Last Friday the TUC published Richard Murphy's analysis of how much extra tax will be collected in 2011-12 by the UK's 50% top rate, assuming zero elasticity of taxable income. His answer (on p19) based on HMRC's published estimates is £6.789bn: on the next page he reduces this to "between £3 billion and £6 billion a year", for reasons not stated.
Earlier in the week I gave a one-line calculation based on data from the same HMRC table which gives the number as £6.1bn. So who is right?
The answer, naturally enough, is neither of us. It's impossible to work this out exactly from the information given, but Murphy fails to account for two relevant factors and I overlook one of them (my excuse is that I was merely showing how Murphy should have avoided the crass error in the calculation he implied he was doing).
First, tax relief, especially on pension contributions. Murphy talks about such reliefs at some length starting on page 26 (oddly, he gives ISAs as an example, but they give tax relief on their returns, not income tax relief at the time of investment). Tax reliefs are a problem for his calculation only because he works from total income, which includes an unspecified amount of tax relief. I avoid this problem because I work from the actual tax paid, which naturally is calculated from income net of tax relief.
(Using the total income gives Murphy an additional minor problem: in the £150k-200k and £200k-£500k ranges there are some taxpayers whose net income is below £150k and who therefore avoid 'additional rate' tax at 50%. For those who do pay tax at 50%, he assumes their average income is the average income of the whole group, but that will be a slight underestimate.)
Second, dividends. My calculation is wrong because I implicity assumed that all taxable income is taxed at the earned income rates. In fact dividend income is taxed at lower rates, at least for the purposes of the HMRC table. This is important, because some high earners own their own companies and pay themselves in dividends rather than salary (I'll go into this in detail later).
Strangely, Murphy seems unaware of the different tax rates for dividends, although he is well aware of the tax avoidance potential, having used the method himself. He mentions the dividend tax rate nowhere in his 35-page report, and says on page 6, writing about deferring tax by delaying dividends from one's own company "These profits can then be left in the company in the hope that one day either the 50% tax rate is abolished or that the company owner’s income has fallen so that the profits can be taken out as a dividend and only be taxed at 40% at most."
The tax treatment of dividends is even given in a footnote to the HMRC table: "Dividend income is charged at 10 per cent up to the basic rate limit of £35,000, 32.5 per cent above £35,000 and 42.5 per cent above £150,000." These will be the rates used for the tax estimates in the table.
So what effect does this dividend tax rate have on the calculation, which I remind you is of how much less tax would be collected if the 50% rate were moved back to 40%, but there was no resulting elasticity effect on reported incomes? Well, that depends exactly what you mean to calculate. If you mean quite literally "How much of tax is paid at 50%" - the quote is from Murphy - then no tax on dividend income is paid at 50% so the answer should be reduced according to how much taxable income above £150k comes from dividends. If you mean, as would be very reasonable, "What would be the effect of cutting top tax rates by 10% (including cutting the top dividend rate back to 32.5%)" then the division between earned income and dividend taxation has no effect on Murphy's analysis (whatever the rate, the 10% change would affect all taxable income over £150k equally - this is an advantage for Murphy's method). It would increase my number, because I am underestimating the amount of taxable income over £150k by assuming too much tax is collected at lower levels.
If you relax the no-elasticity requirement and mean "What would be the effect of cutting the top income tax rate to 40% while leaving dividend taxes unchanged and allowing company owners to switch their income" then it's reasonable to assume that many company owners would take income in salary rather than dividends, and make the reduction in earned income tax collected much smaller - this seems to be the assumption in HMRC's table of "illustrative changes" and at least partly explains why it suggests such a small effect for a tax cut.
However, none of these is really the right question, because we are calculating notional numbers in HMRC's table of projections, not the actual projected income in tax and national insurance. The difference arises because some of the dividend tax comes from a "dividend tax credit". In reality, dividends from UK companies are deemed to be paid net of this notional 10%, so the rates paid in respect of dividends received are 25% above £35,000 and 36.11% above £150,000. The dividend tax credit is in recognition of the corporation tax already paid on company profits, which ranges from 20% at £300,000 profit or less to 26% at £1.5mn profit or more. Hence the marginal tax collected if you sell your services through a company and pay yourself in dividends is 48.89% at £300,000 per year gross and 52.72% at £1.5m gross. If the top dividend tax rate were reduced to 32.5% those marginal tax rates would become 40% and 44.5%. So if the true question is "What would be the effect on total tax take of cutting the top earned income and dividend tax rates by 10%", the cut in tax collected from people taking their income in dividends is about 8.5% of corporate earnings, or 11.11% of dividend income.
Incidentally, for company owners, paying yourself in dividends is always better than salary because a salary attracts tax at a marginal rate of 50% plus employee's national insurance at a marginal rate of 2% plus employer's national insurance at 13.8%, so that the marginal tax rate increases with salary asymptotically towards 57.82% of what it costs the company.*
It's interesting to look at the "income gap" in the HMRC table - the difference between the taxable income that would lead to the projected tax take if there were no allowances and all income were taxed as earned income, and the projected total income.
The 'Implied Salary' column shows the average salary that would be have to be earned with no tax reliefs to pay the amount of tax given - in the first two rows the calculation is done separately for the Higher Rate (40%) taxpayers and the Additional Rate (50%) taxpayers, and a weighted average taken. The 'Gap' column is the difference between the Average Income and the Implied Salary. The 'Implied Dividends' column shows what average dividend income would have to be received if all the income was taken in dividends.
I find it surprising how small the income gaps are other than for people with incomes over £1m. The maximum tax-free pension contribution in 2011-12 is £50k, however this includes contributions made by one's employer. The other significant sources of tax relief are Venture Capital Trusts and Enterprise Investment Schemes: in 2011-12 they will be responsible for an estimated £410mn of tax reductions between them. Even if that were all divided between the 14,000 highest earners it would still save them only £30k of tax each, equivalent to £60k of income. Evidently, the 'Gap' for the highest earners cannot be explained by tax reliefs alone: there is a lot of income being taken as dividends.
To sum up: Murphy's calculation is not very meaningful, as he himself tacitly acknowledges in his arbitrary downward adjustment of his estimate, but it's not in itself badly wrong. However, he does not use the available data to best advantage, and he is remarkably imprecise about tax rates and how the tax system works. As he advises Ryan Bourne on his site "you really do have to improve your maths".
It would be possible to mine the data more deeply to estimate what assumptions HMRC is making about the division between earned income and dividends. But it hardly seems worth the effort, since these are just projections: rather, I'd like HMRC to publish a more detailed breakdown next time.
*Nothing in this post should be taken as tax advice.