ناهمگونی شرکت و ساختار فعالیت های چند ملیتی آمریکا

We use firm-level data for U.S. multinational enterprises (MNEs) and the model of firm heterogeneity presented in Helpman, Melitz, and Yeaple [Helpman, E., Melitz, M., Yeaple, S., 2004. Exports versus FDI with heterogeneous firms. The American Economic Review 94 (1), 300–316.] to make four empirical contributions. First, we show that the most productive U.S. firms invest in a larger number of foreign countries and sell more in each country in which they operate. Second, we assess the importance of firm heterogeneity in the structure of MNE activity. Third, we use the model to identify the mechanisms through which country characteristics affect the structure of MNE activity. Finally, we assess the model's shortcomings in order to inform the development of new theory.

A tiny minority of firms engage in international trade, and a still smaller fraction of firms own production facilities in more than one country. These internationally engaged firms are systematically different from their domestically oriented peers. Firms that export are larger and more productive than firms that do not (Bernard and Jensen, 1999), while firms that open foreign affiliates are still larger and more productive than firms that only export (Tomiura, 2007). A well-developed body of theory explains these phenomena as the sorting of heterogeneous firms into modes of foreign market access.1 In this paper, we investigate how well a model of firm heterogeneity adapted from Helpman et al. (2004) can explain the cross-country structure of U.S. multinational activity. The model is built on two key assumptions. First, firms face a trade-off in serving foreign markets: by opening a local affiliate, firms avoid per unit transport costs associated with trade but must instead incur fixed costs associated with managing a foreign affiliate. Second, firms differ in their productivity. These assumptions imply that for each country there is a productivity cutoff, which is determined by the country's characteristics, such that only those firms whose productivity exceeds this cutoff will open an affiliate in that country. Hence, the model predicts a “pecking order” such that the most productive firms should open an affiliate in even the least attractive countries, while progressively less productive firms enter progressively more attractive countries. In the model, country characteristics affect the aggregate volume of multinational activity, measured as the sales of affiliates to host customers, through two channels. First, country characteristics determine the productivity cutoff, and so affect the productivity composition of the firms that invest there. The key feature of the model is that a change in a country characteristic that encourages a greater number of foreign firms to open a local affiliate must be inducing progressively less productive firms to enter. Second, country characteristics determine the optimal level of sales, holding fixed the set of firms that own an affiliate there. We use the model and firm-level data collected by the Bureau of Economic Analysis for all U.S. multinational enterprises for the year 1994 to make four empirical contributions. First, we show that more productive U.S. firms own affiliates in a larger number of countries and these affiliates generate greater revenue on sales in their host countries. Previous studies, such as Girma et al. (2005), Head and Ries (2003), and Tomiura (2007) show only that firms that become multinational are systematically different from firms that export. Our analyses demonstrate that this sorting extends to the scale and scope of multinational enterprises: more productive firms own affiliates in a larger set of countries and their affiliates are larger than those of less productive firms. This sorting has quantitatively important implications for the aggregate structure of U.S. multinational activity. The second contribution of our analysis is to assess how important firm heterogeneity is in determining the structure of U.S. multinational activity. We show that as a country becomes more attractive to U.S. multinationals, it attracts progressively smaller and less productive firms. For instance, our estimates suggest that a 10% increase in a country's GDP per capita leads to a 7.6% increase in the number of U.S. firms that enter that country, but because new entrants are less productive than old entrants, the average productivity of all entrants falls by 2.0%. Thus, the contribution of the extensive margin, adjusted for the productivity composition of entrants, is 5.6%. Although there are several papers exploring the importance of firm heterogeneity models in the structure of international trade (e.g. Eaton et al., 2008), little has been done to investigate whether this body of theory improves our understanding of the aggregate structure of multinational activity. Our third contribution is to use the structure of the model to disentangle the mechanisms through which individual country characteristics affect the structure of U.S. multinational activity. For example, while it has long been known that horizontal FDI is primarily attracted to developed countries, previous analyses do not shed light as to exactly why this is the case. We show that multinational activity is increasing in a host country's GDP per capita because individual entrants face greater effective demand in richer countries, not because these countries have relatively lower entry costs. The analysis generates similarly surprising conclusions concerning the mechanisms through which physical distance and a shared language affect the structure of U.S. multinational activity.2 The fourth contribution of this paper is to assess the manner in which the model fails. We document systematic deviations from the pecking order. Although multinational activity is highly concentrated in the most productive firms, the model predicts that an even greater concentration of affiliate sales in the largest firms than is actually observed. In particular, larger firms underinvest in the least attractive countries. These observations should prove useful for the future development of models of firm heterogeneity.3 The remainder of this paper is divided into five sections. In section 2, we use a version of Helpman et al. (2004) to derive predictions over the investment behavior of individual firms and to specify a structural econometric model of aggregate multinational activity. In section 3, we describe the firm-level data and the set of firm and country characteristics used to estimate this model. In section 4, we present the main results of our empirical analyses. The results confirm that it is important to account for firm heterogeneity in order to understand the structure of aggregate multinational affiliate sales, and they illuminate the channels through which country characteristics influence multinational activity. In section 5, we calculate the structure of multinational affiliate sales that would be observed if the pecking order were strictly observed and compare this counterfactual measure to the actual structure in order to demonstrate precisely how the model falls short. The final section concludes and presents suggestions for future research.