The winners and losers from tax and spend

Yesterday the ABS released the latest of its surveys of the winner and losers of Australia’s taxing and spending, covering the 2003-04 period.

And the winners are…unemployed single parents with children under 5, taking in on average $878 more a week in income transfers and welfare services than they pay in tax, most of which is indirect tax (it has a statistical caution on the number, but despite low overall tax this group scores the highest payment on ‘tobacco products’, consistent with previous research showing the poor pay a disproportionate share of ‘sin’ taxes).

And the losers are…the top 20% of households by income, who pay on average $571 a week more in tax than they receive back in benefits and services.

As with previous surveys, this one finds that the bottom 60% of households are net beneficiaries of the tax and welfare system, ie they receive more in income transfers and welfare services than they pay in tax. The Australian gave this aspect its lead story, with the heading ‘Tax take helping Howard battlers’. The poorest 20% get more than 40% of all social assistance, and the richest 20% only 9%.

But the redistributive aspect of policy is driven by income transfers rather than government services. Before adjustment for household size, the households with the highest gross private income also receive the highest ‘social transfers in kind’ – though overall these are distributed fairly evenly across income groups. They do particularly well in relative terms out of tertiary education (presumably because children of affluent families are most likely to attend university), acute care institutions, and ‘community health services’. The latter two items suggest that this group probably includes more households with elderly members than the third and fourth quintiles.

Consistent with comments I have made previously on familist policies, couples with kids are not, on average, poor. The weekly private income of the lowest-earning of them, those whose eldest child is under 5, is still $270 a week above the average for all households. Because their private incomes are fairly high, and therefore incur high income taxes, they are not significant net beneficiaries of the tax and welfare system.

That’s not to say that they miss out on substantial government benefits. Especially once kids start school they get far more than average. What we see is the large-scale churn that CIS Peter Saunders writes about. For example, couples with an oldest dependent child aged 15-24 on average pay $583 a week a tax and receive $578 in benefits, mostly education.

Politically, what these statistics tell us is that reforming the welfare state is very difficult. It’s not just that a majority of the population are net beneficiaries of the tax and welfare systems. Even among those groups that on average come out even or as modest net contributors there would be lots of losers as well as winners. Somehow I don’t think a coalition of the biggest losers – the top 20%, single adults, and couples without kids – will ever be enough to secure reform.

33 thoughts on “The winners and losers from tax and spend

  1. “Somehow I don’t think a coalition of the biggest losers – the top 20%, single adults, and couples without kids – will ever be enough to secure reform.”

    Very true and this government has made it worse by doling out cash like it was their’s. However there is an increasing number of Australians who now vote with their feet by leaving the country and return when they have a pile of less taxed capital to play with.


  2. I wonder how much these sorts of figures will eventually be affected by the government’s recent superannuation changes. These represent a major potential tax benefit to the well-off. In the short term, they allow wealthy people to place well over $1m in post-tax dollars into a virtually tax-free investment environment. In the longer term, for a person who is willing to wait to consume, they effectively increase the 30, 40 and 45% marginal tax rate thresholds by $50K. This will become very costly in 20-30 years and could be used as a bargaining chip for broader reform at that time. Don’t know whether Peter Saunders will be around to push for it then though!


  3. JC – Its true that I prefer to pay less tax and avoid HECS by lecturing and studying overseas. Vive La DifFerance! As the post structuralists used to say. Also there is generally less bulldust outside of Australia in many lines of work which is a refreshing change.

    “And the losers are…the top 20% of households by income, who pay on average $571 a week more in tax than they receive back in benefits and services.”

    The roads cost a lot and those are the households most likely to use them. Did you factor the total cost of construction and maintenance into your equation?

    Also the Iraq war is a huge expense, as is the occupation of East Timor. “The top 20%” are most likely to use more oil and in many cases have a vested interest in Timor’s oilfields and companies that stand to gain from conflict.

    “The Top 20%” are most likely to own huge chunks of real estate which are basically ill gotten gains stolen from aboriginals who were in many cases massacred. The total cost of the armed forces that currently protect this property theft should be factored in as benefit for the rich.

    There are so many factors you have failed to include in your equations, that your analysis is somewhere beyond bias.


  4. Actually, I think in the long term the coalition of the biggest losers might form a considerable voting block. Anecdotally (perhaps someone can provide better evidence) — there is certainly some backlash due to even worse tax schemes in some places in Europe, particularily those with low birth rates, where not having children has become much more common, and hence the number of people who are in this type of block is larger. There are also some debates in France (higher birth rate) as to this problem, and NZ also, since people in this block are most likely to vote with their feet and move elsewhere. You occasionly see East-German style solutions suggested, like trying to stop these people moving, forcing them to work in crappy places via indentured contracts (“doctors for the county”) and so on.


  5. Conrad – I think you make a good point in that the ABS describes the different household types in terms of ‘life cycle groups’, which assumes that nearly everyone moves through the cycle of single, couple without kids, couple with kids, etc and therefore at some point cashes in on that third stage, when an increasing number stop at the first or second stage. But they are still a minority.


  6. But the redistributive aspect of policy is driven by income transfers rather than government services
    Yes, but its hard to say how much of this is real or how much is a product of the methodology (many services simply cannot be imputed properly to household types. Which households gains from police? defense? roads? etc). And lots of taxes are hard to impute too (What is the final incidence of corporate tax? Its not necessarily the shareholders).

    Which brings us to:

    Politically, what these statistics tell us is that reforming the welfare state is very difficult …. even among those groups that on average come out even or as modest net contributors there would be lots of losers as well as winners.
    Indeed, which should lead you to understand why massive churning is only unequivocally bad if our sole goal is redistribution of current income between equivalised household current income deciles. Which clearly it is not – in fact it’s not even a goal at all in most government taxing and spending.

    Peter Saunders contuiinually misses this last point.


  7. DD – They don’t even try to impute those things. But given the bottom 60% of the income distribution doesn’t pay for the identified services they receive, any benefit from other government-financed services is also redistribution.

    As to churning in the service sector, my main objection is that people pay via tax and get back sub-optimal public services, especially in education. The redistributive aspect is the main political problem.


  8. Andrew, firstly you talk about unemployed single parents and then talk about couples with kids, as though it were somehow surprising that incomes for households with children should be higher than those without. I would suggest that the increasing number of single-occupant households is skewing average household income downwards.

    my main objection is that people pay via tax and get back sub-optimal public services, especially in education

    Resources that may have been devoted to educating a few are redistributed to educate the many, perhaps not as well as the few may have been educated but nonetheless creating a multiplier effect across the economy/society not allowed for in this analysis.

    I wonder how much these sorts of figures will eventually be affected by the government’s recent superannuation changes.

    As the 2006-7 financial year recedes into history it will be shown the effect is indirect, as those who receive superannuation and those who become parents tend to be different people, just as those with a lazy $1m to drop into super tend not to be unemployed single parents.

    Look, I think the tokenism lavished on families is absurd, as is the situation where people can’t just receive a tax credit rather than the wasteful facade of a transfer payment. I’d love to see a wholesale reform of the tax-welfare systems with a view to incentives. Until then, I’d suggest that analysis such as the above isn’t necessarily helpful.


  9. Andrew – I don’t think it is suprising that families with kids are better off than average households. There are several factors that contribute to this – multiple incomes, the selection effect of men with good incomes being more likely to find partners, parents working harder to finance their family, and economies of scale. My point is only that these factors make additional largesse less necessary, yet families have been a major handout target of this government. Incidentally, equivalised income, which takes into account household size, also shows families to be better off.


  10. Except of course, Andrew, that this last result is sensitive to the equivalising formula used – which is always pretty arbitrary. Still, you’d think selection (poor couples have less kids, contrary to expectation, and are also much more likely to split into sole parent families) and motivation (nothing like having kids and a mortgage to make you work that overtime) would indeed make couples with kids better off on average.

    But there’s a logical flaw in your argument Andrew. Family assistance is based on individual families’ income. So the fact that, on average, families have more income than singles is neither here nor there – well-off families get less. I’m not defending the details, or even necessarily the principle, of family assistance here (I have some issues with both) – just pointing out that the fact that average income of families is high says nothing about an individual families’ income.

    And as for transfer payments versus tax credits, there’s a lot of misunderstanding here. The current system is deliberately designed to let the recipients – not the government – choose whether to have their FTB “churned” (ie taking it out of the pay packet as tax and paid back by Centrelink as a cash payment), have their PAYE withholdings adjusted so as to get the benefit in their pay packet, or claim it all as a tax refund at the end of the year.

    Most choose the first of these three, partly because single income couples prefer to get it paid to the non-earner (the “wallet versus purse” issue) and partly because PAYE changes are a hassle for them and their employer if family status or the job changes. This, frankly, surprised those of us involved with designing the system – it’s called Family Tax Benefit because we thought most people would do it through PAYE or (as with the US EITC) via their tax return. Centrelink, after all, is not an organisation that people find easy to deal with.

    The point is that it is the customer, not the government, who is choosing to churn. And that indicates that this churning must have some advantages.


  11. DD – I know that averages don’t tell you how much individual families earn, and indeed that medians are probably a better measure because some very-high earning families can distort the results. But the fact that there are (by definition) families below the mean also means (by definition) families above the mean and so in even less need of any subsidy that the figures in this publication show. These people won’t get much if any FTB A, but they get lots in education spending, to me the most signficant case of government destroying value through the tax and spend process.

    My principal objection to FTB is not churn but the redistributive aspect.


  12. Anyone got any thoughts on the net effect of the tax cuts in the last three budgets on this?

    And the 2005 HECS increases should have some impact on the level of ‘services’ recieved by high income households.


  13. DD – you’re telling us about the mechanics of churning, not the principle. You’re also just concentrating on the FTB (I understand that follows on from the topic thread) but there is more churn in the Australian welfare system than the FTB.


  14. Leopold – I would think that the trends we’re seeing would intensify. The tax share of the top income earners is increasing and more is being distributed. Middle class welfare is the policy objective of the Howard government and they do it well.


  15. Leopold – The 1997 HECS changes, which were a transfer of expenses from taxpayers to students, would have had a progressive effect. The 2005 changes, however, would have had no effect because the subsidy remained as before, but universities were allowed to charge a slightly higher market price. This survey only picks up government subsidies, not the total cost of education.

    I would imagine that families would have improved their position in the last few years, since they (on average) benefit from both reduced taxes and higher benefits. The increase in labour force participation is probably beneficial for families as well.

    The impact on quintile income and subsidy share is harder to predict; obviously reduced taxes benefit those who pay most tax (the top two quintiles) but there has been modest restraint on the spending side.


  16. No, Sinclair, the great bulk of churn in the welfare system is through FTB. The various tax offsets ensure that there is relatively little churn with the primary income support payments, and getting rid of that remaining churn would in fact be difficult.

    This is because most of it occurs amongst people who work part of the year and are unemployed in other parts of the year, rather than amongst people combining benefit and work at a given point in time. In the former circumstances it is a method of ‘clawing back’ some of the benefit paid in the form of a tax liability, while still providing income support at the time when it’s needed. In effect it converts part of the benefit into a loan.

    Andrew, which redistribution upsets you? That from the childless to child raisers or that from the better-off child raisers to worse off child raisers? If the former, how do you cope with the simple argument that such payments are to correct an obvious externality where the childless benefit from children (not least by them becoming future taxpayers)? If the latter what have you got against poor children (whose circumstances are hardly their own fault)?


  17. I’m sorry DD, I suspect we have differing views as to what constitutes churn at a definitional level (just as I had very different views to your collegues as to what constitutes welfare). From memory the CIS’ Peter Saunders argues there is about $80 billion in churn (this is the number that sticks in my mind – unfortunately my copy of his last book is at the office). The Budget papers, however, identify the FTB as costing $2.4 Billion. Now while each and every cent of that churn is a disgrace, I doubt that is a lot of money in the grand scheme of things to generate the heat that it does. (That works both ways – the Howard policy has detracters on the left too).

    Could you please expand on the externality argument for children? It sounds very implausible. What do childless high income earners care about future taxpayers when they are net losers now (and, no doubt, into the future)?


  18. DD – Churn is principally in services, not income support. For example, a couple family with eldest child aged between 5 and 14 pays on average $495 a week in taxes, receives $118 in cash benefits ($81 FTB), and $405 in services. The actual redistribution is quite small (though of course there is further implied redistribution, as thes families are making no contribution to other government expenses).

    The problem with income churn is that it catches people in the Centrelink bureaucracy and affects their incentives because of the FTB income test; the problem with services churn is its effects on consumer choice and product markets.


  19. Apologies for the long post –

    Andrew and Sinclair (among others)

    I have written about churning (amongst other things) at

    Click to access Paper7.pdf

    There are a number of different concepts and measurements being used here.

    One approach is to view “churning” as the “unnecessary” level of benefits and taxes in any given year – that is you could cut benefits and taxes by this amount and leave no-one worse off – that is, the distribution of disposable income would be the same. On this basis, churning of direct taxes and direct benefits only was equivalent to just under 23% of direct taxes or 5.6% of household disposable income in Australia in 1998-99 (the previous Household Expenditure Survey), which was then the lowest level in the OECD – see Table 4 in the linked paper, and the footnote which describes how churning is calculated. As the paper shows this was the lowest level in the OECD, because of the targeted nature of the Australian social security system, and the fact that the income tax system is also one of the most progressive (compared to income taxes combined with employee social security contributions in most other countries).

    Even so, this is an upper limit of this measure of estimated churning because it is estimated on the basis of grouped data for deciles and not unit record data. In addition, if a person is unemployed for 3 months or whatever and then gets a job for the rest of the year and pays more in tax than they received in benefits then this is treated as churning. Also if an age pensioner or a person with a disability lives with their working children then any benefits received are also included as churning – this is because these results, and those in the latest ABS study are based on household income, while the income tests in the social security system are based on nuclear families, and the tax system largely on individual incomes. On this narrow measure, I think that DD is likely to be correct that most churning is probably associated with the family tax benefit system.

    The second concept/measure which you are largely discussing here properly takes a broader measure of the welfare state and includes non-cash benefits and indirect taxes. This produces much higher measures of churning, because most services are not as targeted as cash benefits – even family benefits – and indirect taxes are much less progressive than direct taxes. Here the three largest sources of churning are likely to be universal health care, public education and the GST (which is the main source of churning for low income groups).

    The third measure, which is the source of the figure of the $85 billion churning, quoted by Sinclair is an estimate of “lifetime churning”. Using microsimulation Ann Harding estimated that on average an Australian would get back in cash benefits about 38% of the direct taxes they paid in their lifetimes. A separate estimate is that for age pensions an average Australian would get back 62% of the effective contributions they paid over the course of their lifetime (since age pensions are less targeted than unemployment benefits or sole parent pensions, for example). The $85 billion estimate by Peter Saunders is based on the arbitrary – but possibly not unreasonable– estimate that 50% of total spending on the welfare state (social security, health and education) is lifetime redistribution, and then applying this 50% to current annual spending on these items.

    On a lifetime basis, the three main sources of churning are likely to remain the public health system, public education and the GST. Essentially on the spending side, most of this occurs at the two extremes of the life course – when people have children and when they are pensioners, since a large share of health spending goes to older people. Indirect taxes affect people across the board.

    Personally, I think the challenge for critics of this “churning” is to come up with practical proposals for changing these policy areas that improves economic efficiency and leaves the distribution of disposable incomes unchanged. Doing proper microsimulation modelling of the winners and losers would also be useful from those making these proposals.

    Two final points: Andrew, I think it is misleading to look at all families as a group and conclude on the basis that average benefits are roughly the same as average taxes that not much redistribution is going on. If you look at Table 25 in the ABS publication you will see that for couples with children where no one is in paid employment, total benefits exceed total taxes by over $800 per week (before equivalisation), while for families where both parents are in paid work, total taxes exceed benefits by between $100 and $200 per week. Similarly, Table 29 shows that for lone parents total benefits exceed taxes by around $450 a week on average – and this will also vary by work status. Even on family tax benefits alone, Table 25 shows that two-earner families receive about one-fifth of what jobless families receive.

    Following on from this, if you replaced the public education system with a set of vouchers, then you would need to treat the vouchers in exactly the same way as the ABS treats the public education system. (Ditto for health of course.) The case for vouchers should be made on the basis that they enhance choice and lead to greater efficiency, not that they reduce “churning”, because it is likely that they will have no effect on this unless overall public support for education is reduced (with implications for the distribution of incomes.)

    None of this means that it isn’t possible to improve the efficiency of the existing system quite significantly, but I think that the way of doing this is to look in a lot of detail at specific policy proposals, rather than getting worried by global estimates of churning, when what they are are spuriously precise measures of complex policy challenges.


  20. Yeah, its a strange definition of “churn” that treats public goods as being churned. Is defence spending churning? What about roads?

    You can certainly oppose individual things such as public education or public health on the grounds that you don’t consider them the government’s business, but that is quite different from the assertion that paying taxes to have services delivered to you is always and everywhere a less efficient way to get those services. But the debate has to be grounded in that case-by-case assessment of the relative efficiencies and equities of the different ways of doing things, not some “churning is inherently evil and indicates massive inefficency” view.

    In fact can’t help but think that most of the churning “debate” is an attempt to avoid the real issues about the specific merits of user pays vs universal provision.


  21. DD – Who is talking about public goods in the context of churn? The discussion is only about publicly-delivered private goods (eg health, education).

    Peter W. is correct that the argument for vouchers is a choice argument, but the churn argument is relevant against the claim that if people want a private alternative they should pay the full cost, which some parents cannot afford, because they are paying so much tax. To put it another way, why should parents be taxed to deliver public schools when they would rather choose a private school?


  22. DD – you are confusing the public goods argument for intervention in education with the assumption that goods with public good characteristics should always be publicly provided (as opposed to merely having their consumption subsidised)


  23. DD
    The cost of one dollar going into government coffers and then sent out to be spent is about 40%. That’s what reasonable people call churn and it ought to be reduced as much as possible. I thought you would agree with that?


  24. Andrew –
    “Why should parents be taxed to deliver public schools when they would rather choose a private school”
    A good question, and we can certainly have a debate about it, but my point is that such a debate is orthogonal to the ‘churning’ argument. We are really talking about whether redistribution of educational opportunities is a Good Thing or not, but as Peter pointed out unless you means test the vouchers then so long as you have the redistribution you’re gonna have the churning (and even then you’ll have to have some churning. People’s circumstances do not stay constant). To put it mildly, it’s far from obvious that means tested provision is superior to universal provision.

    Jason –
    See my response to Andrew.

    JC –
    That’s a very high deadweight cost of taxation you’ve assumed (anyway arguably for these purposes the (much lower) deadweight cost of an average, not marginal, dollar of revenue should be what we consider). Yes, all else equal churn is undesirable. But all else is not equal.

    After all, churning is of the essence in insurance – private or social. I paid plenty in health insurance premiums last year and ended up getting roughly the same amount back in claims. Does that mean that I should now abandon my health insurance because it’s just churning?


  25. DD

    Churn when other more efficient methods are available should always be avoided.

    I don’t quite think your insurance example is good one. Insurance is to reduce risk. Having vouchers for education say would reduce admistrative costs outlaid by the government and improve choice. I can’t quite see how you can compare apples with… not even oranges in this case. I can’t see who you can compare apples with pasta.


  26. Two answers:

    “Churn when other more efficient methods are available should always be avoided”

    Which concedes my point – we should use the most efficient method available that achieves our goals. That varies on a case-by-case basis. Sometimes the costs of churning are far less than the costs of alternatives. Those promoting the alternatives need to compare these other cost with the costs of “churned” public provision on a case by case basis rather than by a blanket condemnation.

    “Insurance is to reduce risk”
    Yes indeed, and the biggest purpose of most real-life welfare states is not lifetime interpersonal redistribution for its own sake but intra;personal redistribution – it buffers risk. Hence the churning.

    Now returning to education, I take it we all agree that on both equity and efficiency grounds (ie lets not waste talent) all kids should get a chance at a good education. My biggest problem with vouchers is that I think they’ll undermine long-run political support for that proposition. I also think they’ll generally promote intolerance and bigotry through sectarian education, but that’s a secondary issue. Given those problems the short-run efficiency advantages of avoiding churning are, IMO, trivial.


  27. DD – Fixed. Preview buttons are Jacques’ area; I am not sure whether they are easy to instal or not.


  28. The GST is a good example of the point DD makes about selecting the most efficient means of achieving whatever one’s objectives are.

    Indirect taxes are the largest source of churning for low income groups (since they don’t pay much income tax). When the GST was introduced there was a compensation package to offset the price effects of the higher tax take. As a result, social security spending increased by more in one year than in any of the previous ten years, and it was also the largest increase in social spending of any OECD country in 2000.

    From one perspective this is nothing but complete churn and should therefore be avoided.

    But the reason why the Treasury proposed this policy is that they believe it to be more efficient than the alternatives – previously variable rates on all sorts of different things with exemptions for “necessities” and higher taxes on “luxuries”.

    As far as I can see the only way of avoiding this churn would be to give low income people a special concession smart card that would enable them not to be liable for the GST component of the price of a good or service and that they would produce to be swiped whenever they made a purchase. (I think Peter Saunders has made a proposal along these lines.) This seems to me to be technically feasible.

    However, I would also imagine that this would be a complete administrative nightmare, and the opportunities for fraud would be enormous. Imagine if you had a pensioner relative with one of these cards – they would do a lot of shopping for their extended families! Then of course you would also need the new branch of the tax office to check suspicious expenditure patterns among pensioners.

    So in this case, the current policy which involves rather a lot of churning is actually the most economically and administratively efficient way of achieving these policy ends.


  29. What rubbish
    My husband left me when I had 3 young children (this is more than 20 years ago) and I got an extra $50 a fortnight on top of the dole. Sure you can get transport, pharmaceutical ‘benefits’ (now $5.80 pf) but the dole is now $424 per fortnight and around $80 pf rent assistance = around $500 per fortnight. With the cheapest rent for a dump being around $160 pw for a one bedroom unit leaves approx $90 per week to buy food, pay for electricity, buy clothes etc and all the other stuff also, eg. doctor’s appointments etc.
    Try living on $90 per week to cover all costs. And regardless of our supposed low unemployment, a lot of employment these days doesn’t pay much more than the dole, especially as the ‘extra’ is needed for uniforms, transport etc.


  30. Oh, and people on welfare simply do not have the resources to defend themselves. Do you think, for example, that anyone on welfare could afford a computer? And let’s say they somehow got hold of a computer do you think they could afford the internet connection?


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s