RESEARCH AND DEVELOPMENT PERFORMED ABROAD BY U.S. MANUFACTURING MULTINATIONALS.

It is worth noting that 82% of the cross industry variation in PFRD has been explained without using any explanatory variable measuring the cost advantages of an American versus a 'rest of the world' R & D location. While such a variable certainly belongs in the regressions run, it is and will remain prohibitively difficult to obtain unless the OECD or some other international body publishes suitable comparative data[38]. Significant multicollinearity between this omitted independent variable and the eight others used in the regressions is, with the possible exception of LWRDI, a very remote possibility.
This study develops and empirically tests a 'scenario' describing the evolution of a typical American multinational manufacturing corporation's (AMMC's) foreign research and development (R&D) activity. It finds that international differences in R & D factor costs probably have very little effect on R & D location decisions. The critical variables appear instead to be less traditional things like the time elapsed since the AMMC first engaged in foreign production. That such variables were discovered and emphasized in the study is a result of the 'satisficing' approach adopted in its 'scenario'. In vindicating that approach, the paper's findings also show the need for economic theorizing to take explicit account of the fact that information costs something to obtain and to use. [ABSTRACT FROM AUTHOR]