Finance & Development, December 2018, Vol. 55, No. 4 PDF version

Shifting Tides

Dramatic social changes mean the welfare state is more necessary than ever

Nicholas Barr

Shifting Tides

The world has changed in ways that affect families, work, and skills. In advanced economies in the early postwar years, most people got married and stayed married. The wife was the caregiver and the husband the breadwinner, generally in a steady job for many years, possibly a lifetime, usually with an unchanging set of skills.

Merely describing that world makes it clear how much has changed. Today, lifetime employment is no longer the norm. Labor markets are increasingly fluid. Rapid technological change requires workers to update their skills. And many more women are in paid work, more marriages end in divorce, and parenthood is less closely tied to marriage.

Over the decades, the welfare state has evolved in response to these changes in economic, demographic, and social circumstances. Those circumstances continue to change in ways that require changes in design while making a welfare state, if anything, more fundamental.

Why a welfare state?

Before addressing specific issues, we should ask the basic question: What is the purpose of a welfare state? One well-known reason is to assist the poor. A second fundamental, but less well understood, reason is to address market failures. Markets can be inefficient for many reasons, which have been addressed by powerful literature on the economics of information, behavioral economics, incomplete markets, incomplete contracts, and optimal taxation.

These problems both explain and justify the existence of welfare states. Imperfect consumer information makes it necessary to regulate health care and pension funds. Imperfect information by insurance companies about the riskiness of different applicants explains why the state or parastatal institutions provide insurance against health risks or unemployment. Behavior that diverges from strict economic rationality is an argument for making pension saving mandatory.

For these reasons, even if all poverty could magically be eliminated, a welfare state would still be necessary to provide insurance and to assist people in planning for their life course by redistributing from their younger to their older selves.

Third, the welfare state is an element in policies to support economic growth (Ostry, Berg, and Tsangarides 2014). Investing in skills is increasingly important for growth and for sharing the fruits of that growth. Income transfers also assist growth; for example, the ability to afford a healthy diet improves educational outcomes.

Encompassing all three reasons, the welfare state can be thought of as a device for optimal risk sharing:

Seen as insurance at birth against unknowable future outcomes, it helps relieve poverty.

Seen as a response to market failures, it addresses technical problems in private insurance, particularly relating to unemployment, medical risks, and social care.

In sharing risks in these ways, it contributes to economic growth. Without a safety net, people are less likely to risk a new start-up. On the other hand, too little risk is also suboptimal: the communist system protected people against almost all risk and thus stifled effort and initiative.

A more detailed look at the welfare state’s role as a device for risk sharing uncovers the starting point as the distinction between risk and uncertainty. The point is central: with risk, the probability distribution of outcomes is known well enough that the actuarial mechanism (that is, insurance premiums related to individual risk) works reasonably well. For example, the data on auto accidents by drivers of different ages and of different types of cars are good enough that insurers can calculate premiums for automobile insurance. But the actuarial model does not cope well with uncertainty, such as about rates of inflation well into the future. In contrast, social insurance can address both risk and uncertainty because a government can require everyone to be in a single risk pool and can adjust contributions over time.

What do these changes to risks and uncertainties for families, work, and skills imply for social policy?

When marriages were mostly stable, the main risk for a family was the death of the breadwinner. Today, more women are highly educated and take on paid work, and family structures are more diverse. These changes point to policies to widen choices between paid work and family obligations, including affordable childcare, and policies such as equal pay legislation to improve gender equity.

In labor markets, the main risk was once short-term unemployment. Today, people connect with labor markets in more diverse ways. They switch jobs more frequently, often with spells of part-time or self-employment, unemployment, or time outside the formal labor force. Employment is more precarious. In the future, technological change, including the spread of artificial intelligence, may make employment even more precarious. With that greater diversity of labor market relations, fewer workers in advanced economies have a record of continuous employment, and so organizing contributions to social security and private pensions through a worker’s employer have become less effective in providing good coverage.

Postwar social welfare systems assumed that a set of skills would serve most workers for life. Today, rapidly evolving technology creates the need for a more highly skilled workforce with a greater diversity of skills, and the speed of change means that skills have a shorter shelf life. These trends require fundamental changes to education and training. There will have to be more of it; it will have to be more diverse in content and methods of delivery, including a larger role for firms; and it will have to be repeated. These activities will have to be financed on a large scale.

As well as addressing these specific risks, social welfare systems also guard against systemic risk, including the risk of a trade war or economic crisis; political instability; environmental damage caused by climate change or nuclear accidents; and a changing age structure.

Not all these issues are new; the economic and political instability of the 1930s was an important driver of postwar reform. Other risks, notably those associated with damage to the environment and technological change, have become more prominent. Critically, not only are these systemic risks, but they are also mostly uncertainties. Both aspects reinforce the centrality of the welfare state.

Policy responses

What policies should we adopt to address these changing risks, and how will we pay for them?

Addressing income risks during working life includes providing income to the jobless and restoring and expanding earning opportunities, for example, through training and childcare. In this context, there has been renewed discussion of some variant of a universal basic income. Its feasibility depends both on the level of benefit and on the distribution of income. Since the distribution is skewed toward lower incomes, net beneficiaries will outnumber net contributors. As a result, the high average tax rate necessary to finance a large benefit would create major work disincentives. On the other hand, if machines guided by artificial intelligence raised growth rates and thus expanded the tax base, fiscal constraints might ease. A benefit of this type might become important for social as well as economic stability.

Addressing income risks in retirement means moving away from reliance on contributions based on employment status. Part of the solution is a flat-rate, noncontributory pension plan financed from taxation and awarded on the basis of an age and residence test, without a contribution requirement. Such plans are spreading in more advanced economies, including Canada, Chile, the Netherlands, and New Zealand, and in developing economies. Noncontributory pensions have twin advantages: they relieve poverty and reduce the gap in retirement income between men and women. A parallel change is to increase the minimum retirement age over time as people live longer. Choices about the level of noncontributory pension and retirement age should be made to relieve poverty without discouraging work and saving.

There is no single best pension system for all countries (Barr and Diamond 2009). Earnings-related plans that work well come in a variety of guises. One example is the notional defined-
contribution plan pioneered by Sweden in the 1990s. The arrangement is pay-as-you-go (meaning that this year’s contributions pay for this year’s benefits), but—unlike conventional pay-as-you-go plans—this one provides benefits that are closely related to a worker’s cumulative contributions. This design also has been adopted in Latvia, Norway, and Poland. Individual accounts, if part of the broader pension system, should be organized through simple, cheaply administered savings plans (mandatory or with automatic enrollment) that offer limited choice and a good default for people who make no choice (Barr and Diamond 2017). In the future, electronic payments open the possibility of basing pension contributions on consumption spending rather than earnings.

Addressing health risks, it is almost universally accepted among advanced economies that intractable market failures make private actuarial insurance a bad fit for medical risks, the United States being unique among advanced economies in its reliance on this approach. A key finding (Barr 2012) is that intervention on the scale necessary to address the slew of technical problems faced by actuarial medical insurance based on individual risk leads to an arrangement that is de facto social insurance, with everyone in a single risk pool.

Addressing the risk of skills mismatch must recognize the increasing complexity of providing appropriate education and training. The range of skills required in the job market is growing, as are ways of acquiring them; given the speed of technological change, workers will have to retrain, sometimes several times, over the course of an increasingly long working life.

Thus, what is needed is a system with at least three strategic attributes:

Emphasis on early childhood development, given powerful research findings that early gaps in cognitive and social development are hard to make up

Flexible choices for individuals over subject, method, and speed of skills acquisition and over pathways through vocational and academic training

A system of financing to support such delivery methods, including a mix of taxpayer money and, where possible, a well-designed system of student loans, as in Australia, New Zealand, and the United Kingdom

What is the role for individual contributions in these new welfare systems? Earnings-related benefits clearly must be contributory. However, where the primary purpose of benefits is insurance (health care) or poverty relief (basic pensions), contributions organized through a worker’s employment are not only less effective than in the past but can also discourage employment in the formal sector. So health care and similar benefits may be better financed from broadly based taxation (Levy 2008) or from a dedicated source of revenue that is unrelated to employment status; for example, from a portion of the proceeds of a consumption tax.

In all these areas, it is important to distinguish between the structure of an activity and how it is financed. Is an activity delivered more effectively by the market or the state? If there are no substantial market failures, market allocation complemented by income transfers is generally superior. How should the activity be financed? If public financing is involved, the answer will depend on a country’s fiscal situation and political economy. For example, Scandinavian countries vote for higher taxes to finance more and better public services in a way that is politically not possible in the United Kingdom or the United States.

Why state involvement?

Finally, why should the state be involved? Good social policy requires that market and state activity be mutually reinforcing, and that policy design go with the grain of economic theory. There are many solutions that respect market failures, recognize changed labor market conditions and family structures, and draw on the findings of behavioral economics—for example, "nudging" people to save more by automatically enrolling them in a pension plan.

All pension designs involve significant state involvement in financing and regulation, and in varying degrees also in delivery. The delivery of health care can be private, as in Canada; public, as in Scandinavia; or a mix of the two, as in France and Germany. Financing of health care can be organized at a national or subnational level or by nonprofits. In all cases, however, systems that work well are based on social insurance or tax financing, not private actuarial insurance.

Much of the debate about social policy is ideological. In the United States, public involvement in health care is often attacked as "socialism"; in the United Kingdom, private involvement is widely abhorred as "privatization." These arguments are not helpful because they locate ideology in the wrong place. The proper (and vital) place for ideology is in setting objectives, the "what." The "how"—or the respective roles of the market and state—should be treated mainly as a technical matter related to the extent of market failure in the face of major risks and uncertainties.

NICHOLAS BARR is professor of public economics at the London School of Economics and Political Science.


Barr, Nicholas. 2012. The Economics of the Welfare State, 5th ed. New York: Oxford University Press, 254–57.

———, and Peter Diamond. 2009. "Reforming Pensions: Principles, Analytical Errors and Policy Directions." International Social Security Review 62 (2): 5–29.

———. 2017. "Designing a Default Structure: Submission to the Inquiry into Superannuation: Assessing Efficiency and Competitiveness." Australia Productivity Commission.

Levy, Santiago. 2008. Good Intentions, Bad Outcomes: Social Policy, Informality, and Economic Growth in Mexico . Washington, DC: Brookings Institution.

Ostry, Jonathan D., Andrew Berg, and Charalambos G. Tsangarides. 2014."Redistribution, Inequality, and Growth." IMF Staff Discussion Note 14/02, International Monetary Fund, Washington, DC.

Opinions expressed in articles and other materials are those of the authors; they do not necessarily reflect IMF policy.