Comments

Gordon H. Hanson:

This paper examines whether inward foreign direct investment (FDI) creates positive spillovers for host countries. This question has generated immense intellectual interest. Many countries demonstrate a strong policy bias in favor of subsidizing multinationals, predicated on the belief that FDI is a source of positive externalities for host-country firms and workers. While this belief has some support in economic theory and in anecdotal evidence, it has not found support in microeconomic-level empirical research. In developing countries, plant-level analysis suggests that the productivity of domestic establishments is low in industries with a large presence of foreign firms.¹ The findings of empirical research are thus at odds with the actions of policymakers.

Alfaro and Rodríguez-Clare propose rethinking the empirical evidence about FDI and spillovers. They correctly observe that much of the empirical literature does not specify a structural relationship between FDI and host-country outcomes and thus may be using reduced-form empirical specifications that are uninformative. They proceed to derive the theoretical conditions under which positive spillovers from FDI obtain and then examine whether this condition is satisfied empirically using plant-level data on manufacturing industries in Brazil, Chile, Mexico, and Venezuela.

Their theoretical framework builds on Rodríguez-Clare (1996), in which the arrival of multinational firms in an industry lowers the price for domestically produced inputs and consequently raises the productivity of domestic plants. These backward linkages amount to a positive spillover from multinationals to domestic firms in the same industry. The existence of positive spillovers requires that domestic inputs are nontraded and that multinationals are sufficiently intensive in their use of intermediate inputs. Alfaro and Rodríguez-Clare find that the second condition is satisfied empirically for Brazil, Chile, and Venezuela, but not for Mexico. For three of the four countries in the sample, then, there is evidence consistent with positive spillovers from FDI.

A main contribution of the paper is to show that previous empirical literature misinterprets evidence about input usage by multinationals. The fact that the share of inputs purchased domestically by multinationals tends to be lower than for domestic firms has been viewed as evidence against backward linkages from multinationals. As the authors nicely demonstrate, however, if multinationals are more intensive in the use of intermediate inputs overall, this makes up for the small share of inputs they purchase locally. Thus, the pessimism in the empirical literature about the potential for multinationals to create backward linkages may have been unwarranted.

The paper stops short of estimating the actual magnitude of productivity spillovers associated with backward linkages, although the authors do provide something of a roadmap for how research might proceed. First, a necessary condition for backward linkages is that a large portion of domestic inputs are nontraded. Much of the empirical literature pools data across industries, ignoring the tradability of inputs. Input tradability is thus a key potential source of industry heterogeneity in terms of the sign and size of spillovers. While many inputs used by firms in developing countries are traded (for example, parts and components), many others are not (such as water, power, and some raw materials). An important task for future empirical work is to estimate the tradability of inputs used in developing countries. If most inputs turn out to be tradable, then the previous pessimism on backward linkages may have been warranted.

Second, the theoretical framework of Alfaro and Rodríguez-Clare also suggests a potential source of cross-country heterogeneity in the strength of backward linkages that has been ignored in previous work. In their model, if the intensity of the use of intermediate inputs by multinationals is increasing in the wage-rental ratio, then there could be poverty-trap equilibrium for poor countries.² That is, the condition for positive spillovers may not be satisfied in poor countries (with low wage-rental ratios), because intermediate-input intensity is too low. The arrival of multinationals would then generate negative spillovers, making poor countries even poorer. The condition for positive spillovers may be satisfied, however, in rich countries (with high wage-rental ratios). In this case, the arrival...