The Global Networks of Multinational Firms (with Maggie Chen)

In this paper we characterize the topology of global multinational networks and examine the macro and micro patterns of multinational activity. We construct indices of network density at both pairwise industry and establishment level and measure agglomeration in a global and continuous metric space. These indices exhibit distinct advantages compared to traditional measures of agglomeration including the independence on the level of geographic aggregation. Estimating the indices using a new worldwide establishment dataset, we investigate both the significance and causes of multinational firm co-agglomeration. In contrast to the conventional emphasis of the literature on the role of input-output linkages, we assess the effect of various agglomeration economies. We find that, relative to counterfactuals, multinationals with greater capital-market externalities, knowledge spillovers and vertical linkages exhibit significant co-agglomeration. The importance of these factors differs across headquarter, subsidiary and employment networks but knowledge spillovers and capital-market externalities, two traditionally under-emphasized forces, exert consistently strong effects. Within each macro network, there is a large heterogeneity across subsidiaries. Subsidiaries with a greater size and a higher productivity attract significantly more agglomeration than their counterfactuals and become the hubs of the network.