under the model treaties
Tax information is valuable in a world where capital markets are integrated but governments seeking tax revenues are bound by sovereign borders. Countries have always tried to bargain with information by agreeing to share it in the context of tax treaties, but while the mechanisms for finance and capital flow have rapidly evolved over the last century, the tax information bargain has remained remarkably stagnant.
This article looks at the information sharing provisions of the OECD and UN model tax treaties and shows that the slow pace of change in tax cooperation is transforming information into a valuable commodity. As the model treaties fall further behind, information becomes ever more valuable to those who can protect it and ever more expensive to those who would seek to gain access to it for the purpose of extracting tax revenues.
A better grand bargain is possible, but now so extremely costly that it is likely all but impossible to achieve.
Information sharing has always been an integral feature of international tax coordination, the “prevention of evasion” half of the raison d’être of tax conventions.
Sharing information is part of the tax treaty bargain because tax information is very valuable. It is valuable to taxpayers, whose ability to control what governments know about their available resources impacts how much of such resources they will be asked to contribute to society. It is valuable to tax advisors, who can trade on their expertise by selling tax planning services to taxpayers.
It is also valuable to nongovernmental organizations and other members of civil society, who can use information to make and express policy positions1.
Finally, it is valuable to governments, which can choose to collect or not to collect it, and share or not share it, as a matter of controlling their own access to taxpayer.
For all these reasons, tax information is very valuable depending on who has it, who can get it, and who can use it. Countries should therefore expect to pay for access when what is sought belongs to another2. But governments do not typically exchange information for cash. Instead, they tend to use information as part of the tax treaty bargain, offering each other reciprocal glimpses into their stored vaults of tax information in connection with their agreements on the sharing of revenues between them.
This basic bargain is entrenched in the international tax community by means of the model treaties promulgated by the Organisation for Economic Cooperation and Development and the United Nations. From the very first model tax treaty produced by the OECD, the concept of sharing information on a regular and reciprocal basis has served as a complement to the revenue sharing to be achieved by international agreement. The text of paragraph 1 of Article 26, the main information sharing provision in both the OECD and UN models, shows the remarkable staying power of the language of information sharing. (See Table 1).
Two features of information sharing are evident from this progression. First, there is cross-institutional influence, but it only travels in one direction: from the OECD to the UN. The UN has sought to add modest clarifications about the connection between information and tax evasion, and the OECD has chosen to completely ignore these modifications. Second, both models are evolving at a glacial pace. This might imply that the sharing mechanism works reasonably well and therefore requires little by way of substantive change.
The evidence of information sharing in practice suggests that this is not an accurate picture. Perhaps the most that can be said is that since treaty terms are subject to interpretation on a case by case basis, the practice of information sharing could be evolving faster than the models themselves.3
The information sharing structure might seem rather unremarkable, were it not for the counterweight that has stubbornly persisted through all the iterations of both Models, namely, the opt-out clause found in paragraph 2. In the 1963 OECD Model, this paragraph read as follows:
In no case shall the provisions of paragraph 1 be construed so as to impose on one of the Contracting States the obligation:
a) to carry out administrative measures at variance with the laws or the administrative practice of that or of the other Contracting State;
b) to supply particulars which are not obtainable under the laws or in the normal course of the administration of that or of the other Contracting State;
c) to supply information which would disclose any trade, business, industrial, commercial or professional secret or trade process, or information, the disclosure of which would be contrary to public policy (ordre public).
This language has appeared, virtually unchanged, in every version of every UN and OECD model treaty through the present. Its basic function is to allow countries to opt out of information sharing on grounds that domestic privacy or confidentiality laws prevent the government from collecting or sharing the information that might be sought by the treaty partner. However, following the OECD harmful tax practices initiative, the OECD attempted to cut off some of the power of the opt-out by adding two more paragraphs to the information sharing article:
4. If information is requested by a Contracting State in accordance with this Article, the other Contracting State shall use its information gathering measures to obtain the requested information, even though that other State may not need such information for its own tax purposes. The obligation contained in the preceding sentence is subject to the limitations of paragraph 3 but in no case shall such limitations be construed to permit a Contracting State to decline to supply information solely because it has no domestic interest in such information.
5. In no case shall the provisions of paragraph 3 be construed to permit a Contracting State to decline to supply information solely because the information is held by a bank, other financial institution, nominee or person acting in an agency or a fiduciary capacity or because it relates to ownership interests in a person.
This language reappears without change in the OECD 2010 model, and the UN 2011 model adopts the new paragraphs verbatim. The addition seems on its face to foreclose the possibility that countries will strategically employ domestic confidentiality laws to avoid the sharing they agreed to in paragraph 1.
If that is indeed the outcome of adding the two new paragraphs, the UN model’s wholesale adoption of the OECD language might seem uncontroversial, at least in terms of scope. But that is in all probability not the case, since the provision only prohibits opting out “solely” because of who holds the information or who it relates to. But even if the opt out problem is solved by the new provisions, the persistent problem of reciprocity has not even been addressed, much less solved in the UN 2011 model.
It has always been the case that the formal provision of reciprocity in information sharing does not always mean that information sharing is mutually beneficial in practice. In some tax treaty relationships, the information sharing mechanism may only benefit the capital-exporting developed country (usually a developed country), while the compensation for the source country (usually a developing country) for providing the information may lie solely in any market signalling effect the tax treaty might produce4.
But the bargain seems to at least acknowledge that information is valuable, even if it does not necessarily provide adequate compensation therefor. The more recent trend involves attempts to extract information from some countries without any quid pro quo, namely, in the form of tax information exchange agreements. These efforts do not seem to be producing significant returns, perhaps because these agreements seem to fail to account for the value of information at all5.
With the value of information going up, it seems increasingly clear that revenue allocations are not the only imbalance in a treaty that involves uneven flows of capital between the parties. The burdens and benefits of information sharing also clearly fall unevenly on the treaty partners.
The UN model was originally promulgated as an answer to the OECD’s model, which generally assumes and is suited to a reciprocal economic relationship between two developed countries6.
The UN Model intended to mitigate uneven tax revenue sharing in the context of uneven development among the treaty partners, mostly by allocating more primary taxing rights to source countries. But on information sharing, as shown above, the UN model is virtually identical to its OECD counterpart.
In a comparison of two prior UN and OECD model tax conventions, Michael Lennard, chief of the International Tax Cooperation and Trade Section at the UN Financing for Development Office, called the practical application of the exchange of information standard “a ‘black box’, operating in the shadows more than the light.”7
That does not seem to have changed with the new UN model. Nor does Lennard’s other concern, that “developed countries will expect [less developed countries] to provide information, and will use their economic power to ensure this happens, whereas some developed countries will not be so forthcoming in providing information to allow other countries to protect their tax bases and reverse ‘encouragement’ will not be possible.”
Lennard identifies this particular example of imbalance as one very good reason to have a tax policy group that is working independently on international tax standards in order to protect the interests of poorer countries. Since the UN is the only institution capable of doing that currently, its wholesale adoption of the OECD standard is therefore discouraging.
An information sharing standard that fails to recognize the uneven benefits and burdens of bilateral information flow will also fail to understand the value – and therefore the cost – of information as currency in international tax agreements. Once tax information is acknowledged as a valuable commodity, we must be concerned that access to it is likely unevenly distributed under a reciprocity-based treaty exchange system.
It seems particularly important then that a model for treaties between developed and developing countries take steps to account for the costs and the values being distributed via tax treaties.
The UN model could do so by approaching tax information sharing in the same manner as it has approached tax revenue allocation, namely, by making space for non-reciprocal arrangements where reciprocity creates imbalanced outcomes. This is admittedly a difficult task since the value of information to any given country may or may not match its ability to extract the desired information, or use it if so extracted.
The slow pace of change in both Models suggests that change in the standards written into model treaty language is hard fought. The evolution of information sharing in practice may be even more contested. In the meantime, the value and the cost of information sharing are quickly climbing.
If countries cannot bring themselves to devise standards for acknowledging the value of information and therefore allocating the burdens and benefits of information sharing, we should not expect information to be freely shared. Instead, we should expect that the problems currently embedded in the models will persist and worsen.
That has been the path to date, but it need not be the path going forward.