Profitability of Joint Ventures Abroad : Explaining a New Empirical Puzzle

We explore empirical patterns in the profitability of foreign affiliates of US multinational companies (MNCs). Historically, affiliates that were organized as 50/50 or minority-owned joint ventures (JVs) have been less profitable than their majority and wholly owned ventures. The latter had roughly a 6 percent return on assets from the 1970s to the 1990s, compared to 4 percent for the joint ventures.This pattern held across most industries and regions, though the size of this “profitability gap” varied. In the 2000s, this profitability gap narrowed and even reversed itself in some years, regions, and sectors.To explain these patterns, we propose a simple model based on economics of project finance, classic FDI theory, and a combination of the resource based and transaction-cost theories of the firm.We argue that both the ownership structure and the profitability of a foreign venture are determined by the resources of the MNCs and of potential host country partners. The profitability gap then shows the revealed competitive advantage of US MNCs vis-à-vis local firms. We consider alternative explanations, which we do not think explain the full pattern observed.The data and the simple model we build have important implications for further research.In particular, we call for researchers to revisit the logic of asset bundling in creating value (Hennart, 2009; Gomes-Casseres, 2015), and for them to link this logic more explicitly than has been done so far with the organizational theories that shape governance structures. The schools of thought based on the works of Edith Penrose and of Ronald Coase must be seen to represent two sides of a coin, not competing theories (Teece, 2014). A proper synthesis of these views will help in explaining foreign market entry strategies, effects of government FDI policies, and the distribution of gains between foreign investors and local economies.The research presented in this chapter is preliminary in that we are continuing to refine our empirical strategy. We do not yet have sufficiently detailed data to support multivariate regression analysis, and so rely here on basic statistics and trends.Full article (chapter) available at https://doi.org/10.1017/9781108236188.020