Universal social insurance is sound economics

I’ve previously argued that the pundit obsession with “targeting” welfare benefits in order to lower taxation is often economically non-sensical. I focused on the most straightforward case: welfare payments with means test withdrawal schedules such as child benefit income tests and age pension asset tests. This argument is easy to make because the disincentive effect of that type of means testing can be shown as equivalent to taxation, and expressed as an “effective marginal tax rate” (EMTR).

But the same argument can be made for social insurance more generally.

In thinking of the tax we pay to support the welfare state in the context of a progressive tax system, we can divide it into two tax-welfare bundles:

1. Tax paid to support social protection for people poorer than us
2. Tax paid to support social insurance for ourselves.

The first sort of welfare is what Nicholas Barr, eminent welfare state economist of London School of Economics, calls the “Robin Hood” function of welfare. It’s about alleviating poverty and promoting equality. The second sort is what Barr calls the “piggy bank” function of welfare. We can think of this as smoothing our living standards, reducing the peaks and troughs of our financial wellbeing across life’s contingencies.

Even if all poverty and social exclusion could be eliminated, so that the entire population were middle class, there would still be a need for institutions to enable people to insure themselves and to redistribute over the life cycle. Though private institutions are often effective, they face predictable problems, and attempts to address those problems inescapably involve state intervention.

The Welfare State as Piggy Bank (Barr 2001)

In economic theory, these two bundles have quite different effects on economic efficiency. Conventional public finance worries that too much tax will create deadweight losses by undermining incentives to work and save. Economists are divided on the magnitude of these effects. Personally I don’t believe it’s anything to worry about at the margin in countries like Australia, or even much higher taxing countries. Even if it was, in a society with a large amount of positional consumption, deterring extra work and money-making can sometimes be good, as Robert H Frank argues:

Traditional models view the income tax as a wedge that causes people to expend too little effort. In contrast, positional models view this tax as a device for mitigating consumption externalities. Available evidence favors the latter interpretation, suggesting that most Americans would be happier and healthier if they worked not more hours but fewer.

But let’s set these arguments aside. From a traditional public finance perspective, the Robin Hood tax-welfare bundle reduces incentives to work. If you slack off and reduce your work hours, you’ll get protected by welfare; meanwhile work becomes less attractive because your pay is docked to support that person’s welfare. That doesn’t mean we shouldn’t do the Robin Hood tax-welfare of course, but from a traditional perspective it comes at a price – diminished work incentives and thus a smaller economy. That’s the basis of Okun’s famous equality-efficiency trade-off, the idea that redistribution takes place with a “leaky bucket”. Again, I think this is a crude simplification; evidence suggests that greater equality can be beneficial for economic output. And even if the simple version of the story is true, we should still promote equalising policies because the benefits to human wellbeing from greater equality are so large, on plausible assumptions, they often dominate efficiency considerations at the margins we see in capitalist societies.

But whatever we conclude about the economics of the Robin Hood tax-welfare bundle, it is a completely different story to a piggy bank tax-welfare bundle. Tax to support redistribution is lost to the taxpayer, but tax to support social insurance comes back. One form of compensation is substituted for another.

We can show this with a trivial example. Tomorrow the Australian Government decides to introduce the largest tax in world history – increasing personal income tax on wages by 50 percentage points across the board. But in order to promote work, it simultaneously introduces a 50% wage subsidy paid to the worker (untaxed). Income taxes go through the roof. Australia shoots from low to the top of the taxation and spending tables. But… in the real world, nothing changes. Work incentives remain identical. Everything carries on the same.

But what if the government benefit wasn’t cash, but rather government services, such as social insurance. We sometimes express this substitutability of wages and social provisions with the concept of the “social wage”. During the 1980s, for example, the Australian Labor Government struck a bargain with unions to contain wages to control inflation in exchange for superannuation retirement provisions, as well as through other social protections such as healthcare.

The money-go-round is sometimes easy to see – almost as easy as our tax-wage subsidy example above. We can tinker with the legal and accounting formalities of super to make it thus. Instead of effectively requiring employers to make 10% super contributions, we could require they pay a 10% payroll tax. In this hypothetical, the government would then make the 10% super contribution to the employee’s nominated industry or retail account. This would be substantively identical to the system we have now. Yet suddenly Australia would look on paper like a much higher taxing country. This is important for understanding (some of) the large gap in taxation collection between Australia (30.5% of GDP) and a country like France (43.4% of GDP). Australia’s retirement scheme is largely off budget, whereas France’s is on budget – people pay more tax, but it comes right back to them through pensions linked to tax contributions.

The point here is that compensation can be paid as (1) wages or (2) a mix of wages and social wages provided by (a) the employer or (b) government. Bundles (1), (2a) and (2b) will vary dramatically in terms of accounting. Superficially they will appeal to different ideological impulses, depending on whether one likes “big” or “small” government. They are very different in what policy experts call “vibe”. But in terms of economic incentives and deadweight losses, it’s swings and roundabouts.

Let’s look at another example. Medicare for All in the USA would require payroll taxes on employers – substantially increasing tax and government spending, while reducing after-tax wages – but on the other hand, it would save employers from having to pay private health insurance premiums – ultimately increasing wages. Now because the government can provide the benefit more efficiently than the private sector can, the savings to employers, and ultimately workers, are greater than the tax impost.

In this case, the tax and welfare system is not a leaky bucket at all; it actually comes back with more! Personally I like to think of it as a KFC bucket with extra chicken pieces, which allows me to segway elegantly another metaphor: “a free lunch”. Tax and welfare – at their best – give a free lunch.

But how could spending by government be more valuable than spending by the individual? Isn’t government inefficient? The government does some things better than the private sector, and one example is the insurance of social risks. These include medical benefits, unemployment benefits, and public age pensions, which are about the risk of running out of money due to old age. Governments have superior ability to pool risks over a large number of people, and thanks to the power of compulsion can overcome problems like adverse selection. Adverse selection refers to the problem of more risky individuals seeking insurance than low risk individuals while the company lacks information that would enable it to identify higher risks and charge them accordingly. Adverse selection results in the low risk people subsidising the high risk people, leading to a bad deal for the low risks, which in turn leads to even higher premiums and even more low risks dropping out, creating inefficiency. Government can overcome this whole problem by forcing everyone to join the scheme.

Nicholas Barr draws on legendary economist Kenneth Arrow, who famously demonstrated the potential efficiency of public health insurance:

Arrow argues that, where markets fail, other institutions may arise to mitigate the resulting problems: ‘The failure of the market to insure against uncertainties has created many social institutions in -which the usual assumptions of the market are to some extent contradicted’ (Arrow 1963: 967). In other words, institutions (public or private) may arise that are insurance in the sense of protecting against risk, even if they are not insurance in a narrow actuarial sense. There is an interesting contrast between Arrow’s arguments and those of Hayek (1945). The starting point for both writers is asymmetric information. To Hayek the fact that different people know different things is an argument in favour of markets: we go to a doctor or lawyer precisely because they have specialist information that we do not. Hayek argued that the market makes beneficial use of such differences by allowing gains from trade to be exploited. Arrow showed that the market is an inefficient device for mediating important classes of differences in knowledge between people.

Barr goes on to cite Robert Lucas, another influential Nobel prize winning economist, extending Arrow’s point to other social risks like unemployment:

 Since . . . with private information, competitively determined arrangements will fall short of complete pooling, this class of models also raises the issue of social insurance: pooling arrangements that are not actuarially sound, and hence require support from compulsory taxation. The main elements of Kenneth Arrow’s analysis of medical insurance are readily transferable to this employment context.

Out of interest, Lucas is not, perhaps, an economist one would intuitively associate with the welfare state. A doyen of University of Chicago macroeconomics, he notoriously described caring about inequality as the “most poisonous” trend in economics. So it’s fair to say he’s not a fan of Robin Hood, but he still saw the efficiency argument for welfare.

So tax supported social insurance bundles have two different, and often conflicting, effects on incentives. Let’s focus on the benefits side for a moment. Benefits can be structured in three different ways, each with different incentive affects:

Any basic benefit involves a “Robin Hood” component, and that requires a redistributive tax and thus a work disincentive. The question is what to do from that baseline. Means testing will minimise extra tax, but the means test itself will reduce benefit eligibility with work (or saving) and thus reduce the attractiveness of work just like a tax. On the other extreme, the welfare benefit could increase with income, incentivising work, but at the cost of much greater taxation (having the opposite effect). The middle ground (2) is a flat level of benefit that goes to anyone in eligible categories. This involves less benefit and less tax than (3), but also avoids the effective tax of means testing from (1). Under (2), a high income earner’s tax is mostly going to support a Robin Hood welfare bundle, but some fraction of their tax (declining with income) is purchasing social insurance at a basic level, and thus a piggybank tax-welfare bundle. Moving from a flat basic benefit (2) to a means tested basic benefit (1) would eliminate the piggybank bundle.

Given the conflicting incentive effects of the benefits and tax side in (2) and (3), how the net effect wrings out will depend largely on whether the government program is adding value. If the program is shifting money from people when they are flushed with cash in a high income job to when they are strapped for cash due to social contingencies (like unemployment), or medical, disability and old age, it’s very useful. We can’t efficiently save for these things, because we don’t know if or when and for how long we’ll be unemployed or sick, and we don’t know how long we’ll live. We may have various reasons for choosing one bundle over another – maybe we like the implicit egalitarianism in (2) for example – but we shouldn’t start with strong preconceptions about efficiency.

In assessing efficiency, we should think of tax not in isolation but in conjunction with the particular welfare it supports, what I have called “tax-welfare bundles”. Whereas traditional public finance models would suggest “Robin Hood” tax-welfare bundles have work disincentives that we accept as a price for greater equality, this theory does not generalise to “piggybank” tax-welfare bundles. In adherence to a low tax ideal, Australian governments have created an elaborate array of means tests to exclude middle and upper income groups. They’ve lowered taxes by cutting out the piggy bank. But this sort of tax cut will not generally improve incentives, and may harm them.

Once you realise this, it becomes clear that so many schticks of the Australian policy establishment make no sense. The knee-jerk hostility to “middle class welfare” – makes no sense. The obsession with “targetting” – makes no sense. Fixation on the evils of “churn” (aka both contributing to and withdrawing from the piggybank) – makes no sense.

Social democrats in the Esping-Andersen tradition argue that welfare universalism is good politics. But broad-based social insurance is also good policy.