Is it or isn’t it a spillover?

Last week’s Global Tax Policy Conference at Maastricht University on “international spillovers in taxation” has got me thinking. In particular, I was fascinated by Belema Obuoforibo’s presentation on the IBFD’s methodology for ‘spillover analyses’ (here is a link to an IBFD Powerpoint describing it). The term ‘spillover’ comes from the IMF, a term they use in economic analysis more generally to refer to “the impact of policy actions in one country on others” and around which the IMF framed a whole policy paper. Since its mention in the famous international organisations’ report to the G-20 [pdf], spillover analysis has become a common civil society demand, and the Netherlands and Ireland have gone on to commission them, with a focus on the interplay between their tax treaties with developing countries and certain provisions of their domestic tax law.

But a ‘spillover analysis’ is not the same as an ‘impact assessment’, and I think it’s important to understand the distinction.

The IMF’s recent paper on spillovers in international taxation distinguishes between ‘base’ spillovers, in which an action by one country affects another country without that country doing anything in response, and ‘strategic’ spillovers, in which the change creates an incentive for the second country to change its own policies – tax competition being the obvious example. All very interesting, but I sometimes find it tricky to see how the IMF’s definition applies to the specific areas discussed later on the same report.

One reason for this is that a ‘spillover’ effect from one country on another implies that the affected country is a passive victim. This is not the only way in which one country’s tax system might affect another’s. The decision by a developing country to sign a treaty, or to adopt an international norm, makes it an active participant, but that doesn’t diminish the impact on it of doing so.

A second issue is that ‘spillover analysis’ in practice has tended to focus on how one individual country’s tax system might be different to the norm. Following the Dutch spillover analysis, the government noted that Dutch treaties with developing countries were generally on the same terms as those countries had secured with comparable treaty partners. The IBFD’s methodology is based on how aspects of one country’s tax system compare with similar countries. So the growing practice of spillover analysis, it seems, considers impacts relative to the international average, not absolute impacts. It is a way of finding out if a country is worse or better than average, rather than seeking out all positive or negative impacts.

Let’s consider some examples from the area of tax treaties, to illustrate how these limitations play out in practice:

1. Is it a ‘spillover’ if the affected country actively opened itself up to the vulnerability?

Ireland didn’t used to be a tax haven. It used to be a ‘high tax’ jurisdiction like most OECD countries. Its transformation into a hub used for base erosion and profit shifting is relatively recent, and many of its tax treaties predate this. The changes to Ireland’s tax system transformed its tax treaty network into a major headache for countries such as Zambia, and it seems quite right to describe these effects as spillovers: treaty partners could not have anticipated them when agreeing terms in their tax treaties.

The same could probably be said of the Mauritius-India treaty, which originally predates Mauritius’ generous offshore regime. But it couldn’t be said of Mauritius’ treaties with many African countries, which they signed up to at a time when the risk of treaty shopping through Mauritius would have been clear. This is not a ‘spillover’ of Mauritius’ tax system, because it was a conscious choice taken by the developing countries. But it would be a shame if any analysis of the impact of Mauritius’ treaty network didn’t consider these costs.

2. Is precedent in tax treaty negotiations a strategic spillover?

When Zambia signed a treaty with China containing much lower withholding rates than it had previously agreed to, it may not have anticipated the British reaction, which was to request a renegotiated treaty on similar terms. Zambia also wanted to renegotiate with the UK, mainly to seek improved powers for cooperation and information exchange with the British tax authorities. The implication of the Chinese treaty, however, was that the UK expected similar terms in return for Zambia’s demands, significantly lowering the Zambian tax take from British investors. It seems to me that this was a ‘spillover’ effect of China’s aggressive negotiating position…but I suspect it would be unlikely to be picked up in any ‘spillover analysis’.

3. Are most-favoured-nation effects spillovers?

In April 2003, Venezuela and Spain signed a treaty with a most favoured-nation (MFN) clause in its interest article, which would be triggered if either country subsequently signed a treaty with a lower maximum rate in the interest article. In May 2006, the bilateral MFN clause in its interest article was triggered through a kind of domino effect: Estonia and the Netherlands signed a treaty granting exclusive residence taxation rights over interest; this activated the MFN clause in the September 2003 Spain-Estonia treaty, which in turn activated the MFN clause in the Venezuela-Spain treaty. As a result, according to an article in Tax Notes International, “Venezuela’s treaty with Spain has undoubtedly become the most favorable tax treaty executed by Venezuela to date.”

This was a policy action by two countries that had ramifications for two other countries, but which they should have anticipated when they agreed to the clause. The inclusion of a symmetrical clause, which could be activated by Spain as well as Venezuela, may have been a negotiating error by the latter. But was the MFN activation a spillover?

4. Is it a spillover if a country is just an aggressive negotiator?

The countries that have tended to do spillover analysis so far are those facing accusations that they’re used for treaty shopping. And the main policy response, at least from the Netherlands, seems to have focused on adding anti-abuse provisions into its treaties. What if there is no treaty shopping, but instead just a set of treaties that take most of the taxing rights away from developing countries? As I noted in point 1, the developing countries actively signed up to these treaties at some point in their histories. The IMF spillover paper sounds a cautious note about tax treaties because of the fiscal costs to developing countries of limiting their source taxing rights, but it’s not clear (to me at least) how the plain vanilla impact of the source/residence split could meet the definition of a spillover.

5. Is it a spillover if it’s no different to the norm?

Consider many developed countries’ unwillingness to share the right to tax their shipping firms with the developing countries whose waters they use, and who are in some cases very keen to tax the profits from shipping. Developed countries have been relatively united on this, so even if we accepted that the case described in point 4 above was a ‘spillover’ effect, I think the implication of the IBFD’s methodology would be to find no negative spillover from the hardline stance taken by, say, the UK. But just because this ‘policy action’ is consistent with comparable treaties and with the model conventions, does that mean countries should not assess the impact on developing countries of the policy stance?

In conclusion, it seems to me there is a risk that the focus on the concept of ‘spillover effects’ might lead to an overly narrow analysis of the impacts of a jurisdiction’s tax policy actions on developing countries. Or, to put it another way, could the choice of terms used to discuss this issue have linguistic spillover effects?

Update: Joe Stead points out that I have slightly undersold the 2011 International Organisations’ report to the G-20: