On 17 February the editor of Taxation magazine, Mike Truman, wrote an article under the title ‚ÄòLack of Evidence’, the summary of which said:
The claim that poor countries lose $160 billion in tax from ‚Äòtransfer mispricing’ has been repeated endlessly. MIKE TRUMAN finds it hard to justify
- Two Christian Aid reports claim $160 billion tax lost.
- Raymond Baker’s 7% claim does not relate to TNCs.
- Problems of methodology in Simon Pak’s study.
- Real shortfall is homegrown tax evasion.
Raymond Baker of Global Financial Integrity, Alex Cobham of Christian Aid and I wrote a response, published this week. It is entitled ‚ÄòLack of Will’.
That response is behind a paywall and so is not on public record, even though the critical article is.
There is also apparently an on-line debate going on about the issue – which none of us can read or contribute to as it is also behind a paywall. So much for debate. In the circumstances I think it entirely appropriate to republish our response, below. I leave it to others to work out the ethics of publishing criticism on open pages and denying response and debate a similar airing.
Lack of will
Transfer pricing abuse is a massive global problem, argue
Richard Murphy , Alex Cobham and Raymond Baker.
Mike Truman, in his comment article ‚ÄòLack of evidence’, Taxation, 17 February 2010, page 6, questioned work we have, in various and different ways, undertaken to estimate the loss arising to developing countries from transfer pricing abuse – or transfer mispricing as we prefer to call it.
We think Mike is saying three things. The first is that Raymond Baker’s work on this issue, published in his 2005 book Capitalism’s Achilles Heel, used an inappropriate interview-based methodology to establish a potential rate of transfer mispricing, which he anyway contends is now out of date.
Second, he challenges Christian Aid’s May 2008 report on transfer mispricing ‚ÄòDeath and taxes: the true toll of tax dodging’, which suggested that the loss to developing countries from transfer mispricing might be as much as $160 billion a year because that reports relies in part on Baker’s work.
Finally, Mike questions the findings of Christian Aid’s second report on the subject (published in March 2009), ‚ÄòFalse profits: robbing the poor to keep the rich tax free’, which relies on the statistical analysis of world trade data using a methodology developed by Professor Simon Pak of Penn State University.
Based on his analysis, Mike concludes:
¬? transfer mispricing is not the issue we claim it is;
¬? country-by-country reporting as proposed as one solution to this problem is not therefore as important as we claim it might be; and KEY POINTS
¬? Illegal flows out of developing countries could be up to $1 trillion annually.
Despite our high regard for Mike, we have to disagree with him on all counts, although in the space available cannot address all the issues he raises.
First let us deal with methodology. Raymond Baker in his book only examined mispricing in arm’s length transactions, i.e. between unrelated entities. Having done so, and based on personal experience, he concluded that while it was highly likely that the rate of mispricing was higher in related party transactions, he would only use the figure his interviews had established to be likely between unrelated entities. Three things should be noted as a result: first this is likely to be a conservative estimate. Second, research based on semi-structured interviews is considered entirely suitable as a basis for research in all social science disciplines, including taxation. Third, while now relatively old research, subsequent work has corroborated the findings .
That subsequent research includes new work published by Global Financial Integrity (GFI), a project Baker now directs. Its study of illicit financial flows, published in 2008, defined illegal flight capital as funds intended to disappear from record in their country of origin, with the earnings on the stock of illegal flight capital outside of a country not normally returning to that country of origin.
The report recognised a number of mechanisms that that can be used for this purpose, of which transfer mispricing was just one. As it noted, since this activity is illicit, available data with which to assess its scale is oft en incomplete or inaccurate: the work accepted that risk, as do all other studies in this area. That said, GFI used several methodologies and databases to estimate both the legal and illegal components of flight capital, including the Hot Money, Dooley, and World Bank residual methods, IMF Direction of Trade Statistics, and the International Price Profiling System. All are widely used, recognised and considered by those bodies that have given their name to some of them as the best available methodologies.
Based on this work, GFI estimated that illicit financial flows out of developing countries are some $850 billion to $1 trillion a year. We believe this estimate is conservative. It does not, for example, include transfer mispricing within the same invoice, which cannot be picked up in mispricing models based on IMF Direction of Trade Statistics.
Such mispricing is entirely possible within multinational corporations which do not need to rely on reinvoicing. Nor does it provide any estimate of the loss due to transfer mispricing on services or intangibles, which are perhaps more open to abuse given the difficulty in identifying comparables to establish an accurate arm’s length price.
The IMF Direction of Trade Statistics on which the estimate of transfer mispricing is primarily based measures the difference in exports out of one country and imports into another country for all pairs of reporting countries. After subtracting the cost of freight and insurance, the only way to get a difference in export and import prices (other than mis-entering the data which might itself be indicative of mispricing) is to reinvoice, for example through tax haven locations. It is this reinvoicing that the GFI data records meaning that mispricing within the same invoices would have to be added to these figures to get a more accurate analysis of total mispricing.
Transfer pricing abuse
The GFI report in 2008 estimated that at least half of all illicit financial flows out of developing countries involved transfer mispricing. In February 2010 a further GFI report, ‚ÄòThe implied tax revenue loss from trade mispricing’ sought to quantify the tax loss arising from these illicit flows and concluded that the average tax revenue loss in developing countries was between US$98 billion and US$106 billion annually over the years 2002 to 2006. This figure represents an average loss of about 4.4% of the entire developing worlds’ total tax revenue.
The methodology used is one some commentators will challenge: it assumes that the identified flows of transfer mispriced funds would have been taxed at the marginal corporate tax rate of the location they fl owed from. This ‚Äòtax gap’ methodology, developed by Richard Murphy, has been challenged by some as misleading since its opponents argue that it ignores the availability of reliefs and allowances that might have reduced the effective tax rate below the nominal tax rate.
We do not agree for two reasons. First, if those reliefs had been available in respect of these profits, it would have been rational to have used them. We assume we are dealing with rational entities. They were not used, so presumably they were not available, meaning that tax would have been paid.
Second, to assume that the allowances and reliefs that multinational corporations enjoy in developing (or other) countries are independent of their considerable economic power in such places when negotiating inward investment, or are even independent of other illicit financial flows such as those resulting from bribery, is untenable. Numerous reports, including some by the authors of this article, for Christian Aid, Global Witness and others attest to this fact. As such we suggest that the methodology records a potentially recoverable loss, and that is its purpose.
Simon Pak’s approach to this issue is different from Raymond Baker’s. Christian Aid notes the OECD estimate that at least 60% of world trade now takes place within multinational corporations rather than between arm’s length bodies. For the years 2005-2007, Simon Pak analyses data on all bilateral trade on commodities with the US and European Union to determine the extent of losses arising on this intra-group trade. The US and EU provided the data for this purpose.
The data is the most granulated available: so detailed that HMRC would not provide it directly for the UK because identification of individual trades was possible in too many cases. 83.7 million EU trades were analysed by Pak in 2007, for example. Only data where price estimates per unit supplied could be calculated was used. By definition services are excluded, and given that the majority of transfer mispricing is now likely to be in this area this will result in any estimate we offer significantly underestimating total losses from this activity.
An important assumption in the price filter analysis method Pak uses on the resulting data is that the estimated inter-quartile price range per unit of product traded is an arm’s length price range. This assumption is suggested by some to be arbitrary. However, the assumption is considered reasonable as the US Internal Revenue Service transfer pricing regulation, Internal Revenue Code 482, specifies that an inter-quartile range is an acceptable arm’s length transaction range. We believe that provides credibility to the approach used but we accept that the point is debatable, but then everything in statistical analysis is. This does not invalidate statistical analysis as the basis of much, if not most, academic tax analysis and in turn a great deal of tax policy worldwide.
Lost tax revenue on capital flows as a result of trade mispricing is then calculated on a country-by-country basis by multiplying the capital flow by corporate marginal tax rate for each country in question: this approach accords with that used by Baker/GFI, noted above and acceptable for the same reasons.
This approach is reflected in the second Christian Aid report noted above, but not the first. As that second report notes, the approach seeks to use Pak’s methodology to estimate how many imports to the EU and US from non-EU countries are underpriced, and how many exports from the EU and US are overpriced to facilitate illicit capital transfer from non-EU countries. In doing so it is likely to underestimate the losses, partly because services are not considered and partly since the techniques used will underestimate mispricing because over and under pricing is aggregated by the methodology. There is also the risk that averagely priced transactions may be mispriced. This possibility is not detected.
In contrast, it is accepted (and noted in the relevant report) that there is an opposite risk with regard to products with highly volatile prices, e.g. oil. There, averaging over an annual period,
as the method does, might produce errors. Across the whole spectrum of trade this is assumed to be a counter-balancing error, but it does also explicitly recognise that the issue raised by Mike Truman in his article is a matter of concern, but not one considered likely to be material.
The result of the work is an estimate of lost tax revenue from all non-EU countries to the EU and the US between 2005 and 2007 of £190.8 billion or about £63.6 billion a year ($127 billion a year at 2007 exchange rates). Given that this implied lost revenue is based on EU and US trade, and assuming that trade between developing countries and the rest of the world is characterised by a similar level of mispricing, Christian Aid extrapolated this figure to find it consistent with their earlier estimate of $160 billion globally.
All of the estimates reviewed fall in the range $100 billion to $160 billion a year. As yet unpublished research by Richard Murphy for the World Bank undertaken in 2009 shows it is plausible for transfer mispricing of this scale to take place within multinational corporations.
Our point now is to suggest that we are presenting broadly consistent estimates within a range. We are not claiming spurious accuracy. As other studies have shown, e.g. that of Clemens & Fuest for the Department for International Development in June 2009, no one outside the small circle of NGO researchers noted here has even sought to do this work. Many have sought to criticise it. We accept it is open to improvement. We also accept, as should any researcher, that the flaws in available data make the results offered estimates. We would however stress, that if the data is fl awed it is likely to be because of trade mispricing, not its absence.
We would also add that the direction of this flow should be noted: overall additional funds arrive in the EU and USA. They may be taxed there, usually at lower rates than would have been paid in developing countries. Many will come through locations such as Switzerland and Hong Kong and in case study after case study we have seen this to be true. This lets us immediately dismiss the main thrust of Bill Dodwell’s assessment of our work as implausible: we do not know of tax authorities which take transfer pricing cases to argue down their revenues. This is what would be required if Dodwell’s assessment assertion was to be correct.
That said, Christian Aid does also show a transfer of capital from the US and Japan to Europe. Given the use made by corporations from both locations of European holding companies to act as worldwide sales agents, nothing surprises us about this. Indeed, work by Martin Sullivan for Tax Notes in the US has long documented this trend, noting in 2004 that American companies were able to shift $149 billion of profits to 18 tax haven countries in 2002, up 68% from $88 billion in 1999. This strongly suggests that this direction of flow is correct, the strength of the transfer pricing regimes of those countries notwithstanding.
All this being noted, the important thing is to ask what does potential transfer mispricing of this scale from developing countries imply? First, the losses are, even if the lower end of the estimate range is considered, more than twice the sum required to pay for the United Nation’s Millennium Development Goals.
In other words, we believe that reducing (but not eliminating) transfer pricing abuse could eradicate extreme poverty and hunger, achieve universal primary education, promote gender equality and empower women, reduce child mortality, improve maternal health, combat HIV/AIDS, malaria and other diseases, ensure environmental sustainability and help develop a global partnership for development. If that is the case, the argument for inaction has to be very strong indeed.
Any action does, however, have to recognise the reality of taxation in developing countries. It is essential to bring the poor into the tax base, as it is likely to result in stronger engagement in political processes, and strengthen accountability between state and citizen.
However, in the short term, income taxation will have limited revenue impact given the weak economic base. Taxing a small elite of individuals, civil servants, major corporations, international trade and natural resources when present is likely to have a much greater revenue impact. To be effective the largest available flows must be taxed.
Stricter tax reporting
We suggest three things to ensure that these flows are taxed as effectively as possible. The first is that, and here we agree with Mike, significant technical support to tax authorities in developing countries is needed – as well as cash to ensure their best staff are not continually poached by the biggest firms of accountants.
Second, we argue for country-by-country reporting by multinational corporations. Mike is entirely wrong to say this cannot help. HMRC now publicly concede that country-by-country reporting by multinational corporations would increase tax yield in the UK. We do not however argue it is the solution to transfer mispricing: it is not. What it does is provide the data that can show whether pursuing a case is likely to be worthwhile. When resources are scarce, as they are in developing countries, this is vital. The tiny experience of transfer pricing litigation in Africa to date suggests that the simple absence of data on differing profit rates by location within multinational groups – data we think was deliberately withheld by those multinational corporations to assist their cases – is a major inhibitor to any chance of success on this issue. Country-by-country reporting would help provide this data.
Country-by-country reporting does much more: it is now seen as a key component in effectively tackling corruption in the extractive industries, for example. It is, therefore, a key component in tackling the very issue Mike says is an impediment to progress. It also provides enormous value to shareholders concerning the timing and location of tax liabilities that their company faces. To dismiss country-by-country reporting because it cannot solve transfer mispricing by itself is absurd.
Lastly we promote massive increases in the range and scope of information exchange agreements available to developing countries so they can secure the data they need to address issues on transfer mispricing, which also impacts revenues from royalties, sales taxes, export levies and more besides. Developing countries are almost entirely excluded from the tax treaty network. They start with a massive asymmetric information disadvantage as a consequence, which makes their current task almost insurmountable. This economic externality has to be removed if they are to have any chance of building successful states.
In these circumstances to suggest the problems faced are the result of home-grown tax evasion misses the largest part of the picture. Nothing but abuse by those unscrupulous businesses can explain the data differences we have consistently found. We can argue about the scale of the abuse but not its existence. Even then, suppose we had overstated the scale of this issue twofold and only half the problem could be effectively tackled using the mechanisms we promote. That would still eradicate extreme poverty and hunger, achieve universal primary education, reduce child mortality and improve maternal health while leaving some over to tackle AIDS and other major diseases.
Can anyone give a good reason why the tax profession would not want to do that when all the evidence suggests that tax compliance by multinational corporations – where tax compliance means seeking to pay the right amount of tax (but no more) in the right place at the right time, where right means that the economic substance of the transactions undertaken coincides with the place and form in which they are reported for taxation purposes – could achieve these aims?
We don’t know of any.