ICTD Working Paper 50
In June 2014, Uganda announced the temporary cessation of bilateral tax treaty negotiations, and a review of its policy towards such treaties. The main effect of tax treaties is to divide up the ‘rights’ to tax cross-border investment between the state parties, which reduces the possibility that businesses will incur double taxation; in doing so, it places significant curbs on the ability of capital-importing countries, such as Uganda, to tax foreign investors.
Uganda’s review follows decisions by developing countries as diverse as Argentina, Mongolia, Rwanda and Zambia to cancel or renegotiate some of their historical tax treaties. These countries, together with some independent commentators, international and non- governmental organisations, have questioned whether the benefits of tax treaties for developing countries outweigh their costs. In Uganda, as elsewhere, tax treaties have always been surrounded by an investment promotion discourse in political debate, yet there is little convincing evidence that they have had a positive effect on investment flows into low-income countries. In contrast, there are some clear aspects of Uganda’s treaties, such as definitions of ‘permanent establishment’ and rules concerning the taxation of capital gains, which cost Uganda significant revenue and are vulnerable to abusive tax planning. A key problem is that Uganda’s negotiating position has been based on the UN model treaty, which embodies a compromise position, rather than an ideal one to be horse-traded during negotiations. The recent East African Community (EAC) and Common Market for Eastern and Southern Africa (COMESA) model treaties also represent compromise positions.
This paper uses a comparative analysis of treaties signed by Uganda and other neighbouring countries, combined with interviews conducted with government officials and private sector tax advisers, to assess whether Uganda’s network of tax treaties is fit for purpose, and to recommend how it could be improved through the policy review.