Lord Agnew resigned yesterday suggesting in the process that government losses to fraud were at least £29 billion a year. I am afraid I have to disagree with him. I believe that it is much larger than that because HM revenue and customs persistently understate the UK, which is a process assisted by the Office for National Statistics refusing to recognise the scale of fraud within the economy when estimating our gross domestic product.
I last analysed this situation in any depth in 2019. As I noted then:
I have explained time and again why HMRC tax gap data is, by definition, wrong. Try this, for starters if you like: nothing has changed since I wrote it in 2014. HMRC's methodology is still as inappropriate now as it was then.
Let me summarise why. First, HMRC's estimate of the VAT gap tends to be sound: it is based on macroeconomic data. It usually suggests that about 10% of VAT is not paid. And I suspect that is fair for reasons I note below. But then HMRC ignore the implication of this in the rest of their work. If you lose 10% of tax on the top line of any set of accounts (which is what is subject to VAT) then that is the minimum that you will lose in other taxes like income tax, national insurance and corporation tax as well. Why? Because there is no way to reintroduce that missing income back into the profit and loss account thereafter once it has been evaded at the top line level so there must be tax losses throughout the entire income statement as a result. But this obvious logic evades (pun intended) HMRC. As a result all their other tax gaps must be understated.
Second, HMRC instead base their estimated losses on the very limited number of audits that they undertake of tax returns each year. There are big problems with this. First, some samples are far too small and so they have to resort to 'illustrative estimates' to publish tax gaps instead. These estimates account for about one-third of all the tax gap, and always seem incredibly low. Second, tax audits are not 100% accurate by a long way: not all tax need be recovered. So HMRC audit inefficiency is built into the estimates as a result. And third, they simply ignore the fact that 800,000 or so income tax returns and at least 400,000 corporation tax returns are simply not submitted each year: the estimates for losses from these are negligible but, most especially with regard to companies may well account for a considerable part of the tax evaded. The estimates of evasion are vastly understated as a result, and HMRC seems ridiculously comfortable with this.
And third, when it comes to tax avoidance some key elements of this activity, like corporate profit shifting by the likes of Google, and losses because of artificial incorporation to save national insurance, are ignored by HMRC when preparing their estimates because HMRC considers them to be legal. This, of course, is absurd: all avoidance is legal but HMRC use a very narrow definition of artificial tax avoidance that has to be notified to HMRC as the basis for this estimate, which means it too is inevitably, and by deliberate intent, massively understated.
So let me suggest an alternative estimate of the UK tax gap in advance of HMRC's claim. This is based on the work I did for the European Union Socialists and Democrats group of MEPs in the EU parliament published in January this year, here. That report started by looking at available estimates of shadow economies. Two academic estimates were used, plus that which in underpins the EU official VAT gap estimate. The latter has a very different basis of estimation to the academic ones. The results were not entirely consistent, of course, but over Europe as a whole they were, as was also the case for the UK. There the estimated size of the shadow economy varied between 8.32% of GDP and 10.88%, with the EU official estimate interestingly being the highest. To provide balance I averaged them. This suggested that the UK shadow economy might be as large as 9.63% of GDP. HMRC said that in 2015, the year to which my data relates, the VAT tax gap was 9.8%. I think we can agree that the difference is pretty insignificant.
What is not clear is whether this sum is included in GDP, or additional to it. A small shadow economy estimate is in GDP: the last time I made enquiry of the ONS on this issue it was about 1%. In that case what this data very clearly implies is that our GDP is bigger than reported because of the scale of the shadow economy. In other words, the vast majority of this sum should be subject to the whole range of taxes owing. I return to the implications of this issue again, below.
To then estimate the tax lost I have to firstly retranslate the UK data back to sterling (the EU Report was in euros throughout) and then multiply it by the average tax rate calculated as the proportion between total tax take and declared GDP for the UK. The exchange rate for the year in question was €1 = £0.72584. The tax yield as a proportion of GDP for the year was 33.1%. The resulting tax gap due to the shadow economy was, based on a euro UK GDP of €2,602bn, between £60.2bn and £66.7bn. The former assumes the shadow economy is within GDP and the latter that it is outside it, as seems much the better assumption. Eurostat has confirmed that the latter is the case to me, albeit informally.
Is this estimate for tax evasion fair? I argue that it is, as far as the domestic economy is concerned. It is consistent with HMRC VAT data for the declared economy. And although some of the shadow economy may already be in GDP there is also tax evasion in taxes not in the GDP base e.g. wealth taxes. Overall I would argue that this is the most reliable domestic estimate available right now and very likely to be understated because all taxes are not charged on bases within GDP.
I deliberately added the word ‘domestic’ in the above paragraph. As Gabriel Zucman has shown there is now considerable evidence that by far the biggest tax evaders in most countries are the very wealthiest, who evade on a scale all of their own. Based on his work I have estimated that evasion by the very wealthiest in the UK may amount to up to £7 billion a year. I am willing to reduce that a little now: automatic information exchange from tax havens may have cut this to £5billion a year. And as this will not impact the domestic economy or GDP it is additional to the estimate I have already made.
These estimates do not, of course, include avoidance. So how much is that?
Based on the work of Alex Cobham and Peter Jansky, in turn based on IMF based estimates, corporate profit shifting from the UK may be a surprisingly low £1 billion a year (other countries, like Germany and France, suffer a great deal more).
And we know that the Treasury has admitted that tax incentivised incorporation costs maybe £3 billion a year.
These figures are outside the scope of HMRC estimates. How much more is there? It’s hard to say. But if HMRC say that there are avoidance costs of at least £1.7bn in previous reports - which will not include the above two figures for reasons already noted. In addition they suggest that legal interpretation disputes that they lose - which is avoidance by any other name - costs at least £5.3 billion a year. For the sake of argument I will take these at face value, but suspect that they drastically understate avoidance on issues such as inheritance tax and capital gains tax. Combine these figures though and the total comes to £11 bn for avoidance.
Then add in HMRC’s admitted rate of non-payment of £3.4 billion, which is tax declared but not paid, and the £3.2 billion not paid due to errors, which would not show up in shadow economy data as this relates to declared activity.
Now we have a total tax gap of:
- Domestic evasion £66.7bn
- Evasion by the wealthy using offshore £5bn
- Avoidance £11bn
- Error £3.2bn
- Non-payment £3.4 bn
Or £89.4 billion. Call it £90 billion for the sake of argument.
This, I readily admit, is smaller than some of my past estimates.
In part this is because I have been generous with my assumptions on bad debts on this occasion.
And in part it is because it does seem likely that both corporate tax abuse and the size of the shadow economy may have reduced over time: there does seem to be evidence to suggest both are true. And I am delighted about that. I celebrate the progress that has been made, especially on international tax into which I and others have put so much effort over the years.
If the shadow economy is also a little bit smaller then I suspect that may well be because of efforts on offshore: I suspect these really are working now. I have allowed for that in my estimates.
But I strongly suspect that domestic tax evasion is an issue where HMRC is not winning. And if their data continues to deny the scale of this problem then they are continuing to fail us all. And that will still not do. £90 billion is equivalent to more than half the cost of the NHS a year, and almost 90% of the education budget. We could have a tax system that is so much more equitable if only we invested more in HMRC. It’s tiresome to have to still argue for this, but it seems I must.
Why is it that government decisions in the UK continue to deny us tax justice? And for how long must this continue? I wish I knew the answer to that. But please remember it when reading HMRC’s latest pleading on its supposed efficiency, which will be wholly misplaced.
I accept these are estimates of course, but so are the government's figures estimates. GDP is an estimate, come to that. So estimates are all we deal with here. But what they imply is that fraud is vastly greater than the figure Lord Agnew uses but that, as he suggests, a blind eye is turned to it but deliberate government choice.
Why is that?
And at what cost to us all does this fraud happen?