Sophistry on redistribution

The Centre for Policy Studies has issued a report [PDF] by one Charlie Elphicke (hat tip to Chris Dillow), suggesting that the government has redistributed from poor to rich: “the poorest households in Britain are now paying a higher share of tax and getting a lower share of benefits than before Labour came to power”.

Which sounds bad. The calculations are based on ONS figures. In more detail: “In 1996/97, the poorest fifth of households paid 6.8% of the total tax take. This rose to 6.9% in 2004-05. Meanwhile their share of benefits has fallen from 28.1% to 27.1%.”

This is an odd way of looking at it. In fact, while I daresay the calculations are technically accurate, the decision to look things in this way is specious, perverse and politically motivated.

Elphicke makes no attempt to explain why these figures are the ones that matter. The share of total tax paid (or benefits received) by a given part of the income distribution has not been an object of government policy under either Labour or the Tories. Even given such a general approach, he does not try to justify the focus on quintiles rather than quartiles, deciles or even percentiles.

He makes no effort to argue that if the tax and benefits system had stayed as it was, then it would have continued to give and take in the same proportions as it did in 1996/97, nor does he try to contend that changes to the tax and benefits system since then have caused the changes he notes.

These particular numbers are the result of the interaction of thousands of economic factors, some affected by government policies other than those on taxes and benefits, and some beyond the control of government. So on top of the spurious choice of statistics for comparison, Elphicke heaps a post hoc ergo propter hoc fallacy.

One of the key changes since 1996/97 is that unemployment is a good deal lower. As such, we should expect that the poorest will now be receiving relatively fewer benefits and paying more tax, because more of them are in work and earning money. A return to mass unemployment would certainly reverse the trend that Elphicke so deplores.

It turns out that ‘the poorest fifth’ in 2004/05 is not nearly as poor as the equivalent group in 1996/97 – in no small part owing to the improved employment situation. Using the figures given by the CPS, we can calculate that their share of total original (pre-tax and -benefits) income has risen from 2.6% to 3.1% – an increase of 19%. Their share of total tax paid has risen by just 1.5% (from 6.8% to 6.9%).

Thus the increase in the poorest quintile’s original income relative to the rest of the population far outweighs the small rise in their share of the tax bill. Their improved position also contributes to the small decline in their overall share of benefits (although the level of benefits they receive has significantly increased).

Elphicke does not mention changes to levels of tax or benefit rates, or shifting thresholds, or altered eligibility criteria. Without this, his case is empty. In fact, his case is also wrong. A simulation by the Institute for Fiscal Studies found that changes to tax and benefits policy between 1996/97 and 2002/03 did reduce inequality by redistributing in the poor’s favour, relative to what would have happened had the system stayed the same.

(Update: I’ve just found where I put this [PDF]. According to research by John Hills, Director of the LSE’s Centre for Analysis of Social Exclusion:

“Comparing the 2004-05 tax and benefit system with the 1997 system adjusted for price inflation, the poorest tenth are on average 24 per cent better off than they would have been, and the top tenth slightly worse off. Against an alternative comparator of the 1997 system indexed by earnings growth but without reform, the structural changes of the last seven years are more clearly redistributive: the bottom tenth is 11 per cent better off than it would have been with this alternative, but the top four tenths are worse off.”)

Elphicke’s judgement is that “the impact of the Government’s policies resembles those of the Sheriff of Nottingham, not Robin Hood”. The rhetoric is as risible as the analysis. Indeed, it seems that the sole purpose of the analysis is to provide an excuse for the rhetoric.