Non-standard forms of work put additional pressure on social protection financing and on its redistributive mechanisms. The self-employed and part-time workers generally pay lower social security contributions than other workers. As this often results in lower entitlements to earnings-related benefits, they may rely on support financed from general revenues when their income falls. This chapter reviews the social protection coverage of the self-employed, the reasons for their low participation in voluntary schemes, their capacity to contribute, and recent and planned reforms to extend coverage. It also looks at the incidence of platform work. The chapter then reviews trends in part-time employment and their implications for social protection financing.
Financing of Social Protection
Abstract
The self-employed and part-time workers generally pay lower social security contributions than other workers often resulting in lower entitlements. They may therefore rely on general-revenue funded support if their income falls. For the self-employed, lower contributions reflect more limited access to social protection, low take-up of voluntary schemes where they exist, and, in some cases, underreporting of earnings (see also Chapter 3). Their incomplete social protection coverage also enables firms to engage in regulatory arbitrage by choosing work arrangements with fewer social protection entitlements, and thus lower social security contributions. For workers, choosing non-standard employment is especially attractive if contributions are lower than for standard dependent work, while entitlements are broadly similar (e.g. subsidised schemes). Aligning social protection coverage across employment forms can help decrease the incentive for regulatory arbitrage.
Similarly, part-time work negatively impacts the aggregate wage sum as it contracts hours worked, already diminished by a shrinking labour force due to below-replacement-level fertility rates, and therefore puts pressure on social protection financing (OECD, 2024[1]). Total hours worked per worker are on a downward trend across EU and OECD countries. Between 1995 and 2024, average annual hours worked by worker decreased by 6.6% across the EU and by 4.8% across the OECD on average. As most social security contributions are levied on income irrespective of working hours, a rising incidence of part-time work can also undermine the capacity to pay principle that is implicit in many earnings-related social insurance schemes that include an element of redistribution.
About one in seven (14%) of all workers are self-employed in the EU on average, slightly fewer than across the OECD (16%, (OECD, 2026[2])). Cross-country variation is wide, with around one in five working-age people active in the labour market being self-employed in Italy and Poland, compared to 5% in Norway, and 8% in Denmark (Figure 2.1). The share of self-employment has been on a downward trajectory since the second half of the 20th century, primarily driven by the decline in the share of the labour force working in agriculture, as well as small craft and retail businesses (OECD, 2024[1]). The share of self-employed workers in total employment decreased by 4.1 ppts across the OECD, and by 4.5 ppts across the EU on average since 1996, with particularly large declines in Romania (-28 ppts), Republic of Türkiye (-23 ppts) and Portugal (-17 ppts), see Annex Figure A A.1.
While the self-employed account for 14% of total employment, social contributions by the self-employed only account for 4% of all contributions across the EU on average, see Chapter 1. This is due to i) the more limited formal coverage by social protection and ii) an overrepresentation at the low and top end of the earnings distribution. This section discusses these two topics, concentrating on formal social protection coverage for the three social protection functions that have been selected as focus areas in this report: family, unemployment and old age benefits.
2.2.1. The self-employed are less likely to be statutorily covered by social protection
Coverage gaps for the self-employed tend to emerge in benefit programmes that are mainly contribution financed, as general-revenue financed benefits typically do not condition on prior contributions. As a result, family benefits, three‑quarters of which are financed by general revenue across the EU on average (see Chapter 1), are the social protection branch for which the self-employed are best covered: they receive child benefits in the same way as dependent employees in all countries. However, in some countries, they do not receive maternity income‑replacement benefits, particularly where these benefits are tied to, or part of, the sickness income‑replacement insurance. For instance, in Germany, self-employed workers do not have access to maternity cash benefits unless they have opted to pay for income replacement in the case of sickness scheme. This also applies to most self-employed in Austria. But self-employed mothers and fathers can still receive parental leave benefits in the same way as employees. In Poland and Romania, the self-employed have voluntary coverage, i.e. may opt in to maternity benefits (SPC and EC, forthcoming[3]).
Coverage gaps are largest for unemployment benefits, and other types of insurance closely tied to the workplace, such as work accident insurance (SPC and EC, forthcoming[3]). Even where self-employed workers are enrolled in mandatory schemes, compliance is typically weaker than for dependent employees (OECD, 2018[4]). Only ten EU countries provided the same access to unemployment benefits to the self-employed as to employees, and six countries provide no access. Across 38 OECD countries, only nine provide full unemployment benefits to self-employed workers, while 17 provide no access. Six countries offer the self-employed the option to enrol in unemployment insurance voluntarily, although enrolment rates are typically low (see below), and six countries provided partial access, e.g. lower benefit amounts or longer contribution periods than for employees (Figure 2.1).
Countries where the self-employed cannot access unemployment benefits often offer means-tested benefits for those without entitlement to insurance benefits or other resources (such as unemployment or social assistance benefits), including for those who have lost self-employed income. These safety-net benefits are typically less generous and less accessible than unemployment insurance benefits because of income‑ and asset tests. Payments are also not balanced by contributions. Excluding the self-employed from contribution-based schemes can therefore increase spending on means-tested benefits and create a degree of free‑riding.1
A lack of coverage affects more workers in countries with a high incidence of self-employment, e.g. Italy, where more than one in five workers is self-employed, than in Norway, where the share of self-employed is very low. In general, countries with a high incidence of self-employment do not seem to offer stronger coverage (Figure 2.1).
Some countries provide partial access to unemployment insurance, e.g. in France, there is a separate scheme for self-employed workers, the Allocation des Travailleurs Indépendants that is general-revenue financed and provides a lower maximum benefit receipt duration as well as lower benefit amounts than for employees.
Providing unemployment benefits to the self-employed can be difficult due to moral hazard considerations, as the self-employed control the success of their business in ways employees do not (OECD, 2024[1]), and it is difficult to establish unemployment without an employer to confirm a layoff (OECD, 2023[5]). But there are also practical difficulties in calculating contributions and benefits – the self-employed have fluctuating earnings as they may be paid by clients for larger chunks of work at irregular intervals. They also often have some control of the timing of their payments and therefore potentially more scope to manipulate their income to maximise their benefit entitlements (OECD, 2018[4]). Similarly, providing work accident insurance to the self-employed is associated with moral hazard since they control their own work environment.
Statutory access to unemployment benefits for self-employed workers, 2025 or latest available year, and incidence of self-employment, 2024
Note: *No data on the incidence of self-employment for Cyprus and Malta. Data for Bulgaria, Chile, Estonia, Iceland, Lithuania, Norway, Romania, the Slovak Republic, the United Kingdom and the United States refer to 2024. Partial access: more stringent entitlement criteria (e.g. longer contribution periods or higher minimum earnings thresholds) and/or less generous benefits (amounts and maximum duration) and/or limited coverage (some types of self-employed not covered). Some countries provide unemployment benefits for selected types of self-employed workers, but not most self-employed, e.g. Italy for some types of para-subordinate workers. Voluntary: in some countries with voluntary participation, self-employed workers who opt into the insurance scheme may also choose their contributions levels.
Source: Incidence of self-employment: (OECD, 2026[2]), “Labour Force Statistics: Summary tables”, OECD data explorer, (https://data-explorer.oecd.org). Statutory access to unemployment benefits: European Union and EFTA countries: (SPC and EC, forthcoming[3]), MISSOC: Comparative tables Unemployment, 2025‑2007‑01, https://www.missoc.org. Non-European/non-EFTA countries: (ISSA, 2026[6]).
These moral hazard considerations are largely absent in the areas of family benefits and pensions. Statutory coverage for old-age pensions is more robust than for unemployment benefits: the self-employed have to contribute to earnings-related pensions schemes in all countries except Denmark and Germany.2
However, they only pay the same contributions as the equivalent sum of employer and employee contributions in Czechia, Estonia, Finland, Hungary, Lithuania, Luxembourg, Portugal and Slovenia. In all other countries, they either have a reduced contribution rate or flat rate contributions or only contribute to the basic pension system. For instance, in Poland, self-employed workers only have to pay contributions on the minimum contribution base (60% of the average wage), although they may increase their contributions voluntary. As a result, pensions of self-employed workers are on average 22% lower than those of dependent workers across the OECD on average. Some countries, including Austria and Costa Rica, also explicitly top up lower contribution rates by self-employed workers (OECD, 2025[7]).
The reason for lower pension coverage is likely the double‑contribution issue (the self-employed would have to pay nominally higher contributions since they do not have an employer), paired with their often low incomes (see below). Historically, the self-employed were also more likely to be able to liquidate their business upon retirement (e.g. farmers or certain craftspeople), but business assets are less substantial among the self-employed that operate in the service sector. The reason for lower pension coverage is likely the double-contribution issue (the self-employed would have to pay nominally higher contributions since they do not have an employer), paired with their often low incomes (see below). Historically, the self-employed were also more likely to be able to liquidate their business upon retirement (e.g. farmers or certain craftspeople), but business assets are less substantial among the self-employed that operate in the service sector.
2.2.2. Participation in voluntary schemes tends to be low
Many countries do offer voluntary schemes for self-employed workers if they are not covered in a compulsory scheme, especially for the risks of sickness (cash benefits) and maternity, but also unemployment, old-age and work accidents. However, take-up (the share of self-employed workers opting to participate) in these schemes tends to be low: for instance, in 2024, in Austria, only 0.6% of workers opted in to the unemployment insurance scheme for the self-employed workers, and in Slovakia 2%. In Finland the take-up rate was 10-15% in 2023, and in Slovakia 2%. In Germany, fewer than 10% of all self-employed who were not already compulsorily covered by law opted to participate in the old-age pension scheme in 2023, even though this scheme benefits from public subsidies, while take-up for supplementary pensions was significantly higher at 57% in Belgium in 2022. Take-up is higher in Spain, where 49% of the self-employed in agriculture participate in the voluntary UI scheme in 2024 (SPC and EC, forthcoming[3]; SPC and EC, 2024[8]; MISSOC, 2025[9]).
One reason for low participation in voluntary benefits is low willingness to pay more for social protection across the board, but especially for unemployment benefits. The OECD Risks That Matter (RTM) survey shows that across 27 participating OECD countries on average, only 19% of all self-employed respondents would be willing to pay an additional 2% of their income for better provision of/better access to unemployment supports, even though only 32% believe that they would receive adequate public support if they were to become unemployed (Figure 2.2, Panel A). Willingness to pay more for unemployment supports is lowest in Belgium, France, and Ireland with under 10% willingness to pay more, increasing up to 30% in Türkiye and Slovenia (Figure 2.2, Panel A).
One reason for low participation in voluntary benefits is low willingness to pay more for social protection across the board, but especially for unemployment benefits. The OECD Risks That Matter (RTM) survey shows that across 27 participating OECD countries, on average, only 19% of all self-employed respondents would be willing to pay an additional 2% of their income for better provision of / better access to unemployment supports, even though only 32% believe that they would receive adequate public support if they were to become unemployed (Figure 2.2, Panel A). Willingness to pay more for unemployment supports is lowest in Belgium, France, and Ireland with under 10% willingness to pay more, increasing up to 30% in Türkiye and Slovenia (Figure 2.2, Panel A).
The willingness to pay more for old-age supports is higher, at 33% on average among RTM participating countries, in-line with the fact that some self-employed have a low unemployment risk (or assess their own risk to be low), whereas most expect to reach retirement age. However, willingness to pay is still low, considering that 29% of self-employed respondents across RTM‑27 countries do not expect adequate benefits in old-age (Figure 2.2, Panel B). Here, willingness to pay is highest in Latvia at almost 50% of self-employed respondents, where 38% of respondents expect adequate old-age benefits, as well as Israel and Slovenia with about 45% of respondents (and a significantly lower satisfaction with income support in old-age, see Figure 2.2, Panel B). Willingness to pay more for social protection, as well as satisfaction with current provision, is broadly similar between employees and the self-employed on the cross-country average and in most countries (see Figure A A.2 for employees).
Myopia (i.e. short-sighted decision making), and inefficient under-saving for retirement because of other behavioural distortions, was one of the main reasons why compulsory social security schemes were initially introduced for dependent employees (e.g. (Diamond, 1977[10])). In the absence of significant business assets, this would also apply to the self-employed.
Voluntary schemes can also struggle with adverse selection problems, where workers more likely to be affected by a risk are more likely to opt in (e.g. those with poorer health are more likely to opt into sickness benefits, or those planning a family more likely to opt into maternity or paternity benefits), which leads to incomplete risk pooling, and the need for higher public subsidies, see Box 2.1.
Voluntary social insurance schemes risk adverse selection of participants: if premiums are uniform, those with the highest risk have the biggest incentive to join. For schemes that do not rely on general-revenue subsidies, this can lead to a vicious circle of contribution hikes and lower-risk members leaving while subsidised schemes may require higher-than-expected inflows of funds. Compulsory insurance limits this by mandating low-risk individuals to participate.
Country examples of adverse selection include:
In 2014, a health insurance reform in the United States limited private insurers’ ability to adjust premiums to health status and medical history. Exploiting geographical differences across counties in Colorado, (Panhans, 2019[11]) shows that, for each USD 1 increase of the insurance premium of an average plan, the average medical expenditures of the insured population increased by USD 0.88 – higher premiums made the insurance pool more expensive, implying that those who expected higher medical costs were more likely to remain insured. Furthermore, for every USD 100 increase in the insurance premium, the likelihood that an insurance holder had a chronic health condition increased by 2 ppts.
The Swedish unemployment insurance system consists of a basic unemployment benefit, that is supplemented by voluntary unemployment insurance funds affiliated with trade unions (OECD, 2018[4]). Landais et al. (2021[12]) show that workers who join an unemployment insurance fund are 2.3 times more likely to become unemployed than workers who do not, with under half of this effect due to adverse selection (workers having private knowledge of their higher unemployment risk) and about half due to moral hazard (workers who have the insurance being more likely to become unemployed). In 2007‑8, a reform linked employee contributions to unemployment risk and raised average premiums by 300%. Membership in the Unemployment Insurance Funds dropped by around 10 p.p. in the following years. Groups that were particularly likely to exit the system included older workers over the age of 60, who have the lowest unemployment risk of all age groups, and young workers under the age of 25, who typically have low earnings and short unemployment durations (OECD, 2018[4]).
In 1998, the Netherlands introduced a new disability insurance scheme that included experience rating for firms: firms whose employees were more likely to become disabled were charged higher premiums, to increase their incentive to improve preventive measures and re‑integration efforts. Employers had the option to opt out of the public scheme and either buy private insurance or carry any disability benefit costs themselves. Benefit levels and entitlement criteria were independent of the employer choice. (Deelen, 2005[13]) shows that firms with a higher share of older and female employees were significantly less likely to opt out of the public insurance, and that opt-out decisions significantly varied by sector.
Self‑employed workers in Austria can opt into a short‑term sickness benefit programme. In 2016, close to half of those who were covered received a benefit. The average benefit duration was 22 days, nearly twice the average sick‑leave duration among dependent employees, who are subject to mandatory insurance, highlighting moral hazard risks. In response to the resulting deficits in the programme, the minimum benefit amount was cut significantly in 2017 (OECD, 2018[4]).
Note: Data refer to the share of respondents who stated that they would be willing to pay an additional 2% of income in taxes/social contributions to benefit from better provision of and access to: Unemployment supports/Old age pensions as well as the share of respondents who agree or strongly agree with the statement “I think that the government does/would provide my household and me with adequate income support in the case of income loss due to unemployment/retirement.” Data refers to self-employed (working-age) respondents. RTM‑27 is an unweighted average of the 27 countries participating in the RTM survey. Countries are ordered by increasing willingness to pay for supports.
* The difference between the self-employed and employees is statistically significant at p < 0.05. For Panel A, this is the case in Ireland for the willingness to pay more and for willingness to pay more for unemployment supports, and for Estonia, Latvia, Lithuania, Korea, and Slovenia for the satisfaction with income support in the case of unemployment. For Panel B, this is the case for Lithuania and Mexico for the willingness to pay more for old-age pensions, and for Denmark, Estonia, Ireland, Latvia, Türkiye, and Slovenia for the satisfaction income support in retirement.
Source: Data refer to pooled waves for 2022 and 2024 of the OECD Risks that Matter Survey (https://www.oecd.org/en/about/programmes/oecd-risks-that-matter-rtm-survey.html).
2.2.3. The self-employed are overrepresented at the bottom earnings distribution, which partly explains low contributions
Another reason for the lower-than-expected share of contributions from the self-employed is that the income distributions of dependent workers and the self-employed are different, with the self-employed being overrepresented at the bottom and top end of the earnings distribution. Across the EU on average, the self-employed are more than twice as likely to be in the bottom 10% of the earnings distribution and are still overrepresented in deciles two and three – even if contributions were equivalised between self-employed and dependent employees, they would therefore pay less in contributions. They are slightly overrepresented among the 10% with the highest earnings too, which can limit contributions in countries with a contribution ceiling (Figure 2.3).
Similar to gender earnings patterns for employees, self-employed women are more concentrated at the bottom end of the income distribution: women account for 35% of all self-employed across the EU on average, but for half of all self-employed in the bottom decile of the income distribution, compared to 27% of the self-employed in the top decile (Figure A A.3).
Cross-country differences in the incidence of self-employment across the earnings distribution are quite significant: The self-employed are more concentrated in the middle of the income distribution than at its extremes only in Belgium. In Hungary, the self-employed are particularly overrepresented among low earners: they account for over 60% of the bottom 20% of earners, but only for 22% of all earners. Overrepresentation at the bottom is also more accentuated than on average in Denmark, Greece Croatia, Lithuania and Poland (see Overrepresentation at the bottom is also more accentuated than on average in Denmark, Greece Croatia, Lithuania and Poland (see Figure A A.4).
These results are under the caveat that income data for the self-employed is more unreliable than that of employees (Zardo Trindade and Goedemé, 2020[14]). The self-employed have a clear incentive to underreport income to tax- and social security institutions, which can lower their income in surveys that rely on administrative data for incomes. But they may also underreport earnings in questionnaire responses because they fear that their incomes will be transmitted to the tax authorities. Recall errors in surveys may also be more important for the self-employed who often have irregular income. Using matched data from the tax authorities and a household survey, (Cabral, Gemmell and Alinaghi, 2020[15]) estimate for New Zealand that the self-employed underreport about 20% of their earnings to the tax authorities, but only about 6‑10% to household surveys, making survey-based income more reliable. In contrast, (Consolini and Donatiello, 2013[16]) show for Italy, that self-employment income in tax data is higher than self-reported income from the EU-SILC.
The self-employed as a share of all workers, by wage / earnings decile, in per cent, EU-average, 2024
Note: The black line indicates the overall incidence of self-employment across the EU on average (13%). Status as self-employed or employee is self-identified. For workers with missing information, employment type is assigned as the type with the higher income (employee or self-employed earnings). Unweighted average.
Source: Secretariat calculation based on the EU-SILC 2024.
2.2.4. Recent and planned policy reforms to expand coverage for the self-employed
Several countries are implementing or considering reforms to improve the social protection coverage of the self-employed. This trend picked up in the wake of the COVID‑19 pandemic, that showed that the self-employed need income support, at least during sudden crises. Emergency programmes designed for immediate relief often suffered from incomplete coverage, delayed payments and design flaws leading to overpayments. They were also necessarily not balanced by contributions (Dely, Hyee and Prinz, 2025[17]).
Incomplete social protection coverage of the self-employed also enables firms to engage in regulatory arbitrage by choosing work arrangements with fewer social protection entitlements, and thus lower social security contributions. For workers, choosing non-standard employment is especially attractive if contributions are lower than for standard dependent work, while entitlements are broadly similar (e.g. subsidised schemes). In addition to the enforcement of the correct classification of workers (through labour inspections and the judicial system), aligning social protection coverage across employment forms can help decrease the incentive for regulatory arbitrage. Trade unions can also play a role in supporting misclassified workers to assert their re‑classification through the judicial process (OECD, 2019[18]).
Voluntary schemes are not generally the best way to achieve broad coverage of self-employed workers, as they tend to suffer from low take-up and adverse selection (those who have the highest risk have the greatest incentive to join), which leads to incomplete risk pooling, see Section 2.2.2.
This section reviews current reform efforts to expand coverage for the self-employed, based on answers to the short OECD policy questionnaire “financing of national social protection systems”, circulated to EU, OECD and OECD accession countries, in September 2025, as well as the MISSOC comparative tables (https://www.missoc.org/missoc-database/comparative-tables/), unless otherwise stated.
Ireland introduced a new earnings-related unemployment benefit, “Jobseeker’s Pay-Related Benefit” in March 2025. For the self-employed, there is a separate benefit, “Jobseeker’s Benefit (Self-Employed)” that also provides benefits tied to previous earnings, for the same duration, but lower amounts. However, self-employed with an annual income below EUR 5 000 may voluntarily opt in to the unemployment benefit, whereas there is no insurance for employees with an income below EUR 1 800.
The Netherlands are considering introducing a compulsory invalidity insurance for the self-employed (who can currently opt in voluntarily), funded by social contributions.
The Czech Republic is increasing the contribution base for the pension insurance for the self-employed gradually between 2024 and 2026 to improve future pension adequacy.
Finland has launched a project to design a “combination insurance model” that is supposed to improve access to unemployment benefits for workers who combine dependent and self-employment.
Finland also reformed pension contributions for self-employed workers, with a view to improving their pension adequacy. Currently, the contribution base for the pension insurance for the self-employed in Finland is an estimate of how much the self-employed worker would earn if they performed the same work as an employee (based on the sector’s median earnings, their working hours and their skills) (Rantala, 2025[19]). This is reviewed by insurance companies every three years. For employees, the contribution base is taxable income. From 2028 onwards, the self-employed will be able to choose whether to base their contributions on the imputed income, or on their actual taxable income. After a transition period, the insured income would have to be at least 50% of actual earned income. The reform will also enable the self-employed to lower their pension contributions if their actual income is below their imputed income. A pension evaluation commissioned by the government also suggested lowering the threshold under which no contributions are due (around EUR 9 400 per year in 2026, compared to around EUR 860 for employees), and to abolish the minimum requirement of four months’ income for compulsory insurance, which does not exist for employees either.
The Slovak Republic increased the minimum assessment base for the pension insurance for the self-employed with an annual income above EUR 8 580 from 50% to 60% of the average wage of two years ago (EUR 914 in 2026). For employees, no such minimum contribution base exists. At the same time, the Slovak Republic also increased the health-insurance contribution rate for the self-employed from 15 to 16%.
Romania significantly increased the ceilings for the contribution base of the self-employed to the health insurance (from 12 times the minimum wage in 2025 to 72 times the minimum wage) in 2026. There is no ceiling for employees.
Italy introduced a new unemployment benefit for para-subordinate professionals (unlicenced freelancers, e.g. graphics designers) on an experimental basis from 2021 to 2023, then made it permanent in 2024 (European Commission, 2025[20]). Freelancers whose income drops by more than 50% compared to their average income over the last three years, and whose annual income is below a threshold, are entitled to 25% of their average income over the last three years. This benefit could only be claimed once every three years and is financed by social security contributions. Recipients have to participate in training courses offered by the PES related to their professional field. Professional associations contribute to the design and choice of the courses (OECD, 2023[5]).
Cyprus extended parental leave and accidents-at-work coverage to the self-employed (European Commission, 2025[20])
Türkiye is planning to increase the coverage of social protection for non-standard forms of work, in particular for the self-employed, low-income and seasonal workers, including special programmes for low-income tradesmen, artisans, farmers and seasonal workers. Legislation is currently in development.
Belgium is considering increasing the minimum contribution rates for self-employed workers both in the standard as well as in the voluntary supplementary pension scheme for the self-employed.
2.2.5. Platform work is a small share of total employment
According to the EUROSTAT pilot survey on digital platform employment, covering 17 EU and EFTA countries,3 3% of working-age people reported to have worked for a digital labour platform at least one hour per month (Eurostat, 2023[21]). This estimate is somewhat lower than the share estimated by the Contingent Worker Supplement of the Current Population Survey (CPS) for the United States in 20174 (6‑7% of workers), but this estimate includes independent contractors and freelancers who do not find clients on digital platforms (Abraham et al., 2023[22]; Bernhardt et al., 2022[23]). In Canada, the share of working-age people who had earned money through a labour platform was 1.7% in 2023, and 0.4% of workers said that work through digital platforms was their main job (Hardy, 2024[24]). In Australia, the share of the population who undertook digital platform work during the last four weeks was estimated to be just under 1% in 2022‑2023 (Australian Bureau of Statistics, 2023[25]). While these measurements are not directly comparable (e.g. some measure platform work undertaken over the past month vs. year), and the measurement of platform work is not straightforward (e.g. there is no single definition of what should be included in platform work (OECD/ILO/European Union, 2023[26])), these measurements indicate that overall, the group of platform workers is small.
However, they seem especially vulnerable, because, like other self-employed, they have limited access to social protection, are not covered by employment protection legislation or minimum wages, and are typically not allowed to engage in collective bargaining, although exceptions exist in the EU for some self-employed, in particular economically dependent solo self-employed persons (European Commission, 2022[27]). Unlike other self-employed, they also operate with little or no business capital, and if they are dependent on a few or only one client, they also do not enjoy entrepreneurial freedoms, e.g. setting their own prices or working hours. From a social protection financing perspective, there is also the concern that firms might substitute employees for platform workers to benefit from lower labour cost (regulatory arbitrage, (OECD, 2018[4])), which can undermine social protection funding base.
In recent years, several court decisions have classified platform workers as employees as many jurisdictions have “primacy of facts” principles that define platform workers as employees if they meet minimum standards of dependence (OECD, 2024[1]). The 2024 EU Platform Work Directive introduces a legal presumption of employment when platforms exercise “direction and control” over workers, which should be integrated into national regulations by the end of 2026. Spain already legislated in 2021 with the “Rider”s law’ that classified platform workers as employees when platforms exercise direction and control (ETUC, 2022[28]; OECD, 2026[29]).
Mexico introduced a range of measures for digital platform workers in 2025, which should benefit up to 700 000 platform workers. Platform workers earning at least the minimum wage are covered by health insurance, maternity, disability and retirement, as well as childcare services, while those earning below the minimum wage are covered only for accidents at work. Platforms must provide information on workers who provide services through their platform to the social security agency and pay contributions (that is, for the purpose of social protection, they act as the employer). Similarly, in Estonia, digital platforms must provide income data on workers to the Tax and Customs Board (although they do not have to pay contributions). This transmission is expected to increase compliance with taxes and social contributions (SPC and EC, forthcoming[3]).
Belgium is planning to introduce an obligation for digital platforms to insure the self-employed who find work through their platform against workplace accidents using private insurance (SPC and EC, forthcoming[3]) in 2028.
Total hours worked per worker are on a downward trend across EU and OECD countries. Between 1995 and 2024, average annual hours worked by worker decreased by 6.6% across the EU and by 4.8% across the OECD on average (see Annex Figure A A.5). Decreases were particularly large (i.e. more than 15 p.p.) in Chile, Ireland and Iceland. Around a quarter of this reduction in hours worked occurred across the EU since 2019, which indicates that i) the decline in hours worked started before the COVID‑19 crisis, and ii) that this is a continuing trend. Evidence from Europe suggests that it may be driven by an increasing preference for leisure as incomes rise (OECD, 2024[1]).
The fall in hours worked per worker negatively impacts the aggregate wage sum as it contracts hours worked, already diminished by a shrinking labour force due to below-replacement-level fertility rates, and therefore puts pressure on social protection financing (see Chapter 3 for an estimation of the potential of increasing hours worked per worker to counteract the effects of population ageing). Part-time workers also accumulate lower entitlements to earnings-related benefits such as unemployment, sickness or old-age pensions, and the resulting benefits may be insufficient to lift them above the poverty line.
As most social security contributions are levied on income irrespective of working hours, a rising incidence of part-time work can also undermine the capacity to pay principle. Many earnings-related benefits contain an element of redistribution, with the implicit assumption that low earnings are connected to low earnings capacity (due to low wage rates, unpaid work, periods of unemployment etc.). For instance, mandatory pension schemes typically provide higher replacement rates for low earners (OECD, 2025[7]). A rising incidence of part-time work will increase the number of low earners and therefore put additional pressure on these redistributive mechanisms.
2.3.1. One in seven workers works part-time in the EU
Around 14% of employees worked part-time across the EU, and 13% across the OECD on average, in 2024.5 Women are more than twice as likely to work part-time across the OECD average (20% vs. 7.5% for men), and more than three times as likely across the EU on average (22% vs. 6.6% for men), connected to the uneven sharing of unpaid work between men and women (see, e.g. (OECD, 2024[1])). Cross-country differences are striking, with over 35% of employees working part-time in the Netherlands, and over 20% in Germany, Austria and Switzerland, while the part-time rate is below 5% in all central and eastern European countries as well as the Baltics and Portugal (Figure 2.4).
About 15% if all part-time workers across the OECD on average were involuntary part-time workers in 2025 (they would have liked to work more hours, but could not find a job offering them). Across EU countries this share is slightly higher at 17%. Underemployment has decreased strongly since the Great Financial Crisis (GFC): in 2014, 25% of part-time workers were involuntary part-time workers across the OECD, and 30% across the EU, on average (OECD, 2026[30]).
Part-time work can help workers who would struggle with full-time work to participate in the labour market. For instance, people with disability are twice as likely to work part-time across 32 OECD countries with available data that people without disability (OECD, 2022[31]).
Over the past three decades, women’s part-time rate declined across the EU and the OECD on average, against a backdrop of increasing female labour force participation. While the share of women working part-time increased slightly at the beginning of the 2000s up until shortly after the GFC, it has been on a downward trajectory since the beginning of the 2010s, dropping from 24% in 1995 to 22% in 2024 across the EU on average, and from 24% to 20% over that period across the OECD on average (OECD, 2024[1]; OECD, 2026[32]).
There is no clear link between rising women’s labour force participation and a higher incidence of part-time work. Some countries, notably Austria, Finland, or Italy, did experience rising women’s labour force participation (in the range of 10‑15 p.p.) combined with rising part-time employment among women (in the range of 9‑14 p.p., see (OECD, 2024[1]; OECD, 2026[32])). However, other countries experienced an equivalent increase in women’s labour force participation combined with a decrease in part-time work. Examples include Canada, Switzerland, France, Mexico, or Israel, where labour force participation increased by 10‑15 p.p., while the share of women working part-time decreased by around 4‑7 p.p. The decrease in part-time work was driven by prime‑aged (25‑54) and older women (55‑64), who show increasingly robust late‑career labour force attachment. In contrast, part-time rates for young women (15‑24) increased even more strongly than for young men, likely driven by higher enrolment in post-secondary education, that has now caught up to, and in places surpassed, male enrolment.
Men, in contrast, increasingly work part-time. The share of men who work part time increased from 4.2% in 1995 to 6.6% in 2024 across the EU on average, and from 6% to 7.5% across the OECD on average. Increases were particularly strong in the Netherlands (8 p.p.), Austria and Germany (6 p.p.), Finland and Norway and Spain (5 p.p.), and Belgium and Denmark (4 p.p.). The part-time rate for men rose in all countries except the United States, Sweden and Greece.
It is not clear from the available data what drives the increasing propensity for men to work part-time. Labour market slack seems to have played a part at the beginning: men’s part-time rates accelerated with the Great Financial Crisis (GFC), in line with indicators of involuntary part-time work. However, while underemployment fell with the economic recovery, men’s part-time rate did not return to pre‑GFC levels. Men may have curtailed work to participate in unpaid care, but this is difficult to assess with available data. Workers may also increasingly value leisure as wages increase. This is consistent with part-time work increasing most in higher-income countries (see (OECD, 2024[1]) for an in-depth discussion).
Should the trend towards part-time work among men and young women continue, it could depress aggregate labour supply and the wage sum, affecting the financial stability of social protection systems. Part-time work exacerbates the squeeze that population ageing puts on labour earnings, the main funding base of social protection systems. The sustainability of pay-as-you-go pension systems in particular hinges on productivity growth compensating for negative population growth even if labour supply is constant (Aaron, 1966[33]). This means that even higher productivity growth would be necessary to preserve social protection funding if the incidence of part-time work increases in the long term.
Part-time employment rate by sex, 15‑64 years, 2024
Note: Part-time in this figure is defined as dependent employees who usually work less than 30 hours per week in their main job. Korea does not have data on dependent employment by full- and part-time status, no data on Japan.
Source: (OECD, 2026[32]), Incidence of full-time and part-time employment based on OECD-harmonised definition (indicator), https://data-explorer.oecd.org/s/438.
2.3.2. Part-time workers sometimes do not fully contribute to social protection
Part-time workers are typically statutorily covered by social protection in the same way was full-time workers, but there are important exceptions.
In Austria, workers earning below the “marginal threshold” for employee contributions (EUR 551 per month in 2025) are not covered by health insurance or pension insurance, and therefore not liable for social contributions (they are covered for accidents at work). Workers with marginal earnings can opt into health and pensions insurance voluntarily, but often, workers with marginal earnings already have health insurance, e.g. because they are students or pensioners. Recent and planned reforms are chipping away at this advantage: In 2024, workers who have marginal employment contracts with two or more employers were included in the compulsory unemployment insurance system following a constitutional court decision (SPC and EC, forthcoming[3]). Austria is also planning to introduce mandatory health insurance contributions for those with marginal earnings during the current government term (2025-2029).
Similarly, in Germany, “mini jobs” with a monthly income up to EUR 603 per month in 2026 are not subject to employee social security contributions except for pensions (they may however apply for an exemption from pension insurance payments), and employers pay a flat rate. “Midi-jobs”, with an income between the “mini” threshold and EUR 2 000 per month benefit from reduced social security contributions, while their contributions to pension insurance build full entitlements. Similar contribution-free lower thresholds exist in the United Kingdom, Ireland and Canada (OECD, 2026[34]).
Some countries offer rebates in social security contributions for low-income workers, but these can be adjusted by working time to improve targeting to low-wage workers, instead of workers working fewer hours. For instance, in France, employers can reduce employer social security contributions for workers earning less than 1.6 times the minimum wage. For part-time workers, this reduction is adjusted by hours worked (OECD, 2026[34]). Similarly, Belgium calculates the potential monthly income for part-time workers using their salary and stated hours of work to determine whether they are entitled to a social security contribution rebate for low-income workers (Harding, Paturot and Simon, 2022[35]).
Japan limits social security coverage to full-time workers for some risks, but current reforms aim to extend coverage. For instance, only workers who work at least 20 hours per week and at least 31 days per month are currently covered by unemployment benefits. Starting in October 2028, employment insurance coverage will be extended to workers working at least 10 hours per week. From 2026, Japan will also extend pension insurance coverage to part-time employees working in smaller firms (up to 50 employees).
Malta has also extended social protection coverage for part-time workers by allowing those with multiple jobs to pay social security contributions on their combined wage. Broadening the contribution base will improve the accumulation of pension rights.
Improving the coverage of non-standard workers is not only important for improving the financing base of social protection. Many non-standard workers likely lack the capacity to accumulate savings to cushion potential earnings shocks. Under the caveat that earnings from self-employment are less reliable than employee earnings (see Section 2.2),6 across countries with available data, part-time workers are 1.8 times as likely to be at-risk of poverty7 than full-time workers, while the self-employed have a 2.1 times higher poverty risk (Figure 2.5). In all countries with available data, full-time employees have a poverty rate of under 15%, and the risk is under 8% across countries on average – full-time work is very protective against poverty. Poverty risks are very high for the self-employed in Poland (where the poverty risk for full-time workers is only 4%), Greece, Lithuania and Estonia with over 20% of self-employed workers.
Thus, incomes of other household members, or earnings-replacement benefits, do not on average compensate for lower incomes of non-standard workers, even though students and pensioners are more likely to work part-time, many women working part-time have a full-time working partner, and many self-employed with low incomes are topping up income from dependent work, including most platform workers (Eurostat, 2023[21]).
Share of non-standard workers at-risk of poverty, in per cent, 2024
Note: Poverty line: 60% of equivalised median household income (Eurostat definition). The standard OECD definition is 50% of equivalised median household income. Part-time work is self-defined. Employees with missing information on part-time work are assumed to work full-time. Status as self-employed or employee is self-identified. For workers with missing information, employment type is assigned as the type with the higher earnings (employee or self-employed earnings). Poverty rates by employment type cannot be shown for Croatia, Cyprus, Czechia, Finnland, Malta, Romania, Slovenia, and the Slovak Republic because the size of at least one employment-type/AROP cell is below the minimum of 30 observations suggested by Eurostat for statistical reliability. EU‑19: unweighted average of countries with available information.
Source: Secretariat calculation based on the EU-SILC 2024.
References
[33] Aaron, H. (1966), “The Social Insurance Paradox”, Canadian Journal of Economics and Political Science, Vol. 32/3, pp. 371-374, https://doi.org/10.2307/139995.
[22] Abraham, K. et al. (2023), The Independent Contractor Workforce: New Evidence On Its Size and Composition and Ways to Improve Its Measurement in Household Surveys, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w30997.
[25] Australian Bureau of Statistics (2023), “Digital platform workers in Australia”, https://www.abs.gov.au/articles/digital-platform-workers-australia.
[23] Bernhardt, A. et al. (2022), “Independent Contracting, Self-Employment, and Gig Work: Evidence from California Tax Data”, ILR Review, Vol. 76/2, pp. 357-386, https://doi.org/10.1177/00197939221130322.
[15] Cabral, A., N. Gemmell and N. Alinaghi (2020), “Are survey-based self-employment income underreporting estimates biased? New evidence from matched register and survey data”, International Tax and Public Finance, Vol. 28/2, pp. 284-322, https://doi.org/10.1007/s10797-020-09611-8.
[13] Deelen, A. (2005), “Adverse selection in disability insurance: empirical evidence from Dutch Firms”, CPB Discussion Paper Nr 46, https://www.cpb.nl/system/files/cpbmedia/publicaties/download/adverse-selection-disability-insurance-empirical-evidence-dutch-firms.pdf.
[17] Dely, B., R. Hyee and C. Prinz (2025), “What worked well in social protection during the COVID-19 pandemic?”, OECD Social, Employment and Migration Working Papers, No. 325, OECD Publishing, Paris, https://doi.org/10.1787/b31ac155-en.
[10] Diamond, P. (1977), “A framework for social security analysis”, Journal of Public Economics, Vol. 8/3, pp. 275-298, https://doi.org/10.1016/0047-2727(77)90002-0.
[28] ETUC (2022), Platform Reps: Spain: Country report 2022, European Trade Union Confederation, https://www.etuc.org/sites/default/files/page/file/2023-05/ETUC_country%20report_Spain-EN.pdf.
[20] European Commission (2025), Joint Employment Report 2025 – As adopted by the EPSCO Council on 10 March 2025, Publications Office of the European Union, https://doi.org/10.2767/9542551.
[27] European Commission (2022), Communication from the Commission. Guidelines on the application of Union competition law to collective agreements regarding the working conditions of solo self-employed persons, https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=intcom:C%282022%296846 (accessed on 18 August 2026).
[21] Eurostat (2023), “Employment statistics - digital platform workers”, Statistics explained, https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Employment_statistics_-_digital_platform_workers.
[35] Harding, M., D. Paturot and H. Simon (2022), “Taxation of part-time work in the OECD”, OECD Taxation Working Papers, No. 57, OECD Publishing, Paris, https://doi.org/10.1787/572b72d3-en.
[24] Hardy, V. (2024), “Defining and measuring the gig economy using survey data”, Labour Statistics: Research Papers, https://www150.statcan.gc.ca/n1/pub/75-004-m/75-004-m2024001-eng.htm.
[37] Immervoll, H. et al. (2022), “De-facto gaps in social protection for standard and non-standard workers: An approach for monitoring the accessibility and levels of income support”, OECD Social, Employment and Migration Working Papers, No. 271, OECD Publishing, Paris, https://doi.org/10.1787/48e282e7-en.
[6] ISSA (2026), Social Security Around the World. Country profiles, ISSA, https://www.issa.int/databases/country-profiles.
[16] Jäntti, M., V. Törmälehto and E. Marlier (eds.) (2013), Improvements of data quality through the combined use of survey and administrative sources and micro simulation model, Publications Office of the European Union, https://doi.org/10.2785/45854.
[38] Königs, S. and J. Terrero-Dávila (2025), “Who climbs the income ladder?: Cross-country evidence on income mobility from tax record data”, OECD Social, Employment and Migration Working Papers, No. 322, OECD Publishing, Paris, https://doi.org/10.1787/eadbbcbf-en.
[12] Landais, C. et al. (2021), “Risk-Based Selection in Unemployment Insurance: Evidence and Implications”, American Economic Review, Vol. 111/4, pp. 1315-1355, https://doi.org/10.1257/aer.20180820.
[9] MISSOC (2025), Missoc Comarative Tables: Financing, https://www.missoc.org/missoc-database/comparative-tables/.
[36] OECD (2026), Average annual hours actually worked per worker, https://data-viewer.oecd.org/?chartId=2326e6d8-c5c1-4be6-b6ef-980c4a52cf6b.
[32] OECD (2026), Incidence of full-time and part-time employment based on OECD-harmonized definition (indicator), https://data-explorer.oecd.org/s/438.
[30] OECD (n.d.), Incidence of involuntary part-time employment (indicator), https://data-explorer.oecd.org/.
[2] OECD (2026), “Labour Force Statistics: Summary tables”, OECD Employment and Labour Market Statistics (database), https://doi.org/10.1787/data-00286-en (accessed on 26 February 2026).
[29] OECD (2026), OECD Employment Outlook 2026: Geographic Disparities in Jobs and Incomes, OECD Publishing, Paris, https://doi.org/10.1787/7e710f54-en.
[34] OECD (2026), OECD TaxBEN policy descriptions, https://www.oecd.org/en/topics/sub-issues/income-support-redistribution-and-work-incentives/how-do-countries-calculate-tax-liabilities-and-social-benefit-entitlements.html.
[7] OECD (2025), Pensions at a Glance 2025: OECD and G20 Indicators, OECD Publishing, Paris, https://doi.org/10.1787/e40274c1-en.
[1] OECD (2024), Megatrends and the Future of Social Protection, OECD Publishing, Paris, https://doi.org/10.1787/6c9202e8-en.
[5] OECD (2023), Benefit Reforms for Inclusive Societies in the United States: Income Security During Joblessness, OECD Publishing, Paris, https://doi.org/10.1787/32d8f005-en.
[31] OECD (2022), Disability, Work and Inclusion: Mainstreaming in All Policies and Practices, OECD Publishing, Paris, https://doi.org/10.1787/1eaa5e9c-en.
[18] OECD (2019), OECD Employment Outlook 2019: The Future of Work, OECD Publishing, Paris, https://doi.org/10.1787/9ee00155-en.
[4] OECD (2018), The Future of Social Protection: What Works for Non-standard Workers?, OECD Publishing, Paris, https://doi.org/10.1787/9789264306943-en.
[26] OECD/ILO/European Union (2023), Handbook on Measuring Digital Platform Employment and Work, OECD Publishing, Paris, https://doi.org/10.1787/0ddcac3b-en.
[11] Panhans, M. (2019), “Adverse Selection in ACA Exchange Markets: Evidence from Colorado”, American Economic Journal: Applied Economics, Vol. 11/2, pp. 1-36, https://doi.org/10.1257/app.20170117.
[19] Rantala, J. (2025), Self‑employed persons also deserve decent pension security: Rapporteur’s report on developing the pension system for the self‑employed (Reports and Memorandums of the Ministry of Social Affairs and Health 2025:35)., Ministry of Social Affairs and Health (Finland), https://urn.fi/URN:ISBN:978-952-00-6000-8.
[8] SPC and EC (2024), Access to social protection for workers and the self-employed. (Partial) Update of the monitoring framework - 2023, https://ec.europa.eu/social/BlobServlet?docId=27549&langId=en.
[3] SPC and EC (forthcoming), Access to social protection for workers and the self-employed. Update of the monitoring framework 2025, European Commission.
[14] Zardo Trindade, L. and T. Goedemé (2020), The comparability of the EU-SILC income variables: review and recommendations. 2020 edition, Publications Office fo the European Union, https://doi.org/10.2785/047001.
Note: The earliest year available is 1998 for Latvia, Luxembourg and Switzerland; 1999 for Romania, 2000 for Bulgaria and Türkiye; 2001 for Colombia; 2002 for Croatia; 2005 for Lithuania; 2007 for Chile; 2010 for Costa Rica. The latest available year is 2018 for Latvia.
Source: Annual labour force survey, summary tables, via OECD data explorer, https://data-explorer.oecd.org/.
Note: Data refer to the share of respondents who stated that they would be willing to pay an additional 2% of income in taxes/social contributions to benefit from better provision of and access to: Unemployment supports/Old age pensions as well as the share of respondents who agree or strongly agree with the statement “I think that the government does/would provide my household and me with adequate income support in the case of income loss due to unemployment/retirement.” Data refers to employees workers (working-age) respondents. RTM‑27 is an unweighted average of the 27 countries participating in the RTM survey. Countries are ordered by increasing willingness to pay for supports. Countries mentioned by * mean that the difference between the self employed and the employees is statistically significant at p < 0.05. For Panel A, this is the case in Ireland for the willingness to pay more and for Estonia, Latvia, Lithuania, Korea, and Slovenia for the satisfaction with adequate income support. For Panel B, this is the case for Lithuania and Mexico for the willingness to pay more and for Denmark, Estonia, Ireland, Latvia, Türkiye, and Slovenia for the satisfaction with adequate income support.
Source: Data refer to pooled waves for 2022 and 2024 of the OECD Risks that Matter Survey (https://www.oecd.org/en/about/programmes/oecd-risks-that-matter-rtm-survey.html).
Breakdown of workers by dependent and self-employment and sex, by wage/earnings decile, in per cent, EU-average, 2024
Note: Status as self-employed or employee is self-identified. For workers with missing information, employment type is assigned as the type with the higher income (employee or self-employed earnings). Unweighted EU average.
Source: Secretariat calculation based on the EU-SILC 2024
Note: Status as self-employed or employee is self-identified. For workers with missing information, employment type is assigned as the type with the higher income (employee or self-employed earnings). To respect EUROSTAT recommendations on minimum observation count per cell, deciles have been combined to quintiles for the country-specific analysis. Romania has been dropped as there were fewer than 30 observations in the top two quintiles.
Source: Secretariat calculation based on the EU-SILC 2024.
Note: OECD and EU averages are unweighted. Data for Estonia refers to 2002 instead of 1995. Data for Korea refers to 2008 instead of 1995. No data for Türkiye.
Source: OECD Average annual hours actually worked per worker (OECD, 2026[36]).
← 1. It is difficult to empirically estimate the extent of this phenomenon, since longitudinal data are required to determine employment status (employee / self-employed) prior to a job- or earnings loss. The longitudinal EU-SILC does contain this information, see (Immervoll et al., 2022[37]), but there are too few observations for self-employed workers experiencing an earnings loss to distinguish between the type of benefit (contributory / general-revenue financed) they receive. Future work could look at this issue using administrative, individual-level data, that contains the universe of self-employed in several OECD and EU countries, collected at the OECD, see (Königs and Terrero-Dávila, 2025[38]).
← 2. In Germany, some self-employed workers are subject to compulsory pension insurance, including craftsmen and midwives, and others are insured within professional schemes, e.g. free professions and farmers.
← 3. Belgium, Denmark, Ireland, Greece, France, Italy, Cyprus, Latvia, Hungary, Malta, the Netherlands, Poland, Portugal, Romania, Slovenia and Finland and Norway (Eurostat, 2023[21]). The survey will be rolled out to all EU Member States in 2026.
← 4. Estimates on the incidence of digital platform work from the 2023 Contingent Worker Supplement have not been released yet.
← 5. OECD harmonised definition of under 30 hours per week in the main job. Using a definition of 35 hours would increase both the level of part-time employment, as well as the trend towards part-time work, see Box 3.1 in (OECD, 2024[1]). Only dependent employees are shown as working time for the self-employed is very unreliable.
← 6. The low share of severe material and social deprivation is also lower among self-employed workers, which may point to their incomes being less reliable.
← 7. Eurostat definition: income below 60% of the median household income. The standard OECD definition is below 50% of median household income.