(For) whom the fisc tolls: negative gearing edition
TL,DR: In the ordinary sense of the word, negative gearing is a regressive, not progressive, feature of the Australian tax code.
Michael Potter of the CIS has an article/press release out arguing that “the average benefit of negative gearing is larger for low income taxpayers, and reduces as income increases, when this benefit is (correctly) measured as a proportion of income”. This would mean that removing the ability to deduct investment property rental losses from one’s tax liability would be a regressive policy (impacting the poor more, proportionally, than the rich). I’m not quite sure I agree with this analysis. I went to the ATO’s sample of individual taxpayer records for the 2013–14 year. Instead of looking at decile averages, as Mr Potter did, I graphed non-linear curves to show the relationship between income and negative gearing benefits as a proportion of income for the ATO’s sample of 2% of taxpayers. A downward sloping line shows a progressive benefit (i.e., a benefit that is proportionally more generous to the poor than to the rich) while an upward sloping line indicates, as you can guess, a regressive benefit.
(I should warn readers at this point that I’m not really a specialist in tax policy, so I may have goofed up in the analysis somewhere. I’d be grateful if someone would point it out if I have. I think I may have acquired a reputation for sarcastic quantitative take-downs; this ain’t that, it’s just a poke around the data.)
Deductions look weird
First off, it’s true that if you look at the rental losses incurred by different sections of the income distribution, negative gearing looks like a hugely progressive element of the Australian tax code:
This is, on the face of it, kind of amazing. On average, the people at the lowest end of the taxable income range are claiming the biggest proportional tax deductions—5 or more times their final taxable income! Note that this curve is for all taxpayers: those who are negatively geared, and those who are not, so if you don’t use negative gearing, then you’re still counted in the calculation that produced that relationship above (you’re counted as having a deduction of $0).
You might wonder why the line goes so flat — it’s because the scale is massively warped by the huge proportional deductions at the lower end of the taxable income scale. We can ‘zoom in’ to the flatter part of the curve, although the line still looks pretty flat, if not downward-sloping:
Tax benefits are what counts
But there are some problems with these graphs, and the kind of analysis that gets us to it—and to Mr Potter’s claims. One is that if we’re interested in the impact of negative gearing, then we’re not really interested in the size of the deductions that people claim, but rather the reduction in the amount of tax they pay. (My countless fans may recall that I had a dig at Troy Bramston last year for mixing these two concepts up.) If you have an income of, say $50,000, and you make a rental loss of $10,000, then you don’t gain $10,000 by claiming the deduction—you gain that fraction of the $10,000 that you would otherwise have had to pay in tax. If you’re paying the Medicare levy, then you’ll pay 39 cents in each additional dollar in tax, so you save $3900 when you’re allowed to claim the rental loss as a deduction. Therefore, we need to calculate the tax benefit of claimed rental losses—not just the size of those rental losses.* Here’s what our curve looks like in that case:
The existence of negative gearing now somewhat less regressive. Now, after a relatively low level of income, the tax benefit as a proportion of taxable income starts to climb as people get a little richer, then flatten off.
Of course, we still have this rather weird portion of the income scale right at the bottom, where we have what seem to be poor taxpayers who benefit quite a bit, proportionally speaking, from negative gearing—at least in comparison to middle-income or rich taxpayers. Nonetheless, when we cut out those really low-income people on the left hand side, which distorts the scale a bit, we can see a rising benefit-to-taxable-income ratio:
This is because richer people are on a higher marginal tax rate, so each dollar that they’re allowed to deduct saves them more money than it would if they were poorer. Negative gearing looks, in other words, like a regressive feature of the tax code, with the exception of people on about $10,000 or less of taxable income a year.
But taxable income not necessarily what we care about
The problem we have though, is that we’ve been considering taxable income—that is to say, income after we’ve deducted the rental losses. But if we think, as many people do, that people deliberately make rental losses in order to minimise their taxable income**, then taxable income may not be a particularly good guide to someone’s actual economic situation. What, then, if we looked at our implicit hypothetical the other way around? Rather than considering the value of the tax benefit compared to your post-deduction income (‘taxable’ income), we can compare it to your pre-deduction income. To be more specific: in the example above, when you claimed a $1000 deduction on an income of $50,000, we classified you as having an income of $40,000; the $10,000 reduction therefore represented a benefit of 25% of your income. Now, in our new ranking of income, we’ll consider you as having an income of $50,000, so your hypothetical deduction would represent only 20% of your income.
On Twitter, Mr Potter said this wasn’t a legitimate method of analysis, since progressivity is usually calculated on the basis of taxable rather than gross income, as is the case for the measurement of the distributional impact of the exemption on fresh food in the GST. This is true, as far as I know. But as we saw above, there is a good reason to suppose that the taxable income benchmark will be misleading when we analyse negative gearing, since the controversy about negative gearing is precisely that it can be used to ‘artificially’ reduce your taxable income. Mr Potter implicitly concedes this when he says that part of the low-income, high-negative-gearing-benefit group ‘could include non-working spouses holding negatively geared property’. An underemployed man with a CEO wife and a negatively-geared investment property is not likely to be ‘poor’, but he will look poor if we look only at his taxable income. Indeed, in the taxpayer sample file, there’s allegedly one man with a taxable income of $0 who is claiming rental losses of $410,864. If he weren’t allowed to deduct those losses from his other income, he would be scrounging around behind his couch looking for about $164,000 to pay in tax. Fortunate for him that he can ask his spouse (taxable income: $157,162) to chip in with the grocery bills.
So let’s add back in the size of the rental loss to our definition of taxable income.
Here’s our curve showing how much of a negative gearing tax benefit taxpayers get on average as a proportion of their pre-deduction income. (This is, again, for all taxpayers, not just those who use negative gearing.) It’s now pretty clear that if we don’t subtract the negative gearing deduction from our measure of income, the deductability of rental losses is a regressive benefit—that is, it benefits higher income earners more, as a proportion of their income, than it does lower income earners, at least up to about incomes of about $100,000, where it may become more proportional.
I would argue then that:
(a) some people seem to be using negative gearing to dramatically lower their taxable income***, and therefore taxable income may not be a good guide to their actual standard of living—they might be living off untaxed income sources, or they may have arranged their affairs so as to enable them to live off someone else’s income, at least officially.
(b) if you look at income before deducting rental losses, negative gearing is a regressive part of the Australian tax code, not a progressive one.
This of course doesn’t mean that any particular proposal to change the negative gearing arrangements is going to be a good idea. For one thing, we can only figure out the legal incidence of the tax benefit here. That is to say, all we know is how it reduces people’s statutory tax bill; we don’t know whether, if the deduction were disallowed, those people may then try to make up for their lost income by, say, raising rental rates. (There are a number of ways that negative gearing influences rents, some raising them and some lowering them, so I think anyone predicting exactly what will happen to rents with 100 per cent confidence is a bit of a charlatan). Nor can distributional analysis like this tell us what a ‘fair’ tax system looks like—that’s a matter for philosophy, not economic analysis. But yeah, I don’t really buy the line that negative gearing is a progressive part of our tax code. Even when we consider its benefits as a proportion of people’s income, it’s a benefit mainly for the rich.
One last graph: the distribution of income (with rental losses not deducted) for those using negative gearing and those not:
I think the case that negative gearing is a feature of our tax code that works in favour of the well-off is pretty solid.
All graphs made using the 2013–14 ATO sample unit file.
* Here, I’m assuming that taxpayers with a taxable income of over $18,000 pay the Medicare levy. This is not strictly true — that low income threshold is not the same as the tax-free threshold, and it’s quite a bit higher for senior citizens. I think this simplication is minor enough not to matter, although if you think it does matter, tell me why and I’ll write some more code to account for the levy more precisely. I’m not counting the deficit levy for high-income earners, nor the increase in the Medicare levy from 1.5 per cent to 2 per cent.
** The theory is to then make back the losses in advantageously-taxed capital gains when they sell the property later. I’m not really an expert on how easy or likely this is to eventuate.
*** To give you an example: about 20% of the negative-gearers with zero taxable income are claiming rental losses of over $18,200, which would bump them up out of the tax-free-threshold zone if the deduction were disallowed.