No, the poorest don’t pay higher taxes than the richest (RF blog)

We all know that parts of the tax system are very progressive – and this should be very apparent in the Autumn Statement when the Chancellor unveils perhaps £10 billion of tax rises that will target the wealthy. But wait, some say, when you factor in taxes such as VAT it’s actually poorer households that pay more tax.

That claim is usually based on the annual “Effects of taxes and benefits on UK household income” release, which was updated today with results for 2016-17. This looks at different sources of household incomes, including employment income and benefits, as well as the amounts paid in direct taxes like income tax and indirect taxes like VAT or tobacco duty. The data also allocates the benefits associated with the consumption of some public services, such as subsidised rail travel or the NHS, across different households. As such, it’s a fairly unique and invaluable resource.

Taken at face value, today’s 2016-17 data seems to confirm that the poorest tenth of households do pay more of their income (49 per cent) in tax than the richest tenth do (34 per cent). But the bottom line is that this data for the poorest is incorrect.

Figure 1: Total tax paid as a share of gross income, by equivalised household disposable income decile, UK

Source: RF analysis of ONS, Effects of taxes and benefits on household income

Using the raw data that underlies the above graph (this time focusing on 2015-16 because the 2016-17 microdata isn’t yet available), we can look in more detail. Figure 2 shows that it’s specifically the bottom 20th of the distribution – the poorest 5 percentiles – that are the outliers. And, zooming in even further, the poorest 1 per cent are off the chart: they seem to pay 265 per cent of their gross income in tax. That means for every £1 of income they seem to be paying £2.65 in tax.

Figure 2: Total tax paid as a share of gross income, by equivalised household disposable income percentile, UK, 2015-16

Source: RF analysis of ONS, Effects of taxes and benefits on household income dataset

These big differences between the poorest 5 per cent and those on slightly higher incomes are not driven by a regressive tax system. Instead, what’s driving this pattern is a mis-match between their reported income and their spending.

Usually, there is a strong relationship between a household’s income and its expenditure. But at the bottom of the income distribution, this relationship appears to break down. As research by Professor Mike Brewer has shown, the bottom few income percentiles appear to be spending far more than could be afforded from their income: indeed, the very poorest appear to be spending around the same as typical-income households, much further up the income distribution. As Figure 3 shows, some households with around zero income are nevertheless spending over £20,000 a year per person. And this spending in turn leads naturally to larger consumption tax bills.

Figure 3: Spending per person (from bottom to top), by disposable income per person (from left to right), UK, 2015-16

Source: RF analysis of ONS, Effects of taxes and benefits on household income dataset

Although the spending data is not perfect, there is good reason to believe that is not the problem here. Instead, there are two main possibilities.

First, the households with the lowest incomes might be drawing down savings or borrowing. To give an extreme illustration, someone who’d previously won the lottery and was now busy spending it might be at the bottom of this graph with zero ‘income’ but paying lots of VAT and other consumption taxes. More realistically, some may be students or others with temporarily low income who take a longer view of their likely lifetime income and so spend more than they have coming in. To the extent that this is what’s going on, it would be a bit misleading to consider the bottom 5 per cent of the income distribution to be truly ‘poor’.

But in any event, that is not the most important reason for the shape of the graph above. Instead, there are very strong grounds to believe the income data there is merely wrong. Recall, this is survey data, and if people don’t fill in enough details about the incomes that they and the other people in their household receive, they can easily end up at the bottom of the income distribution. This is the key driver of the ‘lowest income’ households having disproportionately high spending and so disproportionately high total tax bills.

What should we take from all of that?

First and foremost, it’s a sign that the data is in need of improvement – something the ONS is thankfully committed to. Also released today is its encouraging transformation plan for household income statistics. The first move will be to improve the data for the highest income households (which has its own significant problems), but subsequently: “research into coverage and values reported at the lower end of the income distribution will be prioritised”. With new powers to use administrative data, the ONS should be able to look up sampled households’ earnings and benefit receipts directly rather than relying entirely on personal survey responses.

But the other thing to take away from all of this is that, contrary to what we often hear, the poorest do not pay higher taxes overall than the richest (though this may still be true for indirect taxes alone, as the ONS points out). This is more a measuring problem than a truth.

Does this mean we shouldn’t worry about the progressivity of the tax system? No. While we might not trust the data at the bottom of the income distribution, the other striking thing about Figures 1 and 2 is just how flat they are. As I noted in a previous blog, tax burdens are roughly in proportion to incomes: which may not be in keeping with how progressive people imagine our tax system to be. It’s true that what matters most is the tax and benefit system as a whole, with the whole being progressive (and note, as above, that consumption taxes might be paid at a different time to when the income is earned). But there is no excuse, for example, for council tax having a deliberately regressive structure. With inequality and poverty now likely on the rise, we should certainly not be sanguine about the progressivity of the tax and benefit system, particularly as the government looks for ways to raise taxes for public services.

Originally published at

Leave a Reply

Your email address will not be published. Required fields are marked *