Reproduced below is a joint post by Prof Andrew Baker of Sheffield Political Economy Research Institute and me, originally on the blog of the All Party Parliamentary Group on Inclusive Growth.  The original post sets the context as follows:

Growing interest in tax spillover analysis by NGOs, international organisations and governments is warranted, but there is a need to reframe existing analysis. Our new framework and method would give us a fuller reading of international and domestic tax vulnerabilities.

The blog trails a meeting in the Houses of Parliament tomorrow night at which the paper Andrew and I have written on tax spillover effects will be launched. 

Tax continues to fascinate the media, politicians and international political economists. In that approximate order there is also an increasing demand for data to support action on what many think to be tax abuse. Much of that demand relates to specific abuses of tax systems, whether by tax avoidance or tax evasion. There is, however, a new interest in systemic abuse of one country's tax system made possible by the design of tax law and policy in another jurisdiction. These impacts are called ‘spillover effects' and have been the subject of interest from a number of major NGOs who campaign on tax issues, many of whom think that these spillovers are a feature of tax competition.

The IMF took up the challenge of estimating these spillover effects in a paper they published in 2014. The method proposed by the IMF is highly econometric and essentially depends upon analysis of comparative trade and investment data between states, with the difference between them being compared to the simultaneous differences in tax regimes of the locations in question. Only two countries have actually conducted such analysis to date. They are Ireland and the Netherlands. Somewhat surprisingly the homes of the ‘Double Irish' and ‘Dutch Sandwich' tax abuses were found to have no real identifiable spillover effects in the case of Ireland, while there were some negative and material effects on some developing countries in the Dutch case.

The Irish finding and the general method for conducting spillover analysis has troubled us. Our first concern is that wholly suitable trade and investment data was not readily available in both cases. Second, there was a potential fundamental flaw: if transactions were completely relocated out of jurisdictions, as the Double Irish and Dutch Sandwich both encouraged, then the chance of any tax impact being noted in direct relationships was minimal,  because the impacts would be secondary, transmitted through indirect relationships, and therefore likely to be missed. For these reasons we felt a methodology that was simultaneously more comprehensive, comprehensible and easier to use, that could also produce a wider range of results and indicators, giving us a broader picture of the dynamics of tax competition, was merited.

We have now proposed that alternative methodology in our new paper for the All Party Parliamentary Group on Inclusive Growth (we'll present the key findings at the APPG's event on 14 March in Parliament — register a place here). At its core our new system is a perceptions index which both tax experts and those with informed interest in the tax system of a country could complete. We suggest spillovers are considered, at present, for four taxes. These are income tax, social security, corporation tax on profits and capital gains taxation. The first two are subject to frequent tax attack in our experience; the last two are taxes originally introduced with the primary purpose of defending these taxes from attack. If we want a better sense of the vulnerability of particular tax regimes, and the vulnerabilities certain tax regimes generate for others, we need to look at how tax regimes hang together as a whole and the potential relationships between these different taxes.

Our system ranks the risk of spillovers between taxes on a scale of one to five. If a score of five is awarded then the tax is very vulnerable to spillovers. This means it is likely that taxpayers will try to divert part of the income that should be subject to this tax to another tax or location, or both. If a score of one is awarded then the tax in question in the jurisdiction being assessed is likely to attract the income or gains that should be taxed under the rules of another tax or in another place.

There are several important dimensions to this. First, the risk can be assessed domestically. So, for example, is the design of UK corporation tax likely to undermine the tax base of national insurance? This assessment measures the risk of tax spillovers through tax avoidance within the jurisdiction. Second, the risk can be assessed internationally. So, again as an example, is the fact that the UK has a relatively low corporation tax rate likely to attract the corporation tax base that should be recorded in other countries into the UK? Third, it is not just tax legislation that can be assessed: the spillover assessment can also be extended to tax administration. This process would ask, in effect, whether or not abuse is likely to be detected since this very obviously has an impact on spillovers. The analysis can, fourthly, be extended beyond tax. So, for example, considerations relating to corporate law and its implementation can also be added to the assessment matrix. Each score can be allocated in two ways. One is the domestic impact and the other international. These scores may be different. Taken together, the result is that scores for each tax can be generated domestically, while scores for the jurisdiction as a whole can be marked internationally. The domestic score is a measure of the overall deviance within the system from a neutral score of three between all taxes — which would imply no shifting was likely between tax bases. The spread is the important issue here, indicating the level of risk that exists. Internationally, the overall average may be more important: a low score indicates an aggressor jurisdiction. A high score indicates a vulnerable jurisdiction.

We stress, this idea is at an early stage of development at present. We think the system will work as a means of ranking and evaluating domestic and international tax vulnerabilities. We have designed it so that one jurisdiction might be capable of assessment many times by members of different stakeholder groups whose perceptions may then be weighted to deliver an overall balanced score across a range of opinions. The difference between stakeholders might itself, we suspect, be revealing. But, and we stress this, those trials have not yet happened. We are simply suggesting a design concept at present.

That said, we think this is a design concept of importance. The evidence that the IMF econometric approach to spillovers does not work is already compelling. This does not, however, discredit the idea of spillovers because they undoubtedly exist. What it instead demands is an alternative method. Our proposal is a start to that process. It certainly warrants further refinement, scrutiny and discussion.

Andrew Baker is a Professorial Fellow at the Sheffield Political Economy Research Institute (SPERI) and the Department of Politics at the University of Sheffield

Richard Murphy is Professor of Practice in International Political Economy at City University