To Oxford, a reply

Posted on

As some might know, I and others have been something of a spat with members of the Oxford Centre for Business Taxation at the Said Business School at Oxford University.

Clemens Fuest, deputy director that Centre replied on the TJN blog, saying:

We would like to respond to the two preceding entries in this blog because we disagree with the arguments put forward and we think that it is easy to see why these arguments are wrong.

The main issue seems to be whether it makes sense to take into account what is measured as income shifting out of developing countries and to ignore what the same statistical approach would measure as income shifting into these countries. We restrict our response to this issue (although the other arguments raised in the first entry are equally misguided). In our report, we have made the following point:

The restriction to one direction of income shifting leads to misleading results if the findings are used to estimate the impact of income shifting on corporate income tax revenue collected, as e.g. in recent studies by Christian Aid (2008, 2009), cited in our report. A meaningful estimate of the tax revenue effects would have to take into account shifting in both directions.

It is helpful to demonstrate this using the following simple example: Assume that there are three exporters of a good in country A. Assume all firms have costs of 4 in country A which are deductible from the profit tax base in country A. Firm 1 exports the good at a price of 4, firm 2 exports the good at a price of 8 and firm 3 exports the good at a price of 12. The mispricing approach would identify the transaction at a price of 4 as underpriced and the transaction at a price of 12 as overpriced. The goods are exported to country B, where all three are sold at a price of 14 to consumers in country B.
In this example, the aggregate corporate income tax base in country A is equal to 12. Firm 1 shifts income out of country A and firm 3 shifts income into country A. In the absence of trade mispricing, the tax base in A would be the same. The tax revenue loss of country A due to mispricing is equal to zero. A method which only takes into account firm 1 and neglects the implications of mispricing by firm 3 is clearly misleading. The same applies to the impact of income shifting on country B.

What happens if other taxes are considered? For simplicity, consider the case of tariff revenue losses due to avoidance or evasion (the VAT example used in the first entry is misleading because one would have to take into account VAT paid by the importer; VAT cannot be saved by underpricing imports if the goods are imported and sold on to consumers, at a given price, by a firm which is subject to VAT itself). So assume that there is a proportional import tariff of 10% in country B. Given the prices 4, 8 and 12, the overall import value is 24, and tariff revenue would be 2.4. Now assume that mispricing disappears, i.e. firms 1 and 3 adjust their prices to 8. In this case, import tariff revenue would be 2.4 as well. This example shows that it is of key importance to account for both under- and overpriced imports to B.

How do other economists analyse these issues? The US economist Kimberly Clausing has analysed income shifting through trade price manipulation into and out of the US (Clausing, Kimberly (2003), Tax-motivated transfer pricing and US intrafirm trade prices, Journal of Public Economics 87, 2207-2223.) Her approach differs in many ways from the mispricing approach we have criticized in our report. She identifies mispricing by comparing trade transactions within multinational firms to trade transactions between independent firms. This work has been published in a peer reviewed academic journal, which has a very high reputation.

Clausing’s approach allows for the possibility of income shifting out of the U.S. as well as into the U.S.. She finds that multinational firms manipulate prices to shift income from countries with high tax rates to countries with low tax rates. This may seem obvious, but it is inconsistent with the approach we criticise, and which this blog defends. That approach simply assumes that all profit shifting is from developing countries to developed countries, irrespective of tax rates.

Clemens Fuest and Nadine Riedel

I admit I found this, and an expanded reply of line untenable.

This is my reply, sent today:

Dear Clemens

I have noted your reply to Raymond Baker et al dated 3 September 2009. I am leaving Raymond, Simon and others to comment on their own studies. Here I comment upon the work I have done for the Tax Justice Network and on your methodology.

In writing this mail I am suggesting two things. The first is that the methodology you use for your critique of TJN et al is flawed. Secondly, I want to open discussion on what you think transfer mispricing is.

Let me deal with your comments on the TJN paper ‘The Price of Offshore’ first. You say:

What we consider to be a back of the envelope estimate is that these estimates are employed to derive estimates of the tax gap by multiplying the estimated offshore asset holdings by a rough estimate of an asset return and an assumed average tax rate. We believe that the discussion of the limitations of this approach in our report is moderate and entirely appropriate.

That is an interesting suggestion but let me compare it with this comment of yours:

Estimates of how mispricing affects tax revenue in a country have to take into account all mispriced transactions. A very simple example might help. Assume that there are three exporters of a good in country A. Firm 1 exports the good at a price of 4, firm 2 exports the good at a price of 8 and firm three exports the good at a price of 12. The mispricing approach would identify 8 as the undistorted price. Assume further that all firms have costs of 4 in country A which are deductible from the profit tax base in country A. The goods are exported to country B, where all three are sold at a price of 14 to consumers in country B.

In this example, the aggregate tax base in country A is equal to 12. Firm 1 shifts income out of country A and firm 3 shifts income into country A. In the absence of trade mispricing, the tax base in A would be the same. The tax revenue loss of country A due to mispricing is equal to zero.

A method which only takes into account firm 1 and neglects the implications of mispricing by firm 3 is clearly misleading.

Let me compare these two scenarios. TJN used data you acknowledge has value. It then used verifiably appropriate assumptions on rates of return and tax rates which also accorded with reasonable market knowledge. And it allowed for the existence of compliant tax payers in the sample before suggesting a figure for tax lost. The resulting figure has been widely quoted precisely because whilst the resulting figure is, inevitably, an estimate it is easy for almost anyone to verify its credibility. The very ‘back of the envelope’ moniker you attach to it in pejorative form is in fact its essential strength: it feels right because everything about it is accountable, transparent, verified or plausible and therefore credible.

Now let me look at your scenario. You build an abstract model to dismiss our work. I do not, of course, dismiss the power of scenario building in economic theory: game theory proves its worth. The powerful model of the prisoner’s dilemma shares with the TJN work some features in common. It, for example, models a scenario to which the reader can relate (one hopes in abstract alone) and this gives it an inherent plausibility because the user can relate it to real events.

I am not really not sure that your model shares these characteristics. For example, you present no evidence that the diversity of pricing that you model exists. Nor do you demonstrate that you have calculated a proper transfer price: you have just assumed it is the average of the prices you say have been charged by related entities. Now I may be wrong, but I don’t think this is how the arm’s length model works: that, as I understand it seeks to determine arm’s length prices. In this case that might be anything from 4 to 14. I can’t see the logic for assuming it is 8. Have I missed something?

In addition I admit I am confused by your model. Again, this might just be me, but you seem to suggest that in the face of identical facts one exporting firm shifts income into a territory to secure a tax advantage and another shifts income out of it for the same purpose. I admit I do not subscribe to the view that all human behaviour is rational, but to put forward such a model seems decidedly odd. It seems to suggest you expect irrational behaviour, indeed even inexplicable behaviour in the face of consistent tax incentives. As far as I can see you have provided evidence as to why this perverse behaviour would arise, and I have to admit that without that explanation I am having real problems working out what the use of your model might be. Might you explain it to me? Might you also explain how this proves the work I and my colleagues have done is wrong? Is it that you wish us to build irrationality into our models in a way that we have so far failed to do? No doubt you will let me know.

Let me move now to areas I can speak on with more assuredness since they relate to my particular expertise, and not to economic modelling. Firstly, I note you propose that future work on transfer mispricing be based on micro-economic data based on the accounts of limited liability entities, and especially those in secrecy jurisdictions. I note you suggest that this has been done in the past. I have read some such work. Unfortunately I do not think this terribly relevant. First, as I presume you admit, the data of this sort that is available is very unreliable: only 6 of the secrecy jurisdictions currently being researched by the Tax Justice Network require that accounts be placed on public record. The UK is one of them. This is hardly the basis for a valid sample, I think. If you disagree, might you say why?

Second, the whole nature of accounting has changed since the widespread adoption of International Financial Reporting Standards in 2005. It is no longer true to say as a result of that adoption that reported profit in a set of accounts need bear any relationship at all to taxable profit, even if they ever did. This is because profit can now include fair value movements. This means that accounts are an increasingly poor source of data for this research.

Thirdly, the very issue which this work seeks to highlight, being transfer mispricing, usually results in tax deferral for tax reporting purposes, not absolute tax avoidance. However, academic economists who research this issue seem to persistently use the profit and loss account charge for taxation as a proxy measure for corporation tax paid even though it incorporates that deferred tax charge which might in itself disguise that transfer mispricing. As such unless the methodology you propose to use identifies the current tax charge, or better still, tax paid in cash (allowing for time delay factors) then it is very unlikely that meaningful data on tax paid will be surveyed by any such study.

Finally academic research in this area persists in using database data on accounts. That rarely has the information required to identify deferred tax charges and almost never includes data on tax haven / secrecy jurisdiction companies where, if our hypothesis is correct, much of transfer mispricing is hidden. For all these reasons the basis of analysis you propose seems inherently flawed. If you think I am wrong might you explain why?

This does not, though address another issue. At its core there remains a further methodological issue that needs to be resolved. You are seeking to approach this matter as if it is one of micro-economics. It is not. It is a macro-economic issue. The methodologies we at Tax Justice Network and those our colleagues have used recognises this fact. It is not at all clear you do. We challenge your assumption that any findings based on micro-economic analysis can be extrapolated to indicate the scale of the issue we are addressing. Why do you think that is possible?

Let me turn then to the final remaining flaw that I think is inherent in your logic, which I think of considerable importance. You state that all flows into and out of a state should be considered in any analysis of tax loss to developing countries as a result of transfer mispricing. I, in principle, agree. That would be desirable. But I wholeheartedly disagree with your contention that to measure flows in only one direction will lead to overstatement of the tax losses arising from any analysis. That is not possible. Let me explain why I think the only risk that arises is that a one direction analysis will understate the value of tax lost to developing countries.

Firstly, there is little or no evidence of inward flows to developing countries resulting in the reporting of significant taxable profits. If there were inward profit flows then there is very clear evidence that this would only arise because there was no chance of a corporate profits tax arising upon them as a result e.g. due to tax holidays or excess allowances. In that case there would be no tax gain in the developing country, albeit there might be tax loss elsewhere. As such your first assumption, that inward flows would create revenues that will negate outward flow losses would seem to be wrong: the incentives for each differ because the tax treatment for each differ and therefore it is completely incorrect to assume each might or would give rise to equal and opposite tax impacts.

Second, what evidence there is for the underpricing of inward flows suggests that these arise to avoid or evade duties, tariffs and other charges arising on import and not to secure corporate tax advantages, for reasons noted above. These import tax abuses are not taken into account in any estimate Baker, Christian Aid, Pak et al have made. In that sense those authors underestimate the lost taxes. It is important to note though that such evasion of import tariffs would not rationally arise if it gave rise to overall increase in tax liability i.e. the tariff saving must always be more than any additional tax due on additional corporate profits, if any. Therefore, again, the existence of inward flows does not suggest that losses might be lower than estimated; it would suggest that they are more than estimated.

Thirdly, you seem to assume that transfer mispricing inward takes place on the same trade as the transfer mispricing outward. You therefore imply there is a net offset available between the two. Firstly I suggest that the existence of inward and outward mispricing in developing counties on the same trade is very unlikely because there is no evidence of developing counties being used, unlike tax havens, for repricing of goods and services in vertical supply chains. In other words, there would appear to be little or no chance that any inward transfer mispricing can be netted off against outward transfer mispricing for tax purposes. Second, even if in economic terms you believe that this offset might be reasonable— assuming if, and only if, they arose on the same trade, the reality is that this is not how the two errors would be treated by tax authorities. Those authorities can treat the two transactions as entirely independent variables i.e. they are perfectly entitled to pursue an underpayment arising as a result of transfer mispricing outward even if the goods exported had mispriced inwards. In other words, the two errors are wholly independent of each other, net off is not possible, and correction can be made in both cases, to upward lift the sale price to collect additional taxes on profit and upward lift the inward price to correctly collect taxes on import but without allowing any resulting adjustment for the purposes of taxing profits. Your contention appears to ignore this reality. The result is that that your claim that netting off takes place is, in my opinion, wrong: the errors are additive, not subtractive and accordingly the point you make only serves to indicate the conservative nature of the calculated loss to developing counties inherent in my colleagues’ work.

Given these facts I am finding it very hard to find merit in the arguments you have presented, but look forward to hearing your further opinion on the observations I make.



To put it another way, it seems to me that here we have an economist writing about something about which he seems to know very little, and about which he formulates a series of assumptions to overcome a) the inherent weaknesses in his knowledge and b) the fact that reality does not accord with the world as he would like to see it, rather than  the world as it is.

That is, of course, a good description of most economists. But let’s see what he has to say.

Interestingly, we are both in Washington on Monday at the same World Bank conference where we are both presenting papers. Should be fun.