Spillovers from Multinationals in Developing Countries: The Mechanisms at Work by Richard E. Caves
Summary of WDI Working Paper 247

Early case studies of foreign direct investment identified several channels for spillovers—transfers of knowledge that result in productivity increases—from foreign subsidiaries to local host country enterprises. Statistical analyses, which were initially crude but have recently been based on improved data, have mostly confirmed the positive relation between the prevalence or productivity of foreign subsidiaries and the productivity of local firms that compete with or supply them. Some negative results appear in high-quality studies, however, and other investigations suggest that the incidence of spillovers varies substantially with the country and industry setting in which they (may) occur.

The central function of a spillover is to reduce the inequality of knowledge stocks. Few hypotheses and little evidence exist, however, concerning the factors that determine the leakage of knowledge stocks and their absorption and effective deployment by firms in developing countries. What knowledge-related assets get transferred to local firms, and when are foreign subsidiaries the premier source of supply? This article makes a start at answering these questions by specifying the economic behavior associated with the "spillable" knowledge possessed by foreign subsidiaries and the ability of local firms in developing countries to capture them.

Why local enterprises in developing countries should lack knowledge or skills that could have been acquired from foreign subsidiaries raises questions about the role of the representative firm’s organization and growth as part of the process of economic development. But with a few exceptions, no systematic theory or empirical work has emerged to address this seemingly central issue. The Handbook of Development Economics, 46 chapters filling four weighty volumes, has essentially nothing to say about the subject, for example.

Low levels of national economic development are associated with a paucity of large and complex business organizations. Put another way, the poorer the country, the larger the number of independent business units relative to the working population.

One study (Caves and Uekusa 1976) seeks to explain variation in enterprise densities (the number of employers and own-account workers as a fraction of the labor force) across countries. (An alternative measure added family workers to the numerator.) The sample includes 34 countries with widely varying income levels for which reasonably comparable census-based date for the late 1960s were available. The key explanatory variables are per capita GDP (expressed in U.S. dollars at the current exchange rate) and GDP squared. The results show that enterprise density decreases significantly with GDP per capita, although at a declining rate, leveling off at about the income level then prevailing in Canada and Sweden.

The study also finds evidence of a lagged adjustment process in the decline of the enterprise ratio. The ratio increases with the country’s growth rate in the preceding two decades, which suggests that the squeeze out or consolidation of very small enterprises lags behind the development process (in principle, though apparently not in practice, it could be a leading component of the development process).

This empirical regularity suffices to demonstrate that the capacity to mobilize and run complex business organizations is a correlate of the development process. The correlation between development and the number of large business enterprises underlines the significance of spillovers from foreign subsidiaries and other external sources. It also shows that the role of spillovers depends on exactly what constrains the indigenous development of complex enterprises. We consider some possible explanations, treating them in isolation while recognizing that they are likely to interact:

· Cultural traditions based on nonhier-archical interpersonal relationships may make it difficult to accept the hierarchical structures that exist in large firms. Hierarchical relationships among individuals are surely unnatural in some societies. Rank-order tournaments as a standard way to staff an effective hierarchy require acceptance of interpersonal competition and of ways to compete interpersonally that are hardly natural for all cultures and religions. Such constraints could limit the feasible size of the business organization and promote decisions that diverge from the goal of maximizing value by responding to social norms of conduct formed outside the business environment. (Ponder, for a moment, the revenge culture of the Balkans run rampant within a business hierarchy.)

· Local enterprises in developing countries may suffer from a lack of human capital. Activities such as coordination and supervision require flexible and abstract problem-solving capability, specialized bodies of skill or training, or both. Most developing countries have made rather modest investments in advanced general education; some developing countries have invested more heavily in training, in particular technical skills. But the team aspect of business hierarchies suggests that skills tend to be complements rather than substitutes.

· Local enterprises, lacking in both knowledge and experience, may not even be able to assess what types of information they are missing. Shortages of knowledge are distinguishable from shortages of skills in several ways. Skills are encapsulated in individuals and subject to clear economic definition. Everybody knows that a firm needs a system of accounts and that specialized training is needed to be able to maintain them. In contrast, tacit and unencapsulated knowledge such as the process of forming a business partnership is an "experience good." Unless it invests in information, the firm remains highly uncertain about its value. Significant investment decisions (importing a proprietary technology, entering an export market) demand some stocks of knowledge that are identifiable, albeit probably costly to obtain. However, the firm may not even realize that it is missing other knowledge components salient to the investment decision. Consulting and other firms, of course, stand ready to sell such knowledge stocks, but the pervasive inability to guarantee satisfaction to the buyer in such transactions marks them for market failures of the familiar "lemons problem", in which the seller knows the true quality of a product, but the buyer does not. Low-quality products ("lemons") may drive out high-quality products.

· Enterprises in developing economies may also lack general management and coordinating skills, a problem related to, but distant from, the lack of specific skills and information. That managerial capability limits firm size in developing countries is consistent with a pattern that is well documented in industrial countries. The compensation of chief executive officers increases sharply and regularly with the sizes of the firms they manage, with an elasticity in the neighborhood of 0.3. An obvious explanation of this and related evidence is that the value of talent at the top increases with the volume and complexity of the activities to be coordinated, so that the managerial labor market allocates the best managers to the biggest tasks. Other factors might also explain the relation, including weaker governance by shareholders in the largest firms, but the efficient allocation of talent is surely the main force at work. A national economy less well endowed with managerial talent would likely adjust by reducing the average size of firms and making the widest possible use of the top managerial talent available, which is arguably done by the business groups commonly found in developing countries. In sum, the circumstantial evidence that managerial capability is an important constraint on the productivity and size of local business units in developing countries is compelling.

The four sources of local firms’ shortfalls have different potentials for remedy through spillovers. A business organization based on community relations is probably least prone to spillovers. Such an organization is based on community and collective behavior patterns That punish deviations by individuals. No private benefit rewards those who shift to new ways, even if a collective shift would substantially increase the firm’s value. Firms that operate within the traditional system instead lose market share where such tradition is a less productive institution. Such cultural patterns are not immune to change, of course. They may gradually give way, as individuals exposed to market-based systems (through education and travel, for example) swell to a critical mass that is willing to try different ways. That process is a sort of spillover, but not one that proceeds product market by product market.

Whether spillovers alleviate skills shortages depends on the subsidiaries’ effect on the net supply of skills to local firms. As demanders of skilled labor, foreign subsidiaries tend to drive up its price to local firms. They likely induce positive shifts in supply by providing training themselves, however, or by triggering the establishment of independent training institutions. The degree to which foreign subsidiaries are able to retain workers they have trained becomes an important factor affecting this spillover. The unfeasibility of binding long-term employment contracts works to the disadvantage of subsidiaries and reduces their incentive to provide training; it also increases the expected spillover benefit to local firms. The widespread finding that foreign subsidiaries in developing countries (though not elsewhere) pay higher wages than local firms for labor of a given quality may partly reflect an effort to realize the benefit of investments in training.

Knowledge shortages can be allevaited by the classic public-good property of information and its appropriation. The most helpful spillage from competing foreign subsidiaries may well be local firms’ chance to observe what is feasible. That a better-quality product can be made and sold at the observed price and that locally produced goods can be profitably exported to particular foreign markets, that a foreign-designed product serves well under local conditions (even before adaptive tinkering), are important facts for the local firm contemplating obtaining costly information and otherwise unable to confirm its likely value.

The responsiveness of managerial shortages to spillovers may depend on the situations of local firms and competing foreign subsidiaries. Managerial talent can be obtained or developed from various external sources, such as MBA programs. What a competing foreign subsidiary can supply incrementally is an exemplar of managerial talent at work in the specific context (industry, location) of the local firm’s operations. For this spillover to prove substantial, the two firms must be roughly similar in size. If they are not, the managerial tasks of the firms will differ too widely, and the subsidiary’s experience will be either irrelevant or impossible for the local firm to apply.

In thinking about the potential for spillovers of general management skills, it is important to keep in mind that multinational enterprises tend to operate mainly in industries in which the managerial task is most complex; that is, in the coordination of disparate skills in the pursuit of uncertain outcomes. Because of the complexity of the managerial task, spillovers in developing countries are not likely to occur until the development process is fairly far along. Spillovers of general management capability may be low for countries far from the global frontier of the industrial efficiency, high for countries that have reached a moderate level of development, and low for advanced countries in which local firms’ productivity is close to that of foreign subsidiaries. This is consistent with the fact that multinationals in developing countries flourish in industries that place fewer demands on managerial capability but also flourish in industries with requirements more differentiated between developed-country and LDC environments (Wells 1983).

Interest in spillovers is motivated by two central concerns. First, although productivity spillovers from foreign subsidiaries to local firms are apparently widespread, they are neither ubiquitous nor independent of the market structure in which the firms operate. Second, and more important, the study of economic development has paid little attention to the seemingly vital question of what factors affect local firms’ ability to increase productivity by improving their managerial and organizational capabilities. Further research on both issues is needed to develop effective policies to promote development.

Spillovers may provide a justification for governments in developing countries to encourage inflows of foreign direct investment. The justification is likely to be conditional on the country’s state of development and the structures of industries in which foreign subsidiaries might compete. Favorable treatment might thus be prudent for some countries or industries but not others. Requiring multinationals to provide training or purchase inputs locally could increase the welfare of the host country. The effectiveness of such policies depends, however, on evidence that is not yet in hand (the evidence that does exist suggests that such programs have had very limited success [Conklin and Kecraw 1997]). The finding is thus no open invitation to bureaucratic whim (Caves and Uekusa 1976, Wells 1983, and Conklin and Kecraw 1997).

Richard Caves is Nathaniel Ropes Professor of Economics at Harvard University.