Iain Duncan Smith's uphill battle

From The Daily Mail:

A furious Iain Duncan Smith told Chancellor George Osborne to ‘show more respect’ in an explosive row over plans to slash Britain’s £180 billion welfare bill [Half of this is old-age benefits, of course]. The two men hurled insults at each other when Work and Pensions Secretary Mr Duncan Smith accused the Treasury of trying to block his crusade to end the scandal of welfare dependency...

Mr Osborne is understood to have defended his officials, hitting back: ‘If you come up with proposals that work, they will be treated with respect.’... The confrontation came after the Treasury was told Mr Duncan Smith’s reforms would initially mean a huge increase in the welfare bill – not a cut.

He wants to spend an extra £3 billion on a new benefits system to ensure that people in work are always better off than those who do not work. Mr Duncan Smith says his plan will save tens of billions of pounds – but it could be years before it pays dividends. Mr Osborne gave Mr Duncan Smith an ultimatum: ‘Find £10 billion of welfare cuts – or I won’t give you £3 billion for your new scheme.’

1. One thing that IDS and I can firmly agree on is that marginal deduction rates for welfare claimants/below average earners are far too high - rates of anywhere between 70% and 100% once PAYE is taken into account (i.e. they keep between 30p and 0p for every £1 they earn). One of the advantages of a Citizen's Income scheme is that you are always better off working, as you would get the CI regardless; there would be no means-testing, but you would have PAYE deducted from all your income - there would be no (artificial) distinction between taking money away in tax and taking money away via means testing.

2. Greater minds than mine have calculated that the revenue maximising income tax rate on lower earners is about 60%, which by definition must also be the expenditure minimising rate of benefit withdrawal. Assuming that Employer's NIC is economically borne by employees (about 11% of the wage bill above an exempt threshold), the expenditure minimising PAYE rate to be applied to CI claimants must be about 50% (I'd always go for a lower rate, but that's another topic), which can easily be dealt with via the PAYE system using K-codes.

3. Seeing as the administration and running costs of the DWP are about £10 billion a year with at least another £5 billion lost to fraud and error, doing the means testing via PAYE would enable us to save the bulk of this (so that's the required £10 billion savings in the bag).

4. The big presentational disadvantage of a Citizen's Income is that one government department (the DWP) would be handing out larger sums of money, but the extra revenue - the additional income tax paid by people who otherwise would have stayed on the dole - would be collected by another government department (HM Revenue & Customs).

5. With a bit of willing, this can easily be accounted for - all you to do is work out the difference between the higher PAYE collected (using a K-code) and the PAYE that would have been collected anyway - the difference is treated, for statistical purposes as negative benefits, so would count as a reduction in DWP expenditure rather than an increase in tax revenues. This would understate the overall cost saving (i.e. it would not include the extra corporation tax or VAT generated by CI claimants who are working, which is probably half as much again), but it's a good place to start.

6. Luckily, this is not a new concept. From page 124 of HMRC's 2009-10 accounts and trust statement (pdf)"To be consistent with the Organisation for Economic Co-operation and Development’s classification rules and international practice" HMRC already sub-divide Tax Credits into negative income tax and benefit expenditure; you just have to turn this on its head and split up PAYE received from welfare claimants with K-codes into income tax and negative benefits.

But I'm not sure that IDS is aware of this statistical nicety, and much less whether he can explain it to anybody else, let alone whether he can get his opponents to agree and what the best sort of formula is. Until and unless he wins this argument he is more or less stymied.