Valerie Frey and Jasmin Thomas
More Effective Social Protection for Stronger Economic Growth
Main Findings from the 2024 OECD Risks that Matter Survey
Abstract
Social protection budgets are increasingly squeezed by declining revenues and competing demands for resources. Willingness to pay more for better social protection also remains limited. Governments must therefore look to improve efficiency and effectiveness in social protection, including through the use of digital tools and technologies. This chapter uses data from the 2024 Risks that Matter (RTM) Survey to better understand perceptions of social protection, government service delivery and government digitalisation, including governments’ use of artificial intelligence (AI).
Social protection budgets are increasingly squeezed by declining revenues (e.g. related to population ageing) and competing demands for resources (e.g. national security, pension and long-term care spending).
Public demands for social protection remain high across countries – on average, seven in ten say their government should do more to ensure their economic security – but there is limited willingness to pay more for better provision of social services and benefits. Willingness to pay more for social programmes has gone down since the pandemic. It is therefore increasingly important that governments improve efficiency and effectiveness in social protection.
Fewer than one in three respondents across 27 OECD countries feel they could easily receive public benefits if they needed them, reflecting both concerns around programme eligibility and concerns around navigating administrative procedures.
OECD governments have sought to make social protection systems more efficient in part by digitalising access via websites and applications. About half of all respondents say they use digital tools most or all of the time for accessing public services, and a similar share find these digital tools easy to use.
At the same time, four in ten people continue to prefer paper-based and in-person applications to digital tools. Many people do still use in-person processes, phone calls, mail-in paper applications and forms for government programmes and services.
OECD governments are increasingly using artificial intelligence (AI) in social protection, recognising its tremendous potential to improve programme effectiveness and efficiency. Yet the use of AI is not without risks, especially around privacy and transparency, and many people express concern about their government’s use of AI in social protection. Only 40% of respondents feel that the use of AI to help process and approve social programme applications is good for users, and only 32% say they trust their government with the data collected through AI and other digital tools.
Social programmes in many OECD countries are being squeezed in a narrowing fiscal space. A declining working share of the population means fewer taxes and social contributions are being collected, ageing populations imply higher expenditures on pensions and long-term care (OECD, 2024[1]), and governments face competing demands for resources outside of social protection, for example in the form of military security. These changes are taking place against a backdrop of high levels of debt (following the pandemic and the cost-of-living crisis) and a high-interest rate environment.
A critical question in a time of tight budgets is therefore, “How should we pay for social programmes?” While almost half of all RTM respondents expressed a willingness to pay an additional 2% of their income in taxes and social contributions for better healthcare and pensions in 2020 – during the COVID‑19 pandemic – the share has declined over the past four years. Today, on average, only 38% of respondents say they would be willing to pay an additional 2% for healthcare – the most popular policy area – with rates ranging from 50% in Ireland and Portugal to 26% in Lithuania (Figure 3.1).
Healthcare is not the only policy area to experience declines in willingness to pay, with many social programmes experiencing such declines since the height of the pandemic. Rates have remained stable (albeit low) between 2022 and 2024 in the policy areas of employment and unemployment supports, housing and family programmes. In general, willingness to pay for specific social programmes is correlated with the prioritisation of risks (see Chapter 1), which helps illustrate why policy areas with relatively lower numbers of beneficiaries seem to have the lowest levels of (potential) financial support.
Share (%) of respondents willing to pay 2% more in taxes or social security contributions to benefit from better provision or access to selected social and public benefits, RTM average, 2018‑24
Note: RTM averages are unweighted. Averages refer to the 27 participating RTM countries in 2022 and 2024, the 25 participating countries in 2020 and the 22 participating countries in 2018. Respondents were asked “Would you be willing to pay an additional 2% of your income in taxes/social contributions to benefit from better provision of and access to (a) Family supports (e.g. parental leave, childcare benefits and services, child benefits, etc.), (b) Education services and supports (e.g. schools, universities, professional/vocational training, adult education services, etc.), (c) Employment supports (e.g. job-search supports, skills training supports, self-employment supports, etc.), (d) Unemployment supports (e.g. unemployment benefits, etc.), (e) Income supports (e.g. minimum-income benefits, etc.), (f) Housing supports (e.g. social housing services, housing benefits, etc.), (g) Health services (e.g. public medical care, subsidised health insurance, mental health services, etc.), (h) Disability/incapacity-related supports (e.g. disability benefits and services, long-term care services for persons with disability, community living resources, etc.), (i) Old-age pensions, (j) Long-term care services for older people (e.g. home, community-based and/or institutional care), (k) Public safety (e.g. policing) or (l) Public transportation.” Respondents were also offered the option “not willing to spend an extra 2% on any of these things” (m) or “can’t choose or don’t know” (n). These last two options were fixed at the end and exclusive, while response options (a) through (l) were randomly rotated. Apart from (m) and (n), respondents could select multiple options. Note that some response options changed slightly between survey waves. Refer to the questionnaires at www.oecd.org/en/data/datasets/risks-that-matter-data-and-methodology.html for more details.
Source: OECD Risks that Matter Survey (https://www.oecd.org/en/about/programmes/oecd-risks-that-matter-rtm-survey.html).
Demand for (more) progressive taxation has remained stable since 2022. 58% of respondents, on average across countries, feel the government should tax the rich more to support the poor. Rates are highest in Greece, Türkiye and Italy. When looking at gift and inheritance taxes – a new question for RTM 2024 – only 26% of respondents, on average, say gift and inheritance taxes aimed at the wealthiest are fair, 40% say they are unfair, and 24% select “neither.”
Limited budgets make “value for money” in social programmes more important than ever. Yet despite OECD countries having the most advanced social protection systems in the world, the RTM Survey continues to find a significant gap between the availability of government benefits and services and people’s ability to access them. A high number of respondents across OECD countries report difficulties in navigating bureaucratic processes, understanding eligibility criteria, and completing application procedures – suggesting that needed services and benefits might not be reaching the people who are eligible for them.
To illustrate, only 31% of people, on average across OECD countries, feel they could easily receive public benefits if they needed them (Figure 3.2). This is an imperfect measure of accessibility, of course, as it conflates perceptions of statutory programme eligibility with perceptions of application/renewal processes. Yet digging deeper into possible barriers to programmes reveals (perceived) problems. Only 28% of respondents, on average, say that they think the application process for social benefits would be simple and quick.
Respondents are most confident in their knowledge of how to apply for social programmes, yet only 46%, on average, feel this way. Over half of respondents in Austria, Canada, Finland, France, Germany, Ireland, the Netherlands, Slovenia, Türkiye and the United States agree or strongly agree that they “know how to apply for public benefits”, suggesting the application steps are relatively better known in those countries.
Share (%) of respondents who agree with specific statements regarding government services and benefits, by country, 2024
Note: RTM‑27 is an unweighted average. Respondents were asked to indicate the degree to which they agree or disagree with the following statements: (a) I feel I could easily receive public benefits if I needed them, (b) I am confident I would qualify for public benefits, (c) I know how to apply for public benefits, (d) I think the application process for benefits would be simply and quick, (e) I feel I would be treated fairly by the government office processing my claim. Response options were: “Strongly disagree,” “Disagree,” “Neither agree nor disagree,” “Agree,” “Strongly agree,” and “Can’t choose.” Response items (a) through (e) were randomly rotated. The figure refers to the share of respondents who “agree” or “strongly agree.”
Source: 2024 OECD Risks that Matter Survey (https://www.oecd.org/en/about/programmes/oecd-risks-that-matter-rtm-survey.html).
OECD governments’ efforts to make social protection systems more efficient over the past two decades have largely entailed moving client services online, for example via websites and mobile‑based applications. To a much lesser degree, governments have also started using linked microdata to prompt clients to apply/renew and, in a few cases, automatically enrol clients into social programmes for which they are eligible (OECD, 2024[2]).
RTM 2024 finds that 52% of respondents, on average, say they use digital tools – websites, apps and emails – to access government services most or all of the time. Rates exceed 60% of respondents in Korea, Lithuania, Greece and Israel (Figure 3.3). 52% on average also report that they find such digital tools easy to use (Figure 3.4).
At the same time, a noteworthy share of respondents – 26% on average cross-nationally – say they principally still do in-person applications/renewals for benefits in government offices. 28% rely on phone calls, and 20% use mail-in paper applications and forms when applying for government programmes (Figure 3.3). And a higher share say they would prefer these options, with four in ten (43%) reporting that they prefer paper-based or in-person applications/renewals instead of online.
There is relatively little difference between older and younger people in terms of preferences for paper-based or in-person applications, with 38% of 18‑29 year‑olds and 35% of 50‑64 year‑olds preferring such processes for applications and renewals (see Annex A). There is, however, larger variation in perceptions of the ease of use of digital tools – with 55% of respondents aged 18‑29 years old finding digital tools easy to use, compared to 46% of respondents aged 50‑64 years.
Share (%) of respondents who use selected tools for accessing government services most or all of the time, by country, 2024
Note: RTM‑27 is an unweighted average. Respondents were asked “Thinking about the times you have applied for (or renewed) government benefits or services (e.g. childcare, pension, public healthcare) in the last 12 months, how often did you use these different types of communication methods: (a) Digital tools (e.g. websites, apps, emails), (b) Phone calls, (c) In person and (d) Paper-based processes (e.g. mail-in applications)?” Response options for each of the items were: “Never,” “Sometimes,” “Most of the time,” “Always,” and “Does not apply to me.” Response items (a) through (g) were randomly rotated. The figure refers to the share of respondents who think the item matters “most of the time” or “always” for each of the items.
Source: 2024 OECD Risks that Matter Survey (https://www.oecd.org/en/about/programmes/oecd-risks-that-matter-rtm-survey.html).
Public servants’ human connection helps serve as “guardrails” when modernising social protection (OECD, 2024[2]). These workers are appreciated but often seen as unavailable to meet client demands. Four in ten (43%) respondents, for example, agree that it is easy to reach a human staffer (either in-person or on the phone) when they have a question about benefits or services (Figure 3.1), while 29% disagree. On average, younger people (aged 18‑29 years) find it easier to reach human staffers (46%) than older people (50‑64 years) (40%) (see Annex A).
Share (%) of respondents who agree with specific statements regarding government services and benefits, by country, 2024
Note: RTM‑27 is an unweighted average. Respondents were asked to indicate the degree to which they agree or disagree with the following statements: (a) I feel that digital tools (websites, apps, etc.) for applying for/renewing services and benefits are easy to use, (b) I feel it is easy to talk to a human staffer (e.g. through a call centre or in-person) when I have a question about benefits or services, (c) I prefer paper-based or in-person applications/renewals for programmes instead of online. Response options were: “Strongly disagree,” “Disagree,” “Neither agree nor disagree,” “Agree,” “Strongly agree,” and “Can’t choose.” The figure refers to the share of respondents who “agree” or “strongly agree.”
Source: 2024 OECD Risks that Matter Survey (https://www.oecd.org/en/about/programmes/oecd-risks-that-matter-rtm-survey.html)
Perhaps unsurprisingly, respondents’ perceived ease of using digital tools for social programme is correlated with their actual use of digital tools (Figure 3.5). Among people who most of the time or always use digital tools, 61%, on average, say they find such tools easy to use. By comparison, only 42% of people who only sometimes or never use digital tools say they find them easy to use. A similar pattern emerges when looking at people who rely more on paper-based or in-person applications or renewals: those who use them most of the time or always are about twice as likely as other respondents to say that they prefer paper or in-person processes.
Share (%) of respondents who agree with specific statements regarding government services and benefits, by country and frequency of different styles of interaction with government services, 2024
Note: See Figure 3.3 and Figure 3.4.
Source: 2024 OECD Risks that Matter Survey (https://www.oecd.org/en/about/programmes/oecd-risks-that-matter-rtm-survey.html).
The use of AI in programme design and delivery holds vast potential for improving the efficiency and effectiveness of social protection. Presently, AI is most commonly used for client support (chatbots, digital assistants), automating back-office processes (helping to free up human workers for clients), error and fraud detection by providers, and collecting feedback from clients. AI holds future potential in its predictive analytics capacities, including forecasts of demands and shocks, identification of clients, prevention of harm and better tailoring of interventions to clients’ needs (OECD, 2025[3]). Yet these potential benefits are not without risks, including privacy risks, a lack of transparency, and the potential to reproduce harmful biases.
Perhaps recognising these risks, a sizable share of the public does not yet trust the use of AI in social services. Results from the 27 OECD countries in the RTM Survey highlight skepticism about the benefits of government use of AI, as well as a lack of confidence in how the government will use the data collected about them through digital tools and AI. Only 40% of respondents, for instance, feel that the use of AI to help process and approve social programme applications is good for users (Figure 3.6). 30%, on average, express uncertainty, and 25% do not believe that government use of AI is good for users of social programmes.
Share (%) of respondents according to whether they believe government use of AI for processing and approving applications is good for users of public benefits, by country, 2024
Note: RTM‑27 is an unweighted average. Respondents were asked “Thinking about the use of digital tools and artificial intelligence (AI) by government, to what extent do you agree or disagree with the following statement: Governments using AI to help process and approve applications is good for users of public benefits.” Response options were: “Strongly disagree,” “Disagree,” “Neither agree nor disagree,” “Agree,” “Strongly agree,” and “Can’t choose.”
Source: 2024 OECD Risks that Matter Survey (https://www.oecd.org/en/about/programmes/oecd-risks-that-matter-rtm-survey.html).
When asked whether they trust the government with the data collected about them through digital tools and AI, 37% disagree or strongly disagree, while 32% indicate that they do trust the government with the data collected (Figure 3.7). Fewer than half (44%) of respondents were confident that AI would only be deployed to assess public benefits applications when it is safe and trustworthy to do so (figure not shown). These findings suggest that governments have room to improve when communicating about, and building trust in, the use of AI in social policy.
Share (%) of respondents according to whether they trust the government with the data they collect on them, by country, 2024
Note: RTM‑27 is an unweighted average. Respondents were asked “Thinking about the use of digital tools and artificial intelligence (AI) by government, to what extent do you agree or disagree with the following statement: I trust the government with the data they collect on me through digital tools and AI.” Response options were: “Strongly disagree,” “Disagree,” “Neither agree nor disagree,” “Agree,” “Strongly agree,” and “Can’t choose.”
Source: 2024 OECD Risks that Matter Survey (https://www.oecd.org/en/about/programmes/oecd-risks-that-matter-rtm-survey.html).
References
[3] OECD (2025), “AI and the future of social protection in OECD countries”, OECD Artificial Intelligence Papers, No. 41, OECD Publishing, Paris.
[1] OECD (2024), Megatrends and the Future of Social Protection, OECD Publishing, Paris, https://doi.org/10.1787/6c9202e8-en.
[2] OECD (2024), Modernising Access to Social Protection: Strategies, Technologies and Data Advances in OECD Countries, OECD Publishing, Paris, https://doi.org/10.1787/af31746d-en.