Worldwide versus territorial tax systems: The impact on U.S. foreign direct investment

Reform of the corporate tax regime in the United States is a much debated and contentious issue. One key aspect of a tax regime in a global environment is the assignment of tax jurisdictions. Regarding tax jurisdictions, there are two basic types of tax systems. A territorial tax system is one in which income earned abroad is totally or partially exempt from home country tax. A worldwide system, in contrast, is one in which income earned abroad is subject to home taxes, often with credit for taxes paid abroad. Under a worldwide system, therefore, income repatriated from abroad to a high-corporate tax country such as the United States, can be subject to substantial additional taxes.

            Though the tax regime is a very important issue that affects the international competitiveness of firms, surprising little empirical work considers how the two different regimes might affect the foreign direct investment decisions of U.S. multinationals. The proposed research will empirically model and test the relationship between a host-country’s tax system, either worldwide or territorial, and the pattern of U.S. FDI outflows. The main research hypothesis is that U.S. total FDI outflows and U.S. retained earnings to host countries with a territorial system are, all other things constant, larger than flows to host countries with a worldwide system. The secondary hypothesis is that tax credits offered by countries with a worldwide system reduce or eliminate this differential.

The data considered here consists of annual observations of total U.S. FDI outflows to 53 different OECD and non-OECD countries covering the period 1982 to 2012. In additional to total bilateral flows, the data is disaggregated into its three financing flows; retained earnings, equity capital, and inter-company debt. The main variables of analysis are control variables for the type of tax system (worldwide or territorial) of the host country, a control for those nations with a worldwide system that allow for a tax credit, and the interaction of these two variables. Other control variables based on the empirical literature will be included in the empirical model. Because the research design employs panel data whose distribution suffers from skewness and kurtosis, a non-parametric empirical model will be used.  A two-step approach within STATA to employ this model as a fixed-effects estimator.