Credible tax and welfare reform must tackle the means testing mess

Australia’s tax and welfare systems have evolved into a basket case of contradictory principles that add complexity, undermine fairness and harm gender equity. They do this through means testing policies that have no clear net benefit – and likely net harm – to economic efficiency.

Since the 1980s, Australian governments have repeatedly cut personal income taxes, particularly at the top end, and substituted it with greater implicit taxation through means testing. In general, this process has favoured the well off. It has created an ad hoc implicit taxation regime that overlaps with and contradicts the principles of the explicit taxation regime. (Essentially all the arguments in this post have been made lucidly for decades by Australian economist Patricia Apps – see here and here – who should be considered one of the world’s greats in public finance. I’ve never seen anyone provide a compelling refutation, as opposed to shrug and proceed with business-as-usual low tax means testing policy. It’s like taking a public finance red pill, in the original Matrix sense – it’s easier to live in the world of illusion.)

Means testing reduces the economic reward associated with a particular activity, usually by withdrawing a benefit based on a percentage of each dollar earned. In economic substance it is thus identical to a tax; indeed any means testing regime could be turned in to a tax with some accounting and legal tweaks. On the face of it this is a mundane conventional point. It is recognised, in theory at least, in the orthodox textbooks. However, its implications are rarely followed through in a coherent way in the design of public policy in Australia.

Taxation via “means testing” involves some major constraints on tax design, however – it can only be progressive up to a certain point in the income distribution, and then turns regressive (when a benefit is fully withdrawn); and it can only be imposed on a subset of income earners (those who meet categorical eligibility requirements for a welfare benefit). Generally we would not want to design a tax like a means test, because such a tax would be considered inferior and indeed objectionable compared to more conventional taxation.

In the case of family benefits, the effect of means testing has been to create a de facto joint taxation system that overlays the individual taxation system across the low to middle income range. This effectively introduces a partial form of “income splitting”, an approach that is otherwise rejected in the Australian tax system, and is generally viewed as anti-feminist.

Income splitting allows one partner’s lack of income to reduce the tax obligation of a higher income earner, implicitly assuming they have equivalent living standards to a dual working couple on the same household income. This assumption is wrong. Two full time workers each on $50,000 are not as well off as one full time worker on $100,000 with an able bodied, employable partner not in the formal labour force. This is because the single income couple has an extra human being who can (and the time use surveys show, does) engage in considerable household production, while retaining the option of entering the formal labour market if they choose.

Full time household production is valuable. Conventional welfare economics implies that value of activity outside the formal labour market must be at least equal to the opportunity cost of the foregone wages if the decision to leave employment is voluntary. It could be argued that even an individual-based tax system is biased in their favour, as it ignores the value of this household labour (which would be taxed if it were undertaken in formal labour markets). It is strange to argue they should be given even further tax advantages, as per joint taxation, by reducing the primary income earner’s tax bill as well.

(The exception to the above is when the decision to leave the labour market is not voluntary. The case for favourable treatment of such a family is particularly strong when private household production is constrained due to disability or engagement in socially valuable activities like job search, caring or study. People who experience these social risks and lifecycle contingencies should be protected through the regular welfare payment system, whether singles or couples. As is so often the case, the conceptually correct solution is the simplest: consistent individual assessment of both welfare and taxation.)

Family taxation with income splitting is often viewed as a traditionalist conservative policy as it favours households with a stay-at-home parent. It creates substantial workforce participation disincentives for second earners – disproportionately women with children – as they get taxed at a higher rate due to their partner’s earnings.

Given Australia rejects family taxation in the formal tax system, it is strange we have allowed it to develop as an overlaid de facto system affecting low to middle income working families. There is no rhyme or reason for this inconsistency. It’s an artefact of two policy domains developing along contradictory internal logics without system-level thinking.

The effect of this patchwork has been to push high effective marginal tax rates downwards, and from people with no children to people with children, notably second earners who are mostly women juggling employment with caregiving.

Let’s consider a couple with three children with one full time worker on $50,000 and a part time worker minimum on $21,100. The $50,000 worker earns around 15% higher than the minimum wage, and around half the average full time wage. The second worker is working 19 hours a week at the minimum wage, for $21,000 or about 30% of the average wage for all workers.

Setting aside childcare and the Medicare levy, the full-time earner is subject to an effective marginal tax rate of 52.5% and the part-time minimum wage earner 39%. The low-income full-time worker pays a higher effective marginal tax rate than a worker on, say, $800,000 – within the top 0.1%. The part-time minimum wage worker pays a higher marginal rate than most workers on $170,000.

Let’s assume the full-time worker finds a new contract on $85,000. This worker now pays a marginal tax rate of 62.5%. But this promotion also hits the partner, due to the quasi-joint system. The partner – working 19 hours at minimum wage – now pays a marginal rate of 49%, higher than the top personal income marginal tax rate of 45% facing workers over $180,000.

Now, let’s assume the full time worker’s contract ends and can only find work at $65,000. The second earner moves full time and achieves a promotion to $46,000 (around 9% above minimum wage). They are now both paying marginal rates of 62.5%!

Here we have two battling full time workers – on 71% and 50% of the average full time wage respectively – and we are taxing them 62.5 cents in the dollar!

Until the 1980s we had a universal system of child benefits combined with highly progressive taxation, including marginal tax rates of 60% on high income earners, but much lower marginal taxes on wage earners nearer to the bottom. The combination of tax cuts – supposedly to improve work incentives – and means testing has turned the marginal rate schedule upside down, and pushed the disincentives onto low to middle income families, particularly dual earners.

It seems incredible that such a song and dance was made about work disincentives for the highest earners, only to reallocate them to working poor and middle-income middle women with children, but here we are. Ironically, second earners with children have unusually high labour supply elasticities. The post-1980s economic rationalists, in their wisdom, have directed the work disincentives of high taxation to… precisely where they will result in the greatest reduction of work. Strange!

If we’re trying to promote women’s workforce participation, this is bad in its own right. But it is also a reason to start from a presumption that a tax and welfare regime characterised by low explicit taxation combined with means testing (i.e. high implicit tax) is likely to be less efficient than one with high explicit tax combined with welfare universalism (i.e. low implicit tax).

The other reason is that explicit taxation allows us to spread the tax over a larger population (e.g. people without children, not only people with children), so we can lower the tax rate, whereas means testing of categorical welfare schemes concentrates the tax only on a subset of the working population (e.g. just people with children), thus requiring a higher rate. Assuming deadweight loss is proportionate to the square of the tax rate, a conventional assumption, this would be expected to increase deadweight losses. The efficiency case for universalism is given by the classic economic rationalist heuristic: broaden the base, lower the rate. Empirical evidence from Norway supports that higher tax universalism has superior work incentives compared to lower tax combined with a means tested child benefit. This result applied overall, but particularly to those on low and middle incomes (unsurprisingly).

Estimating the overall economic efficiency of different tax and welfare systems is quite a complex and empirically sensitive task; it’s a case by case exercise (it depends, for example, on how many people are subject to the different marginal rates). So we should perhaps be cautious about making a sweeping declaration here. But we can certainly reject the opposite sweeping declaration: the notion of a dogmatic principle that tax cuts, combined with eliminating the “churn” of “middle class welfare”, is generally good for work incentives and efficiency.

Not true, and we have good reason to start from a presumption of the opposite. The bogus belief in the intrinsic inefficiency of so-called churn has been hugely influential in Australian tax and welfare discourse, even at elite levels – “serious” pundits, policy experts and some academic economists who should know better. These people have commented on issues without thinking deeply about them, and have been tripped up by simple framing effects and accounting conventions.

(The conventional narrative, which is deeply ingrained at a pre-analytical level in people socialised into a neoliberal society, is that welfare payments are big government, which is inefficient because of high tax, whereas withdrawing welfare payments means smaller government, requiring lower tax and thus more efficient. This breaks down once you realise that income testing welfare has the same impact on work incentives as personal income taxation, and the distinction is an arbitrary artefact of accounting and the pull of moral intuitions about market allocated property rights as a natural baseline state. But I digress.)

Universalism is superior public policy for many other reasons:
*More coherently egalitarian as a means testing system fails to consistently favour households with lower living standards over households with higher living standards (it overlooks living standard variations driven by household composition)
*Greater insurance against income fluctuations driven by social risks, increasing utility
*Greater income smoothing for lifecycle variations, increasing utility.

If means testing is viewed as a politicised compromise, an inferior second best in a context where only low taxation is politically possible, it should be explicitly presented this way. On the contrary, means testing has been advanced in policy discussions as first best technocracy, as responsible economic policy. Indeed, Australia formerly had a universal child benefits scheme and was pushing towards universality in age pensions, for a time supported by both sides of politics. Universalism was unwound largely in response to neoliberal policy elites who argued that high taxation and big welfare states were economically irresponsible. This argument was presented as one of sound economics.

It is no such thing. And rationalist policy analysts – in think tanks, central agencies, universities, etc – have a responsibility to help educate public discourse and policymakers on the reality.

Tax isn’t about “the rich”

Public discourse in Australia is back on to the perennially pointless debate about who is “rich”. An argument over a vague and inherently subjective dichotomous variable is bound to generate heat and no light.

The song and dance always emerges when people who resist tax plea “but I am not rich”. A chorus of well-intentioned pro-tax folks respond “actually, you are”, pointing to distributional statistics. The tax whinger counters that current income is only a partial measure of economic means, ignores household composition, and that while they are high in rank the distributional skew is such that their income and living standards are closer to the middle than those who are nearer to the top. Round and round it goes.

While my sympathies are with the pro-tax side, the entire debate is ridiculous. We are furiously arguing and citing statistical evidence over a fuzzy word with no objective meaning. Meanwhile, we are ignoring, and implicitly accepting, the absurd premise: that paying tax is only for the “rich”.

That idea is grounded in a misperception about what the tax system does. The tax system is not primarily about redistribution from high income wage earners to low-income wage earners. The tax system helps here, but it is only a part of its role, and it will only have partial success at it. Tackling entrenched economic power and privilege requires deep political economic changes, linked to unions and ownership. A significant amount of redistributive justice can still be delivered through the tax system, but this should not obscure its other very important job.

The tax system is, in large part, about making a collective contribution to support public goods and services. We all benefit from these, so we all contribute. These general services include a legal system, environment protections, defence, healthcare, schools and other social protections that we lump together as the “welfare state”.

Given the above, tax is largely about redistribution to ourselves – we pay tax in years of relative abundance (when we are healthy, able bodied, full time workers) and it comes back to us in public goods and services that add value to our lives. We benefit at all times, but especially in the lean times – when we’re children, students, unemployed, sick, disabled or old. During these times we draw down on the system. All of us will be both lifters and a leaners, to use Joe Hockey’s famous phrase.

This system has two very nice properties. It helps smooth out lumps in our incomes over the lifecycle, and, relatedly, it mitigates our vulnerability to economic risks. It’s a sort of “insurance” policy. In the Australian context it’s quite limited insurance (apart from health and disability care), but even these plus a system of basic benefits that provide protection from poverty are still a significant reduction in risk and living standard volatility. That’s a valuable service for everyone.

The tax and welfare systems, then, are less about taking from high wage earners to give to low wage earners, and more about taking from wage-earners and giving to non-wage earners (and wage earners who are sick or disabled).

Roughly half the population is not in waged work. This is mostly due to childhood or age, but also disability and unemployment. So if you’re a full time wage earner you’re already relatively high ranking in terms of income. And if you’re healthy as well, you’re going to be a net contributor to the tax and welfare system. And that’s ok, because what goes around comes around. There are periods of your life when this will be flipped around, and you’ll be glad for the system you contributed to.

The above dynamic is why the most successful welfare systems (Nordic countries) have high taxation, especially on the rich, but also high taxes right across the spectrum. This is achieved through a mix of high personal income taxes, social security contributions, payroll taxes and VAT.

It’s also why Curtin and Chifley, long averse to taxing the working class, wound up doing exactly this when faced with the mathematical requirements of building an Australian welfare state. In dramatically lowering the tax free threshold – incorporating the broad working class into the income tax system for the first time – they reassured wage earners that this was fair and reasonable because they would get it back in social benefits. They used a notional politico-accounting construct called the “National Welfare Fund” to make this point. Everyone rags on the National Welfare Fund for what appear to me aesthetic reasons (it offends some wonk sensibilities) but I think it’s awesome.

Similarly, the large welfare state expansions in recent decades – Medicare and the NDIS – have been supported through increases to income taxation right across the spectrum. Highlighting reciprocity between tax contributions and social service benefits seems to be the model for successful tax hikes and social policy expansions. Unfortunately the Australian obsession with “targeting” (e.g. means testing) compromises the ability of social democrats to saliently convey this reciprocity for many programs, even though the social protections do benefit everyone on some level.

The idea that taxation is only for the rich is an absurd notion that socialists and social democratic must debunk. Tax is not a punishment. Tax is not bad. Tax is actually good. It’s the means by which we make shared and reciprocal contributions as citizens to a system which helps us all. Reminding people of this is the proven success model for welfare state expansion.