Megatrends and the Future of Social Protection

Valerie Frey

Raphaela Hyee

Sociodemographic, economic, technological and environmental changes will affect both the need for, as well as the financing of, social protection systems in OECD countries over the coming decades. To provide adequate and sustainable support, the shape, size and funding of social programmes will need to anticipate and adapt to these “megatrends”.

Current economic challenges – including labour shortages and the cost-of-living crisis – brought on the repercussions of megatrends that had been expected in the medium-term. Labour shortages, a result of a drop in labour supply among some groups of workers in the wake of COVID‑19 (Salvatori, 2022[1]), have started to curtail economic activity and public service provision in some countries even before the full effect of demographic change sets in. Social protection systems must also adjust to the continuing trend towards part-time work, which extends beyond mothers temporary reducing working hours to provide care.

Similarly, while planned climate change mitigation policies will result in higher energy prices in the coming years, the sudden and dramatic spike in energy prices in many OECD countries following Russia’s war of aggression against Ukraine demonstrated how keenly energy price rises are felt by households and firms today. Once again, governments mobilised significant resources to cushion the impacts, but going forward, they are attempting to better target support.

This report presents a broad stocktaking of key megatrends impacting the future of social protection in OECD countries – ageing populations, changing patterns of labour supply, new and emerging employment forms, changes in household composition and unpaid work, the effects of new technologies on employment and wages, as well as the effects of climate change and the net zero transition – with an eye towards informing future reforms. In the context of limited fiscal space coming out of the COVID‑19 pandemic and higher interest rates, it puts an emphasis on the funding of social protection. This chapter summarises the report’s main findings.

Shifts in the age composition of the population reflect the combined effect of persistently low fertility rates and people living longer. While the rate of decline in fertility has slowed, fertility rates are nevertheless falling across the OECD on average and are below the replacement rate in almost all OECD countries. At the same time, longevity – remaining life expectancy at age 65 – increased at a pace of about 1.5 years per decade from the 1990s to the 2010s. This pace has slowed down to around one year by decade; the slowdown predates the COVID‑19 pandemic, which amplified it although it is too early to disentangle precisely the structural changes. Longevity is however expected to continue to increase in the future.

These combined trends have resulted in a rapid increase in the number of over 65‑year‑olds, as well as a decline in the working-age population, in most OECD countries. The number of over 65‑year‑olds per 100 people of working age – the old-age to working-age ratio – increased from around 20:100 in the early 1990s to 31:100 in 2022, and this ratio is expected to rise to 54 people over 65 years old per 100 working-age people in 2052.

The main policy lever countries have used to address the pressure population ageing puts on pension budgets is the upward adjustment of statutory retirement ages. Current labour market entrants, expecting to retire in the 2060s, will face a statutory retirement age of 66 across the OECD, on average, if current legislation is implemented, compared to around 64 for those retiring today. Women’s retirement ages will increase more than men’s, as countries increasingly equivalise pensionable ages for men and women. However, this increase in retirement ages falls short of stabilising the split between time spent in work and in retirement – while retirement ages are slated to increase by two years, life expectancy at 65 is predicted to increase by almost five years (OECD, 2023[2]).

As past pension reforms have raised retirement ages and educational attainment has risen among workers across cohorts, the age of labour market exit has increased over the last 20 years, reversing the downward trend that prevailed since the 1970s. However, in many OECD countries, early retirement schemes mean that average labour market exit ages are much below the statutory retirement age (OECD, 2023[2]).

Older people have a higher poverty risk than other age groups in two‑thirds of OECD countries, but in most countries, differences are minor. The average gap is driven by significantly elevated old-age poverty risks in Korea, Estonia, Latvia and Lithuania. Older women are more likely to be poor (17%) than men (11%). Women have lower own earnings-related pension income than men. Also, as they live longer, they are more likely to be widowed and live alone at some point in retirement, possibly reliant on a survivor pension.

Population ageing is a major challenge for social protection systems. It squeezes pension systems from two directions, as rising life expectancy may mean more aggregate expenditure on pensions, and lower fertility means fewer working-age people are available to pay contributions. It is also expected to increase spending in the health and the long-term care system. Increasing effective retirement ages is the main lever currently used to deal with increasing longevity and limit the pressure on public budgets. Higher statutory retirement ages need to be underpinned by occupational health promotion, life‑long learning and other employment measures to enable workers to extend their careers further.

Over the past quarter of the century, women’s labour force participation rate grew from 58% in 1995 to 66% in 2022 across OECD countries. The share of women working part-time is also on a downward trajectory: while it increased slightly at the beginning of the 2000s up until shortly after the Great Financial Crisis (GFC), it has been dropping since the beginning of the 2010s, from 23% to 20% across the OECD on average over that period.

There is no clear link between rising women’s labour force participation and a higher incidence of part-time work. Some countries, notably Austria, Korea, Finland or Italy, did experience rising women’s labour force participation combined with rising part-time employment among women. This is consistent with a narrative of the rising incidence of part-time work being a consequence of increasing women’s labour force participation (a composition effect): as the share of working women rises, the additional women entering the labour force are more likely to have a lower number of desired working hours. 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, and Israel. Indeed, the countries with the highest increases in women’s labour force participation had either low or no increases in the incidence of part-time work (Hungary, the Netherlands), or even strong decreases (Luxembourg).

The incidence of part-time work among men is still much lower than that of women. Nevertheless, men’s part-time rate rose in almost all OECD countries from 6% to 7%, on average, from 1995 to 2022, with most of the increase taking place during and shortly after the GFC. Rises in men’s part-time work were particularly striking in the Netherlands (8 percentage points), Finland (7 percentage points), and Korea, Germany and Austria (6 percentage points). Portugal, New Zealand, and the United States were the only countries where men did not increase their rate of part-time work. Increases in part-time rates among men and women are partly driven by young (15‑ to 24‑year‑old) workers, and likely connected to increasing educational enrolment. However, while prime‑aged (25‑ to 54‑year‑old) women’s part-time rates decreased from 21 to 17% between 1995 and 2022 across the OECD on average, prime‑aged men’s part-time rates increased from 3 to 4%. Late‑career men and women show similar patterns.

Labour market slack did contribute to the rising propensity of men to work part-time during the Great Financial Crisis (GFC), but underemployment cannot explain the persistent increase in male part-time work. At the beginning of the GFC, the share of men working part-time who could not find a full-time job increased significantly, and this labour demand effect is clearly visible in the overall incidence of men’s part-time employment, which increased steeply at the onset of the GFC and dipped in the boom years leading up to the COVID‑19 crisis. However, men’s part-time rate did not decline in line with underemployment during the economic recovery following the GFC.

It may be the case that men are more likely to work part-time because they are taking on more unpaid care duties. This is difficult to assess with available data sources, but there seems to be only a weak relationship between men’s unpaid caregiving and reduced working hours. Only 7% of part-time working men in European countries with available data stated that they work part-time to care for children or adults with care needs, compared to 19% of part-time working women. Another possible explanation for the increasing propensity to work part-time is an increasing preference for leisure – rising real wages as a result of economic growth could mean that workers increasingly value leisure. The fact that increases in male part-time work are concentrated in high-GDP European countries bolsters this idea. Further work, e.g. using time‑use data, should explore the reasons for men working increasingly part-time.

Rising women’s labour force participation can alleviate labour shortages in the short- and medium term. It also means that women increasingly acquire their own social protection rights, though their social security contributions tend to be lower as women are still disproportionately more likely to work part-time.

The increasing propensity towards part-time work among men has much the opposite effect: it diminishes labour supply at a time of labour shortages, and weakens rights to contribution-based benefits, including pensions. Should this trend continue in the future, it would have important effects not only on aggregate labour supply, but also on the financial stability of pay-as-you-go pension systems that are already jeopardised by the shrinking of the working age population.

Self-employment has been on a slow decline in developed economies throughout the second half of the 20th century and into the 21st century, primarily driven by the decline in share of the labour force working in agriculture, as well as small craft and retail businesses. It has declined in almost all OECD countries. There are, however, concerns that the number of independent contractors or freelancers who work for few or even only one client, in particular those whose work is mediated by online platforms (so-called platform workers), is rising. Quantifying the extent of this phenomenon is not straightforward, but available evidence indicates that only a small share (under 10%) of workers are affected, and that earnings are often or even mostly used to supplement other income.

Self-employed workers generally have less access to social protection than dependent employees, which means that they often rely on general revenue‑funded minimum income benefits in the event of an income loss. Self-employed workers control the success of their businesses in ways that employees do not. However, while running a business does – and should – imply risk, not all self-employed activity is equally entrepreneurial. Some self-employed workers are economically dependent on one or very few clients. For instance, platform workers, like other self-employed workers, have limited access to social protection, employment protection or minimum wages, but unlike other self-employed workers, they operate with little or no business capital, and may be dependent on a few or only one client.

In cases of straightforward misclassification of dependent employees as independent contractors, labour law (when properly monitored and enforced) may be sufficient to ensure the adequate protection of workers, and more often than not platform workers have been found to be employees by courts. Ensuring that workers are correctly classified also helps to prevent the erosion of social protection financing bases through regulatory arbitrage – employers lowering labour costs by choosing work arrangements with fewer social protection entitlements, and thus lower social security contributions. Where possible, both contributions and entitlements should therefore be aligned across employment forms.

Recent years have seen impressive advances in robotics and Artificial Intelligence (AI). This rapid progress has been accompanied by concern about the possible effects of AI and robotics deployment on the labour market, including on worker displacement. Whether technological progress leads to worker displacement depends on whether the technologies are substitutes or complements to labour – whether they have the potential to replace workers, or make them more productive, ultimately leading to higher output and employment. Empirical evidence points to technological progress changing the task content of jobs, and the occupational composition of the labour force, but there is no evidence yet pointing to job destruction. These empirical estimates are, of course, necessarily backwards looking and cannot account for the future deployment of new technologies across OECD economies. Evidence on wages is mixed – advances in robotics seem to have somewhat depressed the wages of workers affected by the technology, while advances in AI seem to have had a zero or even slightly positive effect on wages, with workers at the lower end of the wage distribution benefiting more from AI deployment (Georgieff and Hyee, 2021[3]; Georgieff, 2024[4]).

In the absence of population growth, rising productivity through technological progress is the only way to achieve long-term economic growth, which is necessary for the continued funding of public expenditure in a non-zero interest rate environment. Public policies can help to ensure that workers benefit from the productivity gains and cost savings generated by AI, which will also support social protection funding that in many countries mostly relies on labour income. For instance, worker skills will be a crucial determinant of the extent to which new technologies increase worker productivity and wages, and governments should support relevant training programmes. Public policy can also support collective bargaining and a fair distribution of productivity gains within firms (OECD, 2023[5]). Countries may also want to reconsider their tax mixes should technology-driven productivity growth mostly accrue to capital.

Women continue to provide the bulk of unpaid childcare, a barrier to paid employment contributing to lower labour force participation, wages, and women’s higher poverty risks throughout the life course. At the same time, in the absence of significant investments in public provision, the impending surge in demand for long-term care for older people is likely to fall on unpaid caregivers – mostly women.

As populations continue to age in all countries across the OECD, care needs will increase. OECD estimates suggest that employment in long-term care (as a percentage of total employment) would need to rise by 32% over the next ten years to meet the expected increase in demand for caregiving (OECD, 2023[6]). Declining family sizes and increased geographical mobility suggest there may be fewer people to provide unpaid care in the future. If high-quality, long-term care and childcare remain inaccessible or unaffordable, unpaid care obligations will likely increasingly depress women’s ability to work outside the home, at a time when tapping the full potential of the workforce is essential. This highlights the increasing need for public investments in the formal care economy, including home‑based and community care, and in social programmes that (financially) support unpaid caregivers.

Parents of dependent children are increasingly unlikely to be married, which further undermines the rationale of offering tax advantages to married couples regardless of whether or not they have children. Unmarried partners do not have access to alimony in the case of a separation, or survivors’ benefits in the case of the partner’s death, leaving especially women vulnerable, as they are often the main caregivers of children.

The share of adults living alone is also rising, partly driven by population ageing. Without the protective effect of a second earner’s income, and unable to benefit from consumption economies of scale, single adult households face a relatively high risk of poverty across OECD countries. Access to adequate social protection is therefore all the more important. From a social protection funding perspective, the absence of private risk pooling may mean increasing reliance on public income support.

Carbon pricing is a major policy to combat climate change. They are easy to implement and administer, and directly incentivise reductions in fossil fuel consumption by firms at households. In contrast to other taxes, e.g. labour or capital, a reduction of demand is desirable. Carbon tax revenues may therefore be used to lower other (distortionary) taxes, compensate households for carbon taxes paid, and/or subsidise green investments.

However, according to current estimates, carbon prices would have to rise substantially to reach the zero‑emission target – significantly more than the 2022 price surge, that caused a government response of support measures totalling 1.5% of GDP in the median OECD economy. Carbon taxes are highly salient – price increases are immediately obvious to consumers – and perceived to be punitively regressive. Even when they support the net zero transition, people tend to prefer subsidies, bans or regulations that push up prices or costs in a more opaque manner, or in a way that makes it difficult to identify winners and losers (Blanchard, Gollier and Tirole, 2023[7]).

Because lower income households spend a higher share of their income than higher-income households, they are ex-ante more exposed to consumption taxes (such as a carbon tax) in relation to their income. But higher income households may still have higher absolute expenditure (and therefore tax liability in absolute terms) – the relative as well as absolute burden of a carbon tax therefore depends on total expenditure and the carbon intensity of consumption (both directly on household energy and transportation, and indirectly through emissions “embodied” in goods and services throughout the production process) across the income distribution.

Empirically, the richest 10% of households spend about three to nine times as much as the poorest 10% of households, and CO₂ emissions closely track expenditure – that is, the richest households “consume” three to nine times as much CO₂ as poor households. This implies that a carbon tax places a higher absolute burden on high income households – redistributing the carbon tax revenue equally across the population, without any targeting by income, would make households at the bottom of the income distribution better off. Without such a redistributive policy, the relative burden on low-income households would be higher, as they save less, and also tend to spend a higher share of their income on energy and transportation.

To achieve public support for considerable increases in carbon prices, compensation policies need to be immediate, transparent and of sufficient size. Carbon taxes can certainly generate the fiscal space for such compensation. The COVID‑19 pandemic has shown that processing payments to a large share of the population in a short period of time can strain even mature social protection systems. Countries are increasingly using new data and digital technologies to increase their capabilities to provide benefits that are easy to claim (or even paid automatically), timely (that is, any conditionalities are established quickly) and responsive to changing circumstances. These investments will increase the credibility of carbon tax compensation payments.

Carbon tax revenues could also be recycled to lower other distortive taxes, e.g. on labour, to boost economic growth and to counteract lower productivity growth as economies adjust to higher carbon prices. Options include co-financing of social protection schemes, e.g. pensions, to lower contributions.

Housing construction and household-related energy consumption are major contributors to greenhouse gas emissions. Adaptive home renovations require large up-front investments and are often not undertaken even when economically viable. Homeowners may delay profitable investments due to present bias and liquidity constraints. For rented homes, there is a disconnect between landlords shouldering the cost of the investment and renters reaping the rewards of lower energy consumption that might not be well-mediated through the market.

Carbon pricing alone might therefore not be enough to solve the problem of insufficient long-term investments, and additional programmes to support adaptive home renovations for low-income households and those living in rented accommodation, including standards and bans, may therefore be needed. It is, however, essential that the cost-effectiveness of such policies be assessed in advance.

The net effect of the net-zero transition on employment depends on the size and design of climate change mitigation policies, the labour intensity of new technologies and production processes, and consumer and firm responses to changed market conditions. Current estimates forecast a slightly negative net employment impact of the green transition. Before the onset of the COVID‑19 crisis, around 6% of workers work in occupations that are concentrated in greenhouse gas intensive sectors, so-called “brown jobs”. About 20% of all workers worked in “green-driven” occupations, that are either directly connected to the green transition (e.g. Solar Photovoltaic Installers), will see their jobs change because of the green transition (e.g. Architects) or will see the demand for their skills increase (e.g. Construction Workers, (OECD, 2024[8])).

Both high-emissions and green-driven jobs are concentrated in transport, manufacturing, utilities and mining, high-emissions jobs are additionally overrepresented in agriculture, and green-driven jobs in construction. This sectoral overlap means that men are overrepresented in both green-driven as well as high-emission occupations, while women are overrepresented in services, and thus less exposed to the green transition. But this also means that women are worse positioned to take advantage of job opportunities emerging in the green transition – for instance, women are under-represented in STEM fields which may hinder their participation in these expanding industries.

OECD research shows that workers in high emission industries are, on average, more likely to have low educational attainment, and participate less frequently in formal and non-formal education and training programmes. At the same time, they are less likely to earn wages at the bottom of the wage distribution, and more likely to earn higher wages, than other workers, reflecting firm-wage premia (more generous wage‑setting practices in these firms compared to the average firm). They are also somewhat older than the average worker, more likely to be men, and more likely to live in rural areas. This combination of high wages with low skills, low participation in life‑long learning, and living in areas with few other job opportunities, points to high costs of job reallocation: these workers are likely to find it more difficult than the average worker to re‑skill after job displacement, they live in areas with low labour demand, and are more likely to face wage losses after transitioning to a new job.

Accessible unemployment insurance schemes with adequate replacement rates and durations can act as a first line of defence against earnings losses following job displacement. But given the characteristics of these workers, they are at risk to remain unemployed longer, and may require substantial re‑skilling, and may need help to relocate to areas with more job opportunities (Section 6.3). They may also face lower wages after transitioning to a new job in the long term. Thus, in addition to re‑skilling efforts by public employment services, additional income support policies may be necessary. For instance, wage insurance programmes partially cushion wage losses when transiting to a new, lower wage, and have been shown to be effective in incentivising jobless workers to accept new employment.

Additionally, extreme weather events such as floods, fires or extreme heat may cause unsafe working conditions and lead to work stoppages, and thus earnings losses, for workers. Some countries already provide insurance for this risk through their social protection systems.

In regions where high-emission industries are concentrated, local housing markets may deteriorate or even collapse, trapping workers and their families in their mortgages and preventing them from moving towards better jobs. Climate change may also directly damage or destroy homes and communities, e.g. through extreme weather events. So-called “managed retreat” programmes, and/or home buyouts, may therefore be necessary for some areas. Also, regions with a high concentration of carbon-intensive jobs are often not economically diversified. Investments in local economic development may therefore be needed to improve job prospects in these areas. Conversely, in areas with greater labour demand, the need for access to affordable and/or social housing, as well as public transit, educational infrastructure and other social services transport, may increase.

[7] Blanchard, O., C. Gollier and J. Tirole (2023), “The Portfolio of Economic Policies Needed to Fight Climate Change”, Annual Review of Economics, Vol. 15/1, pp. 689-722, https://www.annualreviews.org/content/journals/10.1146/annurev-economics-051520-015113.

[4] Georgieff, A. (2024), “Artificial intelligence and wage inequality”, OECD Artificial Intelligence Papers, No. 13, OECD Publishing, Paris, https://doi.org/10.1787/bf98a45c-en.

[3] Georgieff, A. and R. Hyee (2021), “Artificial intelligence and employment : New cross-country evidence”, OECD Social, Employment and Migration Working Papers, No. 265, OECD Publishing, Paris, https://doi.org/10.1787/c2c1d276-en.

[8] OECD (2024), Job Displacement in High-Emission Industries: Implications for the Green Transition, OECD Publishing, Paris, https://doi.org/10.1787/ac8b3538-en.

[6] OECD (2023), Beyond Applause? Improving Working Conditions in Long-Term Care, OECD Publishing, Paris, https://doi.org/10.1787/27d33ab3-en.

[5] OECD (2023), OECD Employment Outlook 2023: Artificial Intelligence and the Labour Market, OECD Publishing, Paris, https://doi.org/10.1787/08785bba-en.

[2] OECD (2023), Pensions at a Glance 2023: OECD and G20 Indicators, OECD Publishing, Paris, https://doi.org/10.1787/678055dd-en.

[1] Salvatori, A. (2022), “A tale of two crises: Recent labour market developments across the OECD”, in OECD Employment Outlook 2022: Building Back More Inclusive Labour Markets, OECD Publishing, Paris, https://doi.org/10.1787/ddfa64a7-en.