How to Investigate the Impact of Foreign Direct Investment on Development and Use the Results to Guide Policy

My mentors in graduate school were Raymond Vernon, Charles Kindleberger, and Albert Hirschman. They established their reputations by selecting the most important issues and policy dilemmas on the horizon, getting to know the actors and the settings involved as thoroughly as possible, and deriving insights from intensive investigation of (often) small numbers of cases. They then had to persuade their audience that what they had discovered was not idiosyncratic, atypical, or ephemeral.

Their contributions came at the beginning of the revolution in using regressions on large datasets, which was to become the preeminent investigative technique in economics. Their own work would have had more lasting impact if they had been able to combine—as I urge here—statistical analysis with a dedication to understanding how things worked by looking into the innards of the relationships they were trying to understand.

This paper will argue that the investigation of the impact of foreign direct investment (FDI) in manufacturing on development would be much more valid, more convincing—and less prone to error—if multiple techniques of investigation were used to reinforce (or refute) each other.¹ In addition to statistical regressions, such techniques include industry and sector studies, business cases, management interviews, data collection through firm surveys, and cost-benefit analyses of individual projects that are carefully arrayed to avoid selection bias and ensure generalizability. This argument is, I believe, unexceptional and should not even be controversial. The value of making this assertion arises, alas, because the use of multiple contrasting methodological approaches is quite uncommon—nay, vanishingly rare—among economists in the contemporary period. The weaknesses of this self-imposed narrowness plague current policy debates even among informed practitioners.

This paper examines three topics: how to investigate the impact of manufacturing FDI on a developing country host economy, how to search for externalities and spillovers (and identify the channels and mechanisms through which spillovers take place), and how to evaluate the question of whether host governments should devote public sector resources to attracting FDI in manufacturing. In each area, the paper shows how overlapping kinds of evidence can be useful, and are often indispensable, to carry out accurate investigations, to avoid analytical miscalculations, and to design appropriate policies for developed and developing countries.

How to Evaluate the Impact of Manufacturing FDI on a Developing Country Host Economy

The history of efforts to investigate the outcome from multinational manufacturing investment in developing countries begins with cost-benefit analysis of individual FDI projects, industry studies, multinational business cases, management interviews, and firm surveys. Do these approaches lead to reasonably robust general propositions about the impact of manufacturing FDI on the host economy, or are they no more than impressionistic observations—anecdotal in the pejorative sense—that have a significant probability of being overturned by one or two subsequent observations? Can the results from these kinds of investigations be useful in shaping the design of large-N econometric investigations that use plant-level data as well as in steering clear of mistakes that have spoiled earlier efforts?

Early Assessments Using Cost-Benefit Analysis

The earliest systematic attempts to assess the impact of manufacturing FDI on development were carried out by contemporaries of Vernon, Kindleberger, and Hirschman. The investigators used used variations of cost-benefit analysis to measure the effect of the operations of multinational corporations (MNCs) on the host economy. One of the first was conducted by Sanjaya Lall and Paul Streeten under the auspices of the UN Conference on Trade and Development.² They analyzed the contribution to the host economy of 147 foreign investor operations in six developing countries. For approximately 62 percent (ninetytwo projects), the effect of foreign investment on national income was positive; for the remaining projects (fifty-five, approximately 38 percent), it was negative. The key determinant of whether the social rate of return was positive or negative was the extent of effective protection granted to the investors, across all industries and all countries in the sample. A second study, sponsored by the Organization of Economic Cooperation and Development (OECD) under...