$190bn and counting – measuring offshore tax losses


I am delighted to welcome a guest post by Iain Campbell who is a member of the National Committee of the UK Association of Revenue and Customs (ARC). While I have  previously looked at the ‘big numbers’ that shape the debate on corporate tax avoidance, here Iain takes a close look at an estimate relating to offshore tax evasion by individuals:


People who read Maya’s blog probably need no introduction to the name or work of Gabriel Zucman. He has been actively pursuing the issue of offshore wealth and associated tax losses over a number of years and publications, such as The Missing Wealth of Nations (2013),Taxing Across Borders (2014) and The Hidden Wealth of Nations (2015).

I am particularly interested, as an elected official for ARC, a trade union whose members are the senior professionals in HMRC. Our members are naturally interested in discussions of tax compliance and what more can be done to tackle evasion and avoidance, as well as how to measure the scale of the problem.

We should be grateful to Zucman for pushing debate along in this important area. He has published his papers alongside extensive technical appendices and details of sources and calculations, which aim to allow readers to check or reproduce his results. He has also responded in a positive way to some e-mail questions I have raised – but clearly he has not agreed with everything I have said.

The big picture

Assets held offshore are by their nature difficult to assess, but include proceeds of crime, money laundering, corruption and theft of state assets, capital flight driven by concerns about domestic stability and forced appropriation of assets, assets that are evading taxes, and vehicles for legal tax planning, international investment and internationally mobile lifestyles.

The nature of the origin and the motives for assets in tax havens is, to me, likely to have implications for things like the level of returns, the tax lost (or the overall amount that belongs rightly to the state).

What has caught most attention from Zucman’s work is his $190 billion estimate of the tax lost to evasion after the assets have gone offshore:

Screen Shot 2016-05-26 at 11.11.22


[Source: Taxing Across Borders, Zucman, 2014]

As this number is being widely used (such as the basis for figures in the run-up to the recent Anti-corruption Summit) it is worth being clear about what it implies. This blog post is based on my own efforts to understand the basis of Zucman’s estimate, and provides a commentary on the assumptions that underpin it; in particular the rate of return on the wealth and the methodology of calculating the tax, as well as the consistency between the calculations for previous years.

Where does the $190 billion come from?

Offshore assets

Zucman starts by estimating the total amount of private financial wealth held offshore – which he puts at 8 percent of global household financial wealth. This is based on an analysis of anomalies in national economic statistics. He then allocates the total amount ($7.6 trillion) between countries, using global offshore ownership patterns that he infers for Switzerland (using data from the Swiss National Bank) as the general pattern. This seems like a reasonable approach to estimating the unknown, but it is worth noting that it is a top down model based on limited evidence. So that when he estimates, for example, that 30 percent of the financial wealth of households in Africa and Latin America, and 50 percent of those in Russia are held offshore, this is not based on country-by-country evidence of where assets are held and who they belong to, but rather on extrapolating from Switzerland to the global level.

Screen Shot 2016-05-26 at 11.12.48

[Source: The Hidden Wealth of Nations, Zucman, 2015]

Zucman uses two key sets of assumptions to calculate the amount of tax revenue lost because these assets are offshore; firstly the proportion of offshore assets that are undeclared, and secondly the value of the income, wealth and inheritance taxes which would have been payable if they were.

Declaration rates

The fraction of offshore money that evades taxes is a critical assumption. Zucman states that on average, globally 80% of offshore wealth goes undeclared. This estimate is based on data from the Swiss tax administration on the amount of interest which European asset owners either self-report to their home countries, or that is taxed at source by the Swiss. By adding in some assumptions about the rate of interest (2%) and the total amount of European money in Switzerland he concludes that only around 22% of the total amount held by Europeans in Switzerland is declared by one of these two routes, and that the rest is undeclared. He then applies this across offshore assets globally.

Zucman also drew to my attention a paper concerning Swiss Central Bank figures on the total amount of wealth held by foreigners in Swiss banks, and some indication of who owns it. On that basis relatively little wealth seems to be reported to domestic EU tax authorities. (Zucman also assumes that for the UK only 25% of deposits belong to taxable UK residents, and 75% to non-taxable UK residents).

This 80% ballpark figure, however does not take into consideration issues such as account holders having dual residence/non-residence. Other potential figures may provide a counterpoint to the 20%. In the case of the HSBC “Swissleaks” the UK tax authority identified 3,600 Swiss account holders, said that “In around 2,000 cases, we found no evidence of evasion”. We know that HMRC has been criticised for its work but there is very little other hard information to suggest a different (and higher) UK rate. The HMRC figures give 2,000/3,600, or say 55% compliance.

The UK may be an outlier in terms of data because of the rules for non-domiciles. But it does represent an investigation by a tax authority and other countries do not seem to be prosecuting in any significant numbers. They may be settling on a civil basis which to me raises questions about evasion on the 80% level. Countries may not have the capacity to do large numbers of prosecutions but there may not be the evidence needed to prove evasion. There is also some emerging but diverging evidence from the Panama papers. On the one hand, the Colombian Ministry of Finance estimated the overall non-compliance rate for its citizens was 65%, on the other, an early Press report suggests that the Indian tax authority estimated 10% (based on an Indian law on Liberalised Remittance Scheme).  I think a 20% rate of disclosure can be no more, at this stage, than a very early and provisional estimate.

Value of tax losses

Zucman calculates the value of tax losses from income tax, wealth taxes and inheritance tax.

For income tax he assumes that financial assets held offshore are earning average rates of return of 7% (in Taxing Across Borders) and 8% (in The Hidden Wealth of Nations) and ought to be paying tax at the highest marginal rate in each country. To establish the rate of return he draws on benchmark returns from Vanguard. The 2015 technical appendix says, “Vanguard provides a large number of benchmark indexes, with 1, 3, 5 and 10-years return. As of the summer of 2013, all composite indexes show large positive returns over a 10-years period, typically in the 6-8% range. These are nominal returns, before tax and before fees. They are well in line with the 6% return average nominal return on offshore accounts I retain.” The Vanguard link indeed shows there are numerous indices, many of which are not in the 6-8% range. This is even more so for recent investments where some one year returns have even been negative. So I think the 8% is definitely an upper bound (Indeed in the calculation of the amount of money offshore outlined above, Zucman assumes interest rates of 2% for Swiss deposits, which would mean that dividend bearing funds would have to be even higher).

I think taking the upper bound on 10 year returns on a broad, diversified funds and using this to cover a mix of short and long-term deposits and investments is likely to overstate returns in real life where low interest rates, management fees, and the mix of stronger and weaker performing funds must be taken into account.

Another factor to bear in mind is the expectations and behaviours of the people who have placed their wealth offshore. John Henry who has developed an alternative estimate of offshore wealth for the Tax Justice Network, places greater emphasis on the use of tax havens to hide illicit gains. He argues that users of tax havens value secrecy and security far more than market gains and that Switzerland has highest cost banking services among major countries because many of its customers are indifferent to price. (Zucman’s analysis of wealth is not restricted to that held in tax havens. The figures for offshore wealth total all the wealth that is held outside of a person’s residence country, not by a list of tax havens.)

On estate/inheritance tax Zucman applies the tax rate in each country to 3% of the stock of undeclared offshore wealth. The 3% is based on an assumption of the mortality rate of offshore account holders. This is clearly a ballpark estimate which might change given more information on the age profile of the wealthiest, and their longevity. (If the assets are held in Trusts, these may not be affected by the death of any beneficiary.) It looks as if Zucman has applied the highest marginal rate. However I think this is likely to produce an overestimate as it is well known that inheritance/wealth taxes are hard to collect and often relatively easy to avoid, with exemptions, deductions, tax free gifts, etc

How much confidence can we have in this estimate?

Zucman has undertaken important and painstaking work to try to triangulate the unknown from publicly available data. He acknowledges there is a lot of uncertainty. On one hand his $190bn estimate only captures financial wealth (based on failure to declare interest, capital gains, and dividends, plus estate/inheritance tax evasion on those accounts). It disregards real assets such as works of art, jewellery, gold and real estate that would raise the total. On the other hand the estimate of tax loss is highly sensitive to the assumptions that 80% of offshore assets are undeclared, and that they are at the same time earning up 7 or 8% annual returns. If instead we assumed that 60% of offshore assets were undeclared, and average returns were 4%, for example, the overall total tax loss estimate would be halved – but still a very large number. The following chart shows the range of outcomes if the declaration rate, or rate of return, are varied.


Large numbers attract attention. Zucman’s work suggests widespread evasion by a few, while paying little attention to the possible additional sources of crime, money laundering and kleptocracy. It may be that $190 billion is an overestimate for the problem which Zucman has focused on, but at the same time there could plausibly be larger amounts of assets which should be repatriated from the proceeds of corruption/theft of public assets, or from domestic tax evasion where a tax haven nest egg has been built up by illegally diverting pre-tax profits.

While Zucman’s work and his openness about his methodology and calculations is welcomed, this means that changes in the methodology are easy to detect and can themselves generate questions. As this analysis highlights the total from his calculations is highly dependent on a few key assumptions. Between 2014 and 2015 Zucman updated his spreadsheet to take account of various changes:

  • He changed the assumed global non-declaration rate (in 2014 he set the declaration rate at 10% for most countries and 100% for gulf countries where there is no income tax. In 2015 he changed this to a 20% declaration rate across the board)
  • From 2014 to 2015 the nominal rate of return on capital increased from 7% to 8%
  • Some changes were made to inheritance/estate tax (i.e. wealth tax in Spain and marginally higher rates in France and Norway

I intuitively feel that the net effect of these changes should be to produce figures for 2105 that are not the same as 2014. However the resulting headline figures remain the same – suggesting a confident core estimate rather than a wider range of uncertainty.


Readers may be assured they have now reached the end (and hopefully have gained a clearer understanding of how the new ‘big number’ on offshore tax loss has been constructed). There is a lot of very interesting data and analyses, with illuminating material, but some big open questions.

From ARC members’ perspective it is probably enough to know there is likely to be sufficient work to keep us all busy for a quite a while and maybe even enough to justify some more of us?

However it should not be assumed that if Governments had more tax revenues they would invest in schools, hospitals, and so on. In the last parliament HMRC brought in c£100 billion additional yield. But this was used to enable tax cuts, reduce debt interest, etc. Recovering the $190 billion is no guarantee on how it would be spent.

Maya has always been arguing for data led analysis, and better data, so I think she and her readers would hope for more, and more reliable, data. As we move to greater automatic information exchange, data mining and analytics, I think Governments will be increasingly better equipped to tackle issues related to unacceptable uses of offshore accounts. But it would be good if researchers could build on Zucman’s analysis, identify other ways of measuring the compliance/non-compliance rates, and the different types of sources and uses of offshore funds. There is a clear question to be asked of national tax authorities as to whether they feel they have the ability to gather this sort of data and, if not, are they proposing to do anything to rectify this.


2 Responses to “$190bn and counting – measuring offshore tax losses”

  1. I want to thank Ian for undertaking the Herculean task of reviewing Mr. Zucman’s published work. Also Kudos for Maya, as I assume that along with hosting the result of his work, probably contributed here and there.

    As Maya continually points out, having a reasonable estimate of the extent and source of the problem of tax evasion is a necessary first step in coming up with realistic ways of battling this scourge. Too often useful discussion and debate is stillborn by enormous by distracting headlines containing completely unsubstantiated numbers.

    Ian’s analysis may not be the end of the debate on estimates, but it has at least moved the ball from the position that Mr. Zucman originally staked out.

  1. 1 How much money has been recovered as a result of the Panama Papers? | Hiya Maya

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: