The Liberal Democrats have, once again, claimed that the poor pay more of their income in tax than the rich, and that this gap has got larger under Labour. But, by ignoring the fact that the poor get most of this income from the state in benefit and tax credit payments, and by overstating the extent to which indirect taxes are paid by the poor, this comparison is meaningless at best and misleading at worst.

The underlying figures come from the Office for National Statistics, and are not in dispute. As the Liberal Democrats say, in 2007-08, the poorest fifth of households had a gross annual income of £11,105 on average, and paid £4,302 a year in tax, a ratio of 38.7%. Meanwhile, at the other end of the scale, the richest fifth of households had an average gross annual income of £74,247, and paid £25,926 in tax, on average, a ratio of 34.9%. (See Table 1 of this article).

The first key point to note is that benefits and tax credits account for £6,453 of the £11,105 average gross income of the poorest fifth of households. Their original income - in other words, private income from sources such as earnings, private pensions and investments, but not that from benefits and tax credits - was an average of £4,651. So the poorest fifth of households were clearly net beneficiaries from the tax and benefit system, to the tune of £2,151 a year, on average. At the other end of the scale, the richest fifth of households received £1,666 a year in income from the state, and so they are net contributors to the Government's coffers, to the tune of £24,259 a year, on average.

If we define "net taxes" as "taxes paid less benefits received", then the net tax rate of the poorest fifth is -46% of their original income (or -32% of their after-tax income), with the negative number reflecting that they are net beneficiaries. At the other end, the richest fifth have a net tax rate of +33% of their original income (or +50% of their after-tax income). These figures show what one would expect: the tax and benefit system as a whole takes money from the rich, and gives it to the poor.

The combined impact of the tax and benefit system on the distribution of income seems much more enlightening than the impact of the tax system alone when talking about fairness. The Liberal Democrats' analysis does highlight, though, that benefits and tax credits do more to reduce income inequality in the UK than the tax system.

A second difficulty with the numbers relates to the way in which indirect taxes are treated, and this affects whether they appear to be regressive or not. People who are interviewed in the sort of surveys which underpin this analysis, and who report a low current income, tend to spend a lot relative to their current income, and therefore pay a lot of indirect tax relative to their current income. But this arises because the figures are a snapshot view of indirect taxes paid in a given period as a percentage of income in the same period. In reality, much low income is temporary, and people borrow and save to smooth out their living standards; over a lifetime, individuals cannot spend more than their income. As acknowledged by the ONS (see footnote 35 here), we get a different impression of the impact of indirect taxes by ranking people by their level of spending. That shows, for example, that VAT is progressive as a percentage of spending, since zero- and reduced-rated goods (such as food, children's clothes and domestic fuel) are necessities that are bought disproportionately by the poor. IFS researchers have written more about this issue here, although it must be noted that the ONS analysis suggests VAT is regressive even as a percentage of spending.

What about the Liberal Democrats' second claim: that the tax system has got less redistributive over time? Following the definition we introduce above, net tax rates on the poorest fifth have gone up since 1997-98, and those on the richest fifth have fallen very slightly (the 1997-98 figures which underpin this calculation are available in Table B here. But the poorest fifth have seen their own private income rise faster than the richest fifth (masking the opposite trend at the very extremes of the income distribution), so this may not be a particularly enlightening comparison: net tax rates should rise as incomes rise under a progressive tax and benefit system.

A more definitive answer is provided by analysis of the impact of Labour's tax and benefit reforms on the distribution of income, which have been strongly progressive, with the poorest households gaining, on average, and the richest households losing, on average (but with the precise amounts depending on what one considers to be a "change"). It is also clear that these changes acted to dampen down what would otherwise have been a large rise in inequality.

Even on the Liberal Democrats' peculiar definition, a natural follow-up is to wonder whether the tax system would get any less regressive over time under their policies? IFS researchers will provide a fuller assessment of that after their manifesto has been published. It is reasonably clear that the reforms to capital gains tax and the new mansions tax should increase the tax burden on the rich, but it is less clear that the proposal to increase the income tax personal allowance to £10,000 will help many of the poorest households, as the poorest fifth of households will contain those with incomes too low to pay income tax. (In any given year, one third of adults do not pay income tax and one quarter of adults live in a family where no-one pays income tax). The largest beneficiaries of the higher personal allowance will be families with two earners (where both earn less than £100,000).