background and geographical location. Finally, benefits received by individuals are also considered, including disability, old-age, unemployment, sickness, child, social, and housing benefits. These benefits only partially present employment barriers through their disincentive to seek employment and mostly capture specific individual characteristics that sometimes overlap with other socio-economic characteristics. All variables used in this analysis are further defined in table Annex Table 2.A.1.
FGIs were organised for each group and attended by five to ten participants. In total, 35 economically inactive people participated in the interviews. The characteristics of each group in terms of their socio-demographic features were as follows:
Individuals approaching retirement age: The participants in this group were predominantly women (six out of ten people). Most were rural residents. The majority held vocational education with only one person reporting having completed higher education. Respondents indicated having at least 20 years of work experience, with some having up to 40 years. This experience was gained in various professions, including working abroad (two out of ten people). Their unemployment period prior to the interview typically lasted several years. All participants of pre-retirement age were economically inactive but, when possible, they took up casual employment, often without a formal work contract.
Economically inactive individuals above statutory retirement age: In this group, women were the majority (eight out of ten people), with most participants living in rural areas. Most participants held vocational degrees (including technical secondary education), and a few individuals held university degrees. On average, the interview participants had several decades of professional experience. Four individuals in this group were engaged in formal gainful activity because of insufficient pension benefits.
People with a disability: Respondents in this group were people from various age groups (20-57 years old), with a majority being women (seven out of ten people). Most were residents of the region’s capital (eight out of ten people). Only two participants lived in rural areas and had to rely on third-party transportation. The majority of respondents in this group had vocational education, including technical secondary education (six people), and four had completed higher education. Four out of ten individuals had no professional experience, while the others had between several years up to 35 years of work experience, often in jobs that did not require high levels of qualifications. In this group, the period of economic inactivity lasted several years on average and was most often related to deteriorating health.
Women with children aged above 6 years: The respondents were women aged 30-40 years, predominantly living in cities, with most residing in the region’s capital. In one case, the respondent lived in a small town on the province's outskirts. Respondents held either secondary or higher education. Their professional experience was limited, usually ranging from a few months to a maximum of one year of work. Only in one case, the respondent had 20 years of physical labour experience abroad. The period of economic inactivity among the participants ranged from two to eight years, mainly due to taking on childcare duties, though other reasons were mentioned, including low qualifications and health issues. Most respondents manage households with their husbands, raising between one and three children. Most children were in their teenage years. Participants stated that, during their spell of economic inactivity, they sometimes worked without a formal employment contract.
The meetings aimed to address the groups’ motivation for working or finding employment. The discussions held during the meetings involved:
The identification of the main reasons for inactivity in the identified groups, including overlapping factors behind economic inactivity;
The identification of the willingness to work among economically inactive groups;
Determining the need for support offered by companies, Public Employment Services and other institutions to (re-)integrate economically inactive people in the labour market.
The interviews took place in the four Polish cities of Gdańsk, Kraków, Olsztyn, and Opole. Participants were recruited by the Polish Public Employment Services’ Regional Labour Offices who reached out to local NGOs and social economy organisations that work with the target groups. Participation in the interviews was voluntary. While respondents could leave the meeting at any time or refuse to answer questions without consequences, all respondents successfully completed their participation in the interviews. The moderator conducted the analysis immediately after the focus groups. Participants received participation fees and reimbursement of travel expenses.
Each interview lasted about two hours and was conducted based on a script tailored to the specific characteristics of each group. The interview scripts addressed issues relevant to the study's objectives, particularly the respondents' work experiences, their current employment situation, and plans related to seeking employment. The interviews were conducted as informal conversations, allowing the utilisation of effects related to the characteristics of the social group, such as:
Safety effect: Creating a sense of security that allows participants to reveal genuine opinions;
Snowball effect: Deepening of a topic through discussions that build on other participants' comments;
Spontaneity effect: A natural and spontaneous discussion;
Stimulation effect: Eliciting greater motivation and enthusiasm;
Synergy effect: Generation of ideas by the group as a whole.
References
[19] Adams-Prassl, A. et al. (2022), Work that can be done from home: evidence on variation within and across occupations and industries, Labour Economics, Volume 74, https://doi.org/10.1016/j.labeco.2021.102083.
[16] Ameri, M. et al. (2022), Telework during the Pandemic: Patterns, Challenges and Opportunities for People with Disabilities, Institute of Labor Economics, https://doi.org/10.1016/j.dhjo.2022.101406.
[25] Baszczak, Ł. et al. (2021), Dyskryminacja ze względu na wiek na polskim rynku pracy – diagnoza [Age discrimination in the Polish labor market – a diagnosis], Polski Instytut Ekonomiczny, Warszawa, https://pie.net.pl/wp-content/uploads/2022/02/PIE-Raport_Ageizm.pdf.
[22] Bień, I. (2024), Aging labour market in Poland, Bureau of Research Chancellery of the Sejm (BAS), https://orka.sejm.gov.pl/WydBAS.nsf/0/FA64C714F4962F46C1258AC40042C474/$file/Analizy%20BAS%20156.pdf.
[3] Chancellery of the Prime Minister (2024), Projekt ustawy o rynku pracy i służbach zatrudnienia [Draft Act on Labor Market and Employment Services], https://www.gov.pl/web/premier/projekt-ustawy-o-rynku-pracy-i-sluzbach-zatrudnienia (accessed on 8 August 2024).
[10] Chłoń-Domińczak, A. (2019), Impact of retirement age changes on the old-age pension take up in Poland after 1990, https://doi.org/10.32088/0000_10.
[8] Eurostat (2024), Participation rates of selected age groups in education at regional level, https://doi.org/10.2908/EDUC_UOE_ENRA14.
[28] Eurydice, European Commission (2024), Early childhood education and care, Poland, https://eurydice.eacea.ec.europa.eu/national-education-systems/poland/early-childhood-education-and-care.
[30] Frankowski, J. and J. Mazurkiewicz (2020), Województwo śląskie w punkcie zwrotnym transformacji [Silesia at a turning point in its transformation], IBS Research Report, https://www.slaskibiznes.pl/files/grafiki%20r%C3%B3%C5%BCne/biostat/IBS-Research-Report-02-2020.pdf.
[14] Giovanis, E. and O. Ozdamar (2019), Accommodating Employees with Disabilities: The Role of Flexible Employment Schemes in Europe, https://doi.org/10.2139/ssrn.3441925.
[32] Government of Poland (2008), Act of 19 December 2008 on Bridging Pensions, https://isap.sejm.gov.pl/isap.nsf/DocDetails.xsp?id=WDU20082371656.
[21] Government of Poland (1997), Act of 27 August 1997 on Vocational and Social Rehabilitation and Employment of Persons with Disabilities, https://isap.sejm.gov.pl/isap.nsf/DocDetails.xsp?id=wdu19971230776.
[26] Interreg Danube (2024), Comprehensive analysis of work-life balance factors for 55+ employees, https://interreg-danube.eu/storage/media/01J46PBDWW3X65M5C5C6AET5NR.pdf.
[2] Jakubowska, K. et al. (2024), Trends in EU Public Employment Services Capacity 2023, https://doi.org/10.2767/342841.
[7] Kukołowicz, P., P. Leszczyński and J. Lubasiński (2024), Konsekwencje zmian demograficznych dla podaży pracy w Polsce [Consequences of demographic changes for the labor supply in Poland], Polish Economic Institute, Warsaw.
[12] Magda, I. (2020), “Increasing Female Labour Force Participation in Poland”, Instytut Badań Strukturalnych, https://ibs.org.pl/app/uploads/2020/01/IBS-PP_01.2020_EN.pdf.
[4] Ministry of Family, Labour and Social Policy (2024), Projekt ustawy o rynku pracy i służbach zatrudnienia [Draft Act on the Labor Market and Employment Services], https://www.gov.pl/web/premier/projekt-ustawy-o-rynku-pracy-i-sluzbach-zatrudnienia (accessed on 24 January 2025).
[20] Myles, A. et al. (2004), “An introduction to decision tree modeling”, Journal of Chemometrics, Vol. 18/6, pp. 275-285, https://doi.org/10.1002/cem.873.
[5] OECD (2025), The Role of Public Employment Services in Promoting an Inclusive Nordic Common Labour Market, Connecting People with Jobs, OECD Publishing, Paris, https://doi.org/10.1787/54cec4d6-en.
[31] OECD (2022), Reaching Out and Activating Inactive and Unemployed Persons in Bulgaria, Connecting People with Jobs, OECD Publishing, Paris, https://doi.org/10.1787/7b91154a-en.
[1] OECD (2021), Regional Economic Inactivity Trends in Poland, OECD Reviews on Local Job Creation, OECD Publishing, Paris, https://doi.org/10.1787/483a6a25-en.
[11] OECD Statistics (2024), Pensions at glance.
[17] Özgüzel, C., D. Luca and Z. Wei (2023), “The new geography of remote jobs? Evidence from Europe”, OECD Regional Development Papers, No. 57, OECD Publishing, Paris, https://doi.org/10.1787/29f94cd0-en.
[18] Radziukiewicz, M. (2021), Remote work in Poland and its perspectives, Economic and Regional Studies, 14(4), 409-427, https://doi.org/10.2478/ers-2021-0029.
[27] Reine, I. and O. Rajevska (2024), “Enhancing Labor Market Participation of Older Adults: Health and Workplace Adaptations”, European Journal of Public Health, Vol. 34/3, https://doi.org/10.1093/eurpub/ckae144.2176.
[9] Ruzik-Sierdzińska, A. (2018), “An Attempt to Identify Factors Influencing Retirement Decisions in Poland.”, Acta Universitatis Lodziensis. Folia Oeconomica, Vol. 4(336), pp. 43-59, https://doi.org/10.18778/0208-6018.336.03.
[13] Sewdas, R. et al. (2017), Why older workers work beyond the retirement age: a qualitative study, BMC Public Health, https://doi.org/10.1186/s12889-017-4675-z.
[29] Statistics Poland (2024), Education in the 2023/2024 school year, https://stat.gov.pl/en/topics/education/education/education-in-the-school-year-20232024-preliminary-data,13,2.html.
[6] Statistics Poland (2024), The demand for labour in 2023, publication tables.
[15] Taylor, H. et al. (2022), The changing workplace: Enabling disability-inclusive hybrid working, Work Foundation, Lancaster University, https://www.lancaster.ac.uk/media/lancaster-university/content-assets/documents/lums/work-foundation/TheChangingWorkplace.pdf.
[24] The Social Insurance Institution (2024), Struktura wysokości świadczeń wypłacanych przez ZUS w marcu 2024 [The structure of benefit amounts paid by ZUS after the March 2024 indexation], The Social Insurance Institution, Warsaw, https://www.zus.pl/documents/10182/39637/Struktura+wysoko%C5%9Bci+%C5%9Bwiadcze%C5%84+wyp%C5%82acanych+przez+ZUS+po+waloryzacji+w+marcu+2024+r..pdf/2cf8f296-1f4f-0ea0-ab73-ed395fba478b?t=1715935406058.
[23] The Social Insurance Institution (2023), Struktura wysokości świadczeń wypłacanych przez ZUS po waloryzacji w marcu 2023 roku [Structure of the benefit amounts paid by ZUS after the valorization in March 2023], https://www.zus.pl/documents/10182/39637/Struktura+wysoko%C5%9Bci+%C5%9Bwiadcze%C5%84+wyp%C5%82acanych+przez+ZUS+po+waloryzacji+w+marcu+2024+r..pdf/2cf8f296-1f4f-0ea0-ab73-ed395fba478b?t=1715935406058.
← 1. OECD Household indicators dashboard – country view. Measure: Real gross domestic product per capita growth rate. Chain linked volume (rebased), calendar and seasonally adjusted.
← 2. This projection is based on current employment rates, assuming that workers will retire at the statutory retirement age (60 for women and 65 for men) and that the employment rate, including for individuals aged 15-26, will remain stable until 2035 (Kukołowicz, Leszczyński and Lubasiński, 2024[7]).
← 3. The comparison excludes the OECD countries Columbia, Estonia, Lithuania, Luxembourg and Portugal due to either too small number of regions for comparison or lack of data. OECD calculations based on OECD Statistics, Labour indicators rates – Regions.
← 4. All numbers presented here are calculations based on EU-SILC data.
← 5. Based on the Act of December 19, (2008[32]), on Bridging Pensions, in Poland to receive a bridging pension one must:
have completed a period of at least 15 years in employment involving special conditions or of a special nature,
have reached at least the age of 55 (women) and 60 (men),
have completed contributory and non-contributory periods of at least 20 years for women and 25 years for men,
have performed work in special conditions or of special character after 2008.