The Colombia UK tax treaty: 80 years in the making

Today at 2.30pm, the UK parliament’s Third Delegated Legislation Committee will debate tax treaties with Lesotho and Colombia. It will be interesting to see how much debate really takes place, a matter on which I’ve commented before once or twice.

The hearing gives me a chance to plug my article in the British Tax Review last year [pdf], which traced the UK’s attempts to obtain a tax treaty with Colombia over 80 years. Its overtures were frequently rejected, at first because Colombia was not interested in tax treaties, then because it was bound by the terms of the Andean pact, and finally because it could not agree on terms with the UK, especially over technical service fees, an area where the UK position has changed. Since the article was published I had the chance to speak with a Colombian tax official, who told me that Colombia’s change of heart on technical service fees is a change of view about tax policy, rather than a concession forced by OECD membership, as I speculated in the article. Of course, the two developments might not be totally independent.

Here is how the article concludes:

The demands of OECD membership, combined with the unusually liberal use of MFN clauses during an era of less-than-strategic negotiation, seem to have backed a country once insistent on a “red line” over technical service fees, and before that sceptical of accepting the limitations on its taxing rights that come with a tax treaty, into a corner. Having been constrained in its negotiating position by the pro-source taxation stance of the Andean community, Colombia now finds itself pulled in the other direction by the OECD. Is this further proof that the world is moving inexorably towards an OECD-type tax system? The gradual but steady expansion of the OECD, given a fillip most recently by the announcement that Brazil would begin accession talks, might lead us to such a conclusion. In contrast, however, the continued expansion in the use of the technical service fees Article by developing countries, together with its imminent introduction into the UN Model Treaty, point towards a growing divide between states on this topic.

The long history of negotiations between the UK and Colombia perhaps demonstrates more than anything the extent to which the tax treatment of international transactions today is a product of historically specific events. Each side’s positions changed radically over time, from a refusal to accept each other’s terms to a willingness to concede them outright. The UK’s constant enthusiasm for a treaty with Colombia stands in contrast with the latter’s oscillation between hot and cold. If Colombia turns cold again, however, it will be left with a fossilised relic of its negotiating position in 2016. Given the rarity with which tax treaties are terminated or their terms substantially renegotiated, treaty networks are collections of these fossils. Hence Colombia is stuck with its MFN [most favoured nation] clauses, regrettable outcomes of its negotiating spree in the 2000s. The biggest irony, however, is reserved for the UK. Despite its apparent willingness in the 2000s to forgo a treaty with Colombia over withholding taxes on technical service fees, Britain retains, as a legacy of its negotiations from 1973 until the turn of the century, the largest number of treaties of any OECD Member containing just such a clause

Advertisements

Certainty in the tax treaty regime

Here’s the text and slides of a talk I gave yesterday at an event called Harnessing the Commonwealth Advantage in International Trade.

I want to talk today about issues related to tax treaties in developing countries, and their impact on tax certainty for multinational investors. To do this I think we have to consider two aspects of the tax treaty regime: the multilateral norm-setting processes at the OECD and United Nations, and the individual bilateral treaties negotiated by pairs of countries. The key point I want to make is that, at both these levels, the elaboration of a regime that constrains developing countries’ source taxation rights in ways that risk being seen as excessive is not sustainable in the long term.

Consider first the multilateral level. Last week I was reading a PWC document, ‘Navigating the Maze: Impact of BEPS and Other International Tax Risks on the Jersey Funds Industry [pdf].’ It notes that:

Countries are already diverging from suggested guidance from the OECD, which was meant to bring coherence and consistency.

This does not only apply to developing countries, but there is plenty of evidence to suggest that in emerging markets there is a growing dissatisfaction with the OECD approach, as illustrated by the ongoing row over the status of the UN tax committee, and India’s recent financial contribution to its trust fund, which until then had been empty for over a decade.

Here are two quotes that illustrate this sentiment further:

“For developing countries the balance between source and residence taxation [is] very crucial. International tax rules with its preferences for residence based taxation [are] not in interest of developing countries.”

Eric Mensah, Ghana Revenue Authority, 2017 [pdf]

“The global tax system is stacked in favour of paying taxes in the headquarters countries of transnational companies, rather than in the countries where raw materials are produced.”

Francophone LIC Finance Ministers Network, 2014 [pdf]

It seems that, to maintain the integrity of the international tax system as emerging market voices become stronger, countries that favour residence-based taxation will need to accept greater flexibility within the instruments agreed at multilateral level.

Turning to the bilateral treaties that developing countries have negotiated, here I want to introduce you to some research I conducted at the LSE, funded by an NGO called ActionAid. ActionAid used it to inform a campaign that has targeted individual governments and treaties, calling for renegotiations.

Slide2

I took 500 tax treaties concluded by developing countries and had a group of LLM students code them for the main clauses that could vary on a source-residence axis, using an International Bureau of Fiscal Documentation analysis. We can use that data to plot each treaty along a simple axis from 0 to 1, where 0 means an overwhelmingly residence-based treaty, and 1 a more source-based treaty. Remember that 1 here represents the presence of the most source-based clauses within existing treaties, and doesn’t take into account the concerns about inherent bias in the parameters for those treaties set by the OECD and UN models. In this first slide you can see that treaties among developing countries, in light blue, are becoming marginally more source-based over time, while treaties between developing countries and OECD members are becoming more residence-based.

Slide3

The next chart shows some of the underlying drivers of those trends. You can see that permanent establishment definitions are becoming more expansive, perhaps reflecting changes to the model treaties, while withholding tax rates are trending downwards. There are diverse trends in different clauses within areas such as capital gains tax and taxation of services.

I want to talk to you about a few examples.

Slide4

Here we see Vietnam’s treaties taken from the same dataset. Vietnam has actually expressed a comprehensive set of observations on the OECD model convention, broadly following the UN model. So here a zero on the vertical axis means the treaty contains none of those positions and instead follows the OECD model, while 1 means it includes all of Vietnam’s observations. You can see that in the 1990s Vietnam signed a number of more residence-based treaties that are completely the opposite of its stated negotiating position. And of course, these are with many of its biggest sources of investment.

More recently, Vietnam has come to regret those earlier treaties, and has chosen to interpret certain provisions on PE and technical services in the way it wished it had signed them, rather than the way it did. Businesses are very unhappy, and in the words of the Vietnam Business Forum, it has:

made the application of DTA[s] of foreign enterprises impossible, effectively it obliterate[s] the legitimate benefit of enterprises.

The residence-based treaties that Vietnam signed when it was inexperienced and urgently in need of investment are creating uncertainty, rather than the stability that investors are looking for.

Slide5

You might be aware that a few years ago Mongolia tried to renegotiate a few of its treaties, and when it was unsuccessful it terminated them. They’re the treaties with the Netherlands, Luxembourg, Kuwait and the UEA, marked in black on here. But if you look on the bottom left, you see a number of treaties with OECD countries, including the UK and Germany, that have even more limited source taxing rights. Indeed, according to an IMF technical assistance report from 2012 [pdf]:

The Mongolian authorities are currently considering cancelling all DTAs and start building up a new DTA network with countries based on trade volumes and reciprocity in economic relations.

I’m told the IMF talked them out of this, but it is worth knowing that they considered it.

Slide6

Here is Zambia, a Commonwealth example. You can see the same pattern. Its earlier treaties were very residence-based. I did some archival and interview work on those early treaties, and you can see that when they were first signed, Zambia had a hugely under-resourced civil service, with no experience of negotiation, and other countries took advantage of this. The most egregious example is its treaty with Ireland, which had zero withholding tax rates on all types of payment. That’s in contrast to the East African community countries, which had very strong negotiating red lines, and as a result either walked away, or obtained more source-based treaties that today appear quite generous, but have stood the test of time.

Slide7

This chart shows a few renegotiations that have taken place in response to government and civil society concerns. You can see that Zambia’s renegotiations have focused more on updating treaties and closing loopholes, not dramatically shifting the balance of taxing rights. In contrast, Pakistan and Rwanda have both negotiated big overhauls.

So in conclusion, as the politicisation of the international tax regime continues, especially in developing countries, I think we’re likely to see growing demands for a rebalancing between source and residence not just in the multilateral setting, but also in individual treaties. My advice to OECD governments, and businesses who engage with them, is that tax certainty in the future depends on an enlightened approach to the tax treaty regime that leaves more developing country taxing rights intact.

European Economic and Social Council hearing on tax treaties and development

I’ve posted below the slides I just used for a presentation here at the European Economic and Social Council. The EESC has formed a study group to consider the question of “EU development partnerships and the challenge posed by international tax agreements.”

Interesting discussions included the evidence base for the effect of tax treaties on investment into developing countries. Here I think the key question is what provisions of tax treaties are relevant to investment flows, and in what circumstances, rather than simply whether tax treaties per se attract investment.

Tax certainty is the new buzzword, and it was interesting to think about how it applies here. On one hand, a treaty provides greater certainty because it commits its signatories to tax investors in a certain way as long as the treaty is in force. But that certainty relies on ongoing support for tax treaty norms. Developing countries feel unhappy with the content of the international tax norms on which bilateral treaties are based, as well as the institutions that develop those norms. (Here, for example, is a recent presentation by Eric Mensah from the Ghana Revenue Authority that makes these points). Countries such as Mongolia, Vietnam and Uganda are beginning to question the constraints imposed on their tax policy by treaties signed in the past. There is perhaps a trade-off between developed countries’ desire to defend the content of existing norms and the role of the OECD, and developing countries’ willingness to abide by those standards in the long term.

Link to presentation on Slideshare

Visualising Uganda’s (and others’) tax treaties

Interesting news from Uganda, where the government announced in its latest budget that it has finished formulating its new tax treaty policy, and will be renegotiating treaties that don’t comply. Seatini and ActionAid Uganda will no doubt chalk this up as a success! The news report linked to above also states that the the government plans to amend the awkwardly-worded anti-treaty-shopping clause in its Income Tax Act, although there are clearly still doubts about its application. According to a report in Tax Notes International, there’s an ongoing mutual agreement procedure between the Netherlands and Uganda to try to settle the ongoing Zain capital gains case, which turns on the applicability of that clause. 105_screen_shot_2016_04_29_at_6_11_10_am

So this is good timing for my working paper with Jalia Kangave, based on a submission we made to the Ugandan government’s review, to have been published by the International Centre for Tax and Development.

Here’s a link to that paper on Researchgate

When writing that paper, I thought that Uganda had a pretty good record of tax treaty negotiations, but some new visualisations of the ActionAid Tax Treaties Dataset suggest otherwise. For these I am indebted to Zack Korman, and to tax twitter for introducing me to him. Below are some maps Zach has made using the ‘source index’ I developed for the dataset (read more about that here). Red means a residence-based treaty that gives fewer taxing rights to the developing country, while green means a source-based treaty that gives it more taxing rights.

This slideshow requires JavaScript.

Links to high-res versions of individual images: Uganda map, Uganda bar chart, Vietnam, Mauritius, UK, Nordics

Uganda’s treaties are pretty red, meaning that most of its treaties restrict its taxing rights much more than average. Looking at the breakdown of the index shows that Uganda has some above-average withholding tax provisions, but its treaties are quite a lot worse than average in other areas. The slide show also gives some other countries for comparison. Vietnam’s treaties are mostly green, while Asian countries have got better deals from Mauritius (an offshore financial centre, not a developing country, in this context) than African ones. The UK’s treaties are pretty red, while the Nordics are very interesting: diverse in content, but consistent among themselves, giving good deals to Kenya and Sri Lanka, and worse ones to Tanzania and Bangladesh. This suggests that more source-based treaties with Nordic countries have been up for grabs for tough-negotiating developing countries.

Below I’ve posted some of Zach’s animated maps, on which it’s easier (and interesting) to follow the developments at earlier stages. There’s lots to comment on, but mostly I just keep watching them. The technical service fees map, at the bottom, is especially interesting, as it shows how countries have changed attitudes over time: watch how Pakistan suddenly changes position in the mid 1980s, for example.

World2

Above: All treaties in the dataset (red=residence-based, green=source-based)


Asia

Above: Asia (red=residence-based, green=source-based)


Africa2

Above: Africa (red=residence-based, green=source-based)


Vietnam2

Above: Vietnam (red=residence-based, green=source-based)


UK2

Above: UK (red=residence-based, green=source-based)


Nordic2

Above: Nordic countries (red=residence-based, green=source-based)


Netherlands2

Above: Netherlands (red=residence-based, green=source-based)


Slow WHT

Above: Management, technical service and consultancy fees WHT (green=included, red=excluded)

 

New data and working paper: measuring tax treaty negotiation outcomes

Today the International Centre for Tax and Development (ICTD) and ActionAid are launching a new dataset that I’ve developed over the last year. It’s the product of over a year’s work, mostly by an intrepid team of research assistants in the LSE law department, to code the content of over 500 tax treaties signed by developing countries. ActionAid, who funded the research assistants’ time, have also used the dataset to construct a league table for their campaigning work, and there’s a piece in the Guardian based on that. (I’m sure the headline, which criticises the UK, will provoke some debate.)

The aim of the dataset project is to make possible a whole lot of new research to get to the bottom of questions that have not been satisfactorily addressed in the literature to date. The jury is still out on whether tax treaties affect investment flows into developing countries, but do certain clauses matter more than others to investors? What determines the outcome of tax treaty negotiations: is it cooperation, power, or competence? What is a good outcome for a developing country from a tax treaty negotiation? Work on trade and investment agreements has already moved from a simple ‘is there a treaty?’ binary to studies based on the content of those agreements. With this dataset, we can do the same for tax treaties.

There’s also a working paper that I’ve written for ICTD, which sets out some of the initial conclusions I drew from simple descriptive work using the dataset. For example, here’s the chart showing how common treaty provisions that allow developing countries to tax foreign service providers have been over time. The difference between the two is that with a ‘service PE’, developing countries can only tax the company’s net profits, and only if it is physically present for a certain length of time; with a WHT clause, they can tax gross payments to foreign service providers, as most do in their domestic law, regardless of whether there is a physical presence.  The growth in both the provisions highlighted is quite notable.dataset1

The story for capital gains is more mixed. Article 13(4) is an anti-avoidance tool that allows a developing country to tax capital gains from the sale of a company whose value is mainly physical assets in the developing country, even if the company is located abroad. It’s becoming more common, which is not surprising since variations on it are included in both the OECD and UN model tax treaties. Article 13(5) gives a developing country the right to tax gains from the sale of any company resident in that country, even if the sale takes place abroad. Interestingly, this provision is becoming more scarce, though countries such as Vietnam have only recently begun to tax in this way in their domestic law. dataset2

I also constructed an index that evaluates the content of 24 different clauses within each tax treaty to assess how much of the developing country’s taxing rights it leaves intact (a higher value means a bigger share of taxing rights for the developing country). This chart shows that treaties between developing countries and OECD members are gradually becoming more favourable to the latter, while in contrast developing countries are starting to get better deals with countries outside the OECD.

Overall negotiated content by date of signature and type of treaty partner

dataset3

The downward trend is driven by falling withholding tax (WHT) rates, while the upward trend is primarily in permanent establishment (PE) provisions, as we can see by disaggregating the index into these different components.

Average values of category indexes for treaties signed in a given year

dataset4

This is new information that I don’t think people have been aware of before, so I’m quite excited about it. Here’s the summary of the working paper:

Measuring Tax Treaty Negotiation Outcomes:the ActionAid Tax Treaties Dataset

This paper introduces a new dataset that codes the content of 519 tax treaties signed by low- and lower-middle- income countries in Africa and Asia. Often called Double Taxation Agreements, bilateral tax treaties divide up the right to tax cross-border economic activity between their two signatories. When one of the signatories is a developing country that is predominantly a recipient of foreign investment, the effect of the tax treaty is to impose constraints on its ability to tax inward investors, ostensibly to encourage more investment.

The merits of tax treaties for developing countries have been challenged in critical legal literature for decades, and studies of whether or not they attract new investment into developing countries give contradictory and inconclusive results. These studies have rarely disaggregated the elements of tax treaties to determine which may be most pertinent to any investment-promoting effect. Meanwhile, as developing countries continue to negotiate, renegotiate, review and terminate tax treaties, comparative data on negotiating histories and outcomes is not easily obtained.

The new dataset fills both these gaps. Using it, this paper demonstrates how tax treaties are changing over time. The restrictions they impose on the rate of withholding tax developing countries can levy on cross-border payments have intensified since 1970. In contrast, the permanent establishment threshold, which specifies when a foreign company’s profits become taxable in a developing country, has been falling, giving developing countries more opportunity to tax foreign investors. The picture with respect to capital gains tax and other provisions is mixed. As a group, OECD countries appear to be moving towards treaties with developing countries that impose more restrictions on the latter’s taxing rights, while non-OECD countries appear to be allowing developing countries to retain more taxing rights than in the past. These overall trends, however, mask some surprising differences between the positions of individual industrialised and emerging economies. These findings pose more questions than they answer, and it is hoped that this paper and the dataset it accompanies will stimulate new research on tax treaties.

Revenue foregone through tax treaties in context

In the recent Tax Justice Network Africa report on tax treaties, I had a go at estimating some costs to governments, based on a back-of-the-envelope figure for cross-border dividend and interests payments.  This is similar to the methodology used by SOMO and the IMF. (It’s a bit rough and ready, because some of the return on FDI figures I used will include reinvested profits, not cross-border remittances, but I’ve seen some other more sophisticated working recently that produces roughly similar figures.)

What I didn’t realise when I wrote that report is that both Uganda and Zambia break out withholding tax revenue (which includes taxes on domestic as well as cross-border payments) in their budgets, so we can set these very rough estimates in context. I’ve been curious for a while to know the order of magnitude of the importance of tax treaties.

The upshot is that revenue foregone from the lower tax rates on qualifying dividends and interest in tax treaties (which is just one part of the revenue foregone through tax treaties) is about 15 percent of withholding tax revenue. As WHT revenue is about 40 percent of corporate tax revenue and five percent of total tax revenue, this means the revenue foregone is something like five percent of corporate tax revenue, and a little less than one percent of total tax revenue.  Here’s how I get there.

Estimating revenue foregone

Here’s the table from the TJN-A report. I take figures for the primary return on foreign direct investment and assume that investment from each country gets the same return. Then I apply the tax rate discount in the treaty to that those estimated flows. The revenue foregone using 2012 data is about US$17m in Uganda and US$42m in Zambia.

wht table

Uganda

Here’s the table from Uganda’s budget. The 2011/12 withholding tax (WHT) outturn works out at about US$130m using the exchange rate on 1st January 2012. The revenue foregone of US$17m is about 13 percent of total WHT revenue, or 0.7 percent of Uganda’s total tax revenue.

uganda

Zambia

For Zambia I only have the 2014 budget figures, which we can assume with inflation will be larger than those for 2012, the year for which the revenue foregone is estimated, and hence this will be an underestimate of the proportions. WHT foregone of US$42m is 15 percent of the total WHT revenue of US$280m, or 0.8 percent of total tax revenue (using the exchange rate on 1st January 2014).

zambia

Since these calculations don’t include the revenue foregone through reduced rates of other withholding taxes on portfolio dividends, royalties and technical service fees, never mind all the other ways in which tax treaties curb taxation of foreign investors, it seems reasonable to conclude that the total of all revenue foregone from Uganda and Zambia’s tax treaties is of the order of several percent of their total government revenue. There may be benefits to offset these costs, but the starting point for a cost/benefit analysis of tax treaties is certainly to estimate the costs!

Tax treaties in sub-Saharan Africa: a critical review

The report I authored for Tax Justice Network-Africa is now available. It’s based on field research done a year ago and has been a little while getting into print.

Here’s a link to read it online at academia.edu

Here’s a link to download the PDF

Tax treaties in sub-Saharan Africa report cover

And here’s the introduction:

There is growing attention on the question of tax treaties signed by developing countries. The costs of tax treaties to developing countries have been highlighted in recent years by NGOs such as ActionAid and SOMO. During 2014, an influential IMF paper warned that developing countries “would be well-advised to sign treaties only with considerable caution,” and the OECD, as part of its Base Erosion and Profit Shifting (BEPS) project, proposes to add text to the commentary of its model treaty to help countries decide “whether a treaty should be concluded with a State but also…whether a State should seek to modify or replace an existing treaty or even, as a last resort, terminate a treaty.”

Meanwhile, some developing countries seem recently to have become concerned by the negative impacts of some of their treaties. Rwanda and South Africa have successfully renegotiated their agreements with Mauritius. Argentina and Mongolia have cancelled or renegotiated several agreements. Responding to this pressure, two of the developed countries whose treaty networks have raised concerns, the Netherlands and Ireland, have begun a process of review.

To investigate this apparent shift in opinion among policymakers, and to see what lessons can be drawn by other developing countries, Tax Justice Network Africa commissioned this study of current policy towards tax treaties in Uganda and Zambia, two countries that appear to be questioning past decisions. Fieldwork, which consisted of interviews with government officials and private sector tax advisers, took place in Kampala and Lusaka in September 2014.

Uganda has announced a review of its policy towards tax treaties, while Zambia is renegotiating several of its treaties. The Ugandan review has several motivations, according to finance ministry officials. The lack of a politically enforced policy to underpin negotiations is one concern. “When I go to negotiate, all I have is my own judgement,” according to a negotiator. “We thought that cabinet should express itself.” Officials are also concerned about the taxation of technical services provided by professionals in the oil industry, and are asking questions about the relatively poor deal Uganda got in its as yet unratified agreement with China.

Zambia, it seems, is keen to update very old treaties that were negotiated on poor terms by over-zealous officials in the 1970s. But a recent treaty signed with China on poor terms has created a difficult precedent, dragging down the terms of its recent negotiation with the UK. Zambia is also encumbered with several colonial-era treaties that need urgent attention.

This report is divided into four following sections. Section 2 describes the historical development of sub-Saharan Africa’s tax treaty network, including some of the reasons given for its development. Uganda and Zambia are used as examples. Section 3 looks at some of the core vulnerabilities in the content of tax treaties signed by African countries, set in the context of weaknesses in their domestic laws. Section 4 provides a critical perspective on recent initiatives taken by individual countries, regional organisations and other international organisations.

Section 5 provides recommendations for African countries. In summary, they should:

  • Review all their existing tax treaties and domestic legislation, to identify areas where they are most vulnerable to revenue loss. This should include permanent establishment definitions, protection from treaty shopping, and withholding and capital gains taxes.
  • Formulate ambitious national models by applying a “best available” approach to existing models (EAC, COMESA, UN), current treaties, and domestic legislation, none of which are currently adequate.
  • Identify red lines for negotiations from within these models.
  • Based on investment and remittance data, request renegotiations of treaties that have the greatest actual (or potential in terms of capital gains) cost. These renegotiations should be conducted on the basis of an improved distribution of taxing rights, not a “balanced” negotiation.
  • Cancel these high-impact treaties if the red lines cannot be obtained.
  • Incorporate an assessment of tax foregone due to tax treaties into an annual breakdown of tax expenditures.
  • Ensure that all tax treaties are subject to parliamentary approval as part of the ratification process.
  • Ensure that future updates to provisions of the UN and OECD model treaties, or to their commentaries and reservations/observations, reflect the positions set out in their national models.
  • Strengthen the African model treaties (EAC, COMESA, SADC) so that they act as opposite poles to the OECD model, rather than compromises between the UN and OECD models.