Jeff Stein has a piece in the Washington Post about trapezoid programs. These are programs that phase in and then phase out based on current earnings, which has the effect of concentrating the benefit among middle earners while depriving the poor and the rich. I am quoted in the piece saying leftists and liberals should generally oppose trapezoid programs on the grounds that it is wrong to deprive the poor.

The piece and the discourse following it reminded me that I have been meaning to write a piece detailing the different types of welfare benefits. There are, of course, a lot of ways to group different welfare benefits. What follows is just my current way of thinking about the question based on familiarizing myself with a dozen or so welfare states across the world.

1. Income Replacement

The first category is income replacement benefits. These are generally cash benefits that are based on prior earnings levels to some degree. The primary purpose of these benefits is to provide income to people who face a drop in earnings for one reason or another.

The benefits that typically fall under this heading are old-age pension, disability pension, survivor’s benefit, paid leave, and unemployment compensation. In all of these cases, individuals who fall into a specific status receive a cash benefit that is based in part on their prior earnings.

There are two ways to structure income replacement benefits: (a) solely earnings-related, (b) earnings-related with a minimum benefit. Naturally (b) is the better approach of the two when it comes to reducing poverty, inequality, and income insecurity.

US commentators often claim that our income replacement programs fall into category (a), but this is not exactly true. Old-age and disability pensions through Social Security could be described as solely earnings-related, but in practice, elderly and disabled people with insufficient earnings records become eligible for Supplemental Security Income (SSI). Thus the public old-age and disability system as a whole looks more like (b).

Unemployment benefits are currently solely earnings-related in the US and thus fall into category (a). But it would be possible to add a “jobseekers allowance” for all unemployed people who are not eligible for the ordinary unemployment benefits, transforming unemployment compensation into a (b)-type benefit.

2. Income Supplement

The second category is income supplement benefits. These are cash and in-kind benefits that are not based on prior earnings, but are instead based on current income (if anything). The primary purpose of these benefits is to supplement someone’s current income.

The benefits that typically fall under this heading are in-kind benefits like health care, child care, social care, education, and housing. Also included are cash benefits like child allowances and wage subsidies.

There are three ways to structure income supplement benefits: (a) flat grants where everyone gets the same benefit, (b) means-tested benefits where benefit levels start high for those with low incomes and then phase out for those with higher incomes, (c) trapezoids where benefit levels phase in based on income and then phase out based on income, thereby concentrating their benefits on those a rung or two above the very poor.

Generally speaking, the left favors (a) because it wants to create a “universal” social democratic welfare state; the center favors (b) because it wants to target spending to the neediest with the lowest budgetary cost; and the right favors (c) because it incentivizes work and excludes the undeserving poor. But these are only generalities. In practice, you never see a system where all the income supplements follow one of the structures. Instead, you get a mix.

An example of a flat grant in our current system is K-12 education, which every child of a certain age is equally eligible for without any user fees. An example of a means-tested benefit is food stamps, which start out high for those with very low incomes and then phase out for those with higher incomes. An example of a trapezoid is the Child Tax Credit, which phases in and then phases out based on income.

Universal v. Means-Tested

The welfare discourse among internet pundits often takes the shape of arguing about universal benefits versus means-tested benefits. But this is not an entirely clear-cut distinction.

The only universal benefit in the sense that it covers literally everybody is public health insurance in some countries, but even then only if you conceive of the benefit as insurance rather than care (not everyone uses the medical system in a given year even if they are insured). But this is obviously not what people mean by universal. Instead, they mean to say benefits that, at minimum, flow to everyone within a specific population.

But this definition also runs you into some trouble. For instance, in other countries it is common for every parent to receive very large child care subsidies, but also to require parents to pay a means-tested user fee that maxes out at a few hundred dollars a month. This kind of benefit is universal in the sense that it provides benefits to everyone within a specific population but also means-tested in the sense that the benefit partially phases out based on income (just not completely).

Probably the cleanest way to define “universal” then is as benefits that flow to everyone within a specific population but also do not phase in or phase out based on current income. In that sense, there are two types of benefits in the taxonomy above that qualify: income replacements that are earnings-related with a minimum benefit (1b) and income supplements that are done via flat grants (2a).

This definition works fairly well but does not by itself resolve the question of what a “specific population” (i.e. universe) should be. You could imagine, for instance, a benefit that provided a flat payment to investment bankers. That would be universal insofar as investment bankers are a specific population, but we might also say that it isn’t really universal because investment bankers are not a relevant population that make sense as objects of welfare concern. According to this reasoning, excluding everyone who is not an investment banker actually makes your program non-universal in the intended sense of the word.

This latter point might seem a bit hypothetical, but actually seems to be where a lot of the unstated battle lines are when it comes to, for instance, benefits for students. Free college is a flat in-kind income supplement for a specific population (college students), but the implicit criticism is that college students are not a relevant population that makes sense as objects of welfare concern. This argument is in part why I have argued that we should instead identify the population as “young adults seeking to attach to the labor force” and create “universal attachment benefits” that flow to students and non-students alike. Students would get college and non-students would get training, apprenticeship, vocational education, in-work subsidies, and similar.

Regardless of the definitional fuzziness around the edges, I still nonetheless think the concept of universal benefits is a useful one and worth supporting as a policy matter. Flat grants and earnings-related income replacements with minimum benefit levels are simple to understand and maximize the number of people who are formal beneficiaries of the welfare state. I believe this makes for better living in and of itself and has secondary political effects that increase the size of the redistributive budget, improve the resiliency of welfare programs, and promote widespread positive feelings about the welfare regime. These things are key to the creation of a stable, egalitarian society.