{"id":6974,"date":"2022-05-27T13:50:15","date_gmt":"2022-05-27T12:50:15","guid":{"rendered":"https:\/\/www.taxpolicy.org.uk\/?p=6974"},"modified":"2022-11-09T09:13:13","modified_gmt":"2022-11-09T09:13:13","slug":"crs-evasion","status":"publish","type":"post","link":"https:\/\/heacham.neidles.com\/2022\/05\/27\/crs-evasion\/","title":{"rendered":"Tax Policy Associates report: UK taxpayers have \u00a3570bn in tax haven accounts, and HMRC has no idea how much of this reflects tax evasion"},"content":{"rendered":"\n

FOIA requests made by Tax Policy Associates reveal that \u00a3570bn is held in tax haven bank accounts by UK taxpayers, but HMRC has made no attempt to estimate how much of this is undeclared in UK tax returns, and therefore reflects tax lost to criminal tax evasion.<\/strong><\/p>\n\n\n

The background<\/h2>\n\n\n

Since 2018, almost all UK residents with overseas bank accounts have had their name, tax ID, and the balance and income on those accounts automatically reported to HMRC every year. That’s part of a revolutionary global project – the OECD Automatic Exchange of Information<\/a>\/Common Reporting Standard (CRS), under which \u20ac10 trillion of accounts were reported worldwide in 2019<\/a>.<\/p>\n\n\n\n

This should be a bonanza for HMRC. We’re all (except non-doms<\/a>) supposed to declare<\/a> income from foreign accounts in our tax returns, with up to 300% penalties<\/a> if we don’t. So if someone has foreign account income reported under CRS which wasn’t included in their tax return, and they’re not a non-dom, HMRC should be able to immediately identify potential tax evasion.<\/p>\n\n\n\n

The amounts are very large – in 2019 UK taxpayers had over \u00a3850bn in foreign accounts, of which \u00a3570bn was in tax havens – see chart above (and my definition of “tax haven” below). And the average account size in tax havens is considerably larger than the average in the rest of the world1<\/a><\/sup>A tax professional with a huge amount of experience in this area made the excellent point that we need to be careful to distinguish actual account holders vs UK controlling persons of companies (“Passive NFEs”) are accountholders. Otherwise we can be double or multiple counting – i.e. because one company can have multiple controlling persons, and some of those (trustees) won’t be taxed on the account. My FOIA should have extracted the accountholders only, but absent full disclosure by HMRC it is possible they have mistakenly given me UK controlling person data as well<\/span>:<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

Update<\/strong>: the FT story on our report is here<\/a>.<\/p>\n\n\n

Why is it important to have an estimate of the proportion of the \u00a3570bn that reflects tax evasion?<\/h2>\n\n\n

Because the scale of the estimate has enormous public policy implications.<\/p>\n\n\n\n

If 30% of the accounts are undeclared (as some<\/a> have previously suggested), then it’s a massive scandal and immediate action of the most serious kind will need to be taken. I’d say the same if the figure were 3%.<\/p>\n\n\n\n

If, on the other hand, it’s 0.1% then all we should expect is efficient and effective HMRC enforcement. <\/p>\n\n\n

The options<\/h2>\n\n\n

When HMRC started receiving this data back in 2016, it had several options:<\/p>\n\n\n\n

  1. The traditional approach<\/strong>. Build a computer system to cross-check the CRS data with everyone’s tax return, to see who has an offshore account on which they’ve received income that wasn’t declared on a tax return. Version 1.0 doesn’t need to be very expensive or complex – just pull out matching tax IDs (national insurance numbers or UTRs<\/a>) from the two sets of data, where CRS data indicates large tax haven accounts but nothing is declared in the “offshore income” section<\/a> of the tax return. See what comes out. Build up the complexity and sophistication from there.<\/li>
  2. The quick option. <\/strong>Set a bunch of junior tax inspectors to work manually pulling out a random sample of say 1,000 high-value individual CRS accounts in tax havens, locate those individuals’ tax returns, and look for anything suspicious. Then apply statistical tools to the result so you can estimate the overall level of non-compliance (albeit with significant uncertainty). Use that to inform whether you build a big system to automate this, and how you do it. Or maybe you find it’s slim pickings, and so just continue random manual checks for the largest accounts.<\/li>
  3. The nerd option.<\/strong> Anonymise all the CRS data, hashing<\/a> names, addresses, and tax IDs. Do the same with the overseas income section from tax returns. Hand both datasets over to a bunch of smart academics and data scientists, so they can do all the hard cross-checking for you. For free.<\/li>
  4. The Thatcherite option. <\/strong>Do the anonymising thing again. But ditch the nerds – instead, offer private sector companies the chance to crunch the data and identify tax evaders, in return for payment of 10% of tax evasion recoveries. The nerds could apply too.<\/li>
  5. The implausible option. <\/strong>Don’t do any cross checking, automated or manual, large scale or small scale. Make no effort to match up CRS data with tax returns.<\/li><\/ol>\n\n\n\n

    Right now it’s unclear quite what HMRC have done – they’re refusing to say. They’ve told the FT’s excellent Emma Ageyemang<\/a> that they “systematically” check the CRS data, but won’t reveal more. We can speculate they’re looking only for huge anomalies (e.g. millions in CRS accounts but nothing in a self-assessment return) but not cross-checking everything (or they would surely say so). <\/p>\n\n\n\n

    That speculation is consistent with the letters<\/a> we know HMRC have sent to people with foreign accounts disclosed under CRS, reminding them of their responsibility to declare foreign income. HMRC has made clear these letters are sent<\/a> on the sole basis of the CRS data, and without cross-checking with actual tax returns.<\/p>\n\n\n\n

    What is clear is that HMRC have made no attempt to estimate the scale of the offshore tax evasion problem, or even determine if it is a problem. The FOIA response says<\/a>:<\/p>\n\n\n\n

    \"\"<\/figure>\n\n\n

    Why?<\/h2>\n\n\n

    We’ve had several justifications from HMRC.<\/p>\n\n\n\n

    First<\/strong>, in the FoIA response:<\/p>\n\n\n

    “A number of the accounts reported through the Common Reporting Standard (CRS) are not actually chargeable to UK tax, nor is every account of a suitable value to impact on a persons’ UK tax position”<\/em><\/p>\n\n\n

    These, and many other factors, are why the CRS data has to be used with care. They are not reasons to discard the statistical value of the data altogether. You’d expect any kind of audit or cross-check to exclude non-taxable accounts and low-value accounts. (Although the obvious reason why some accounts of UK residents are not taxable is that they are held by non-doms, and given HMRC normally receives no data at all on non-doms’ foreign assets, you’d think this itself would be very useful data.)<\/p>\n\n\n\n

    Second<\/strong>, when speaking to the FT, HMRC said:<\/p>\n\n\n

    Not all of the worlds\u2019 jurisdictions are signed up to CRS and we could not say with certainty whether each account was \u2018properly disclosed\u2019. We would not publish a figure where we did not have certainty that is accurate.<\/em><\/p>\n\n\n

    Here HMRC seem to be taking the odd position that no estimate can be made unless it is certain to be accurate. That is not what the word “estimate” means, and statistical methods have been developed to deal with errors for at least two hundred years<\/a>. <\/p>\n\n\n\n

    HMRC is certainly familiar with producing estimates which are highly uncertain – they do so very competently in their annual tax gap report<\/a>. This contains estimates for tax evasion and avoidance which have very large uncertainties – HMRC describes them as the \u201cbest estimates based on the information available\u201d. And that’s fine. What’s not at all fine, and in fact inexplicable, is using uncertainty as a reason for making no estimate at all.<\/p>\n\n\n\n

    Furthermore, the stated reasons for uncertainty do not seem very hard to overcome. Whilst there are some jurisdictions <\/a>not signed up to CRS\/FATCA, they are generally ones where rational tax evaders would not risk keeping funds. They’d make any estimate a baseline\/lower bound, but that would not diminish its usefulness.<\/p>\n\n\n\n

    “Properly disclosed” is also not terribly challenging to overcome. For example: focus on CRS accounts where the income\/gains is larger than the annual exemption, and where there is a straightforward national insurance number\/UTR match with the tax return of a taxpayer who is not a non-dom. Again that would result in a lower bound estimate, but still a useful one.<\/p>\n\n\n\n

    So HMRC’s explanation for the lack of any estimate doesn’t make sense.<\/p>\n\n\n

    Can we use other sources to estimate how much of the \u00a3570bn reflects tax evasion?<\/h2>\n\n\n

    I don’t think we can – which is why HMRC’s failure to product estimates is so unfortunate.<\/p>\n\n\n\n

    I am very sceptical of some of the claims<\/a> that have been made that over 30% of offshore accounts are undeclared. Tax authorities have now had the CRS data for five years or more, those that aren’t asleep at the wheel have been cross-checking against tax returns, and there have been a handful of prosecutions rather than thousands. And the existence of CRS means that pre-2016 estimates are of limited value.<\/p>\n\n\n\n

    It’s also important to remember that there can be many entirely legitimate reasons for having a financial account in a tax haven:<\/p>\n\n\n\n