Foreign Direct Investment

Bruce A Blonigen

• Publisher: WSPC
• Year: 2019
• eBook ISBN: 9789813277021
• Topic: Volkswirtschaftslehre
• Subtopic: Internationale Wirtschaftswissenschaften

This book is a collection of important reference works by Bruce Blonigen on foreign direct investment (FDI). The book is primarily composed of empirical analyses of foreign direct investment behavior from an industrial organization and international trade perspective. These studies both examine determinants of FDI, as well as the effects of FDI on host and parent countries. The work examines FDI from a firm-level perspective and uses unique micro-level datasets to further understand firm-level FDI decisions and their impact on local economies. The volume should be valuable to both existing and new scholars in the field of international trade and international business. Much of the work is policy-oriented and so it will also be valuable to policymakers who follow the academic literature.

Contents:

• About the Author
• Introduction
• Determinants of FDI:
  • Determinants of Foreign Direct Investment (Bruce A Blonigen, Jeremy Piger)
  • FDI in Space: Spatial Autoregressive Relationships in Foreign Direct Investment (Bruce A Blonigen, Ronald B Davies, Glen R Waddell, Helen T Naughton)
  • Firm-Specific Assets and the Link Between Exchange Rates and Foreign Direct Investment (Bruce A Blonigen)
  • Cherries for Sale: The Incidence and Timing of Cross-Border M&A (Bruce A Blonigen, Lionel Fontagné, Nicholas Sly, Farid Toubal)
  • Tariff-Jumping Antidumping Duties (Bruce A Blonigen)
  • Endogenous Protection, Foreign Direct Investment and Protection-Building Trade (Bruce A Blonigen, Yuka Ohno)
  • The Differential Effects of Bilateral Tax Treaties (Bruce A Blonigen, Lindsay Oldenski, Nicholas Sly)
  • Industrial Groupings and Foreign Direct Investment (Bruce A Blonigen, Christopher J Ellis, Dietrich Fausten)
• Effects of FDI on Home and Host Countries:
  • In Search of Substitution Between Foreign Production and Exports (Bruce A Blonigen)
  • Foreign-Affiliate Activity and U.S. Skill Upgrading (Bruce A Blonigen, Matthew J Slaughter)
  • The Effects of Foreign Direct Investment on Local Communities (David N Figlio, Bruce A Blonigen)
  • Please Pass the Catch-Up: The Relative Performance of Chinese and Foreign Firms in Chinese Exports (Bruce A Blonigen, Alyson C Ma)
  • Voting for Protection: Does Direct Foreign Investment Influence Legislator Behavior? (Bruce A Blonigen, David N Figlio)
  • Tariff-Jumping FDI and Domestic Firms' Profits (Bruce A Blonigen, KaSaundra Tomlin, Wesley W Wilson)
  • Measuring the Benefits of Foreign Product Variety with an Accurate Variety Set (Bruce A Blonigen, Anson Soderbery)

Readership: Graduate students and researchers in international economics, globalization and industrial organization.Foreign Direct Investment;Multinational Enterprises;Mergers and Acquisitions;Greenfield;Trade Policy;Taxation;Spillovers;Offshoring;Wage Inequality;Firm-Specific Assets;Antidumping;Tariff-jumping;Industrial Organization;Ownership-Location-Internalization Theory0 Key Features:

• Provides a rich set of empirical analysis of foreign direct investment (FDI) behavior from an industrial organization, firm-level perspective as opposed to the substantial scholarship on FDI that is more focused at the country- and macroeconomics level
• Uses unique, micro-level datasets
• Contains a number of key works in the literature examining the interactions of FDI and trade policy

Part I

Determinants of FDI

Chapter 1

Determinants of Foreign Direct Investment^∗

Bruce A. Blonigen, Jeremy Piger^†

University of Oregon

^†[email protected]

Abstract. Empirical studies of bilateral foreign direct investment (FDI) activity show substantial differences in specifications with little agreement on the set of included covariates. We use Bayesian statistical techniques that allow one to select from a large set of candidates those variables most likely to be determinants of FDI activity. The variables with consistently high inclusion probabilities include traditional gravity variables, cultural distance factors, relative labour endowments and trade agreements. There is little support for multilateral trade openness, most host-country business costs, host-country infrastructure and host-country institutions. Our results suggest that many covariates found significant by previous studies are not robust.

Résumé. Les déterminants de l’investissement direct à l’étranger. Les études empiriques des déterminants des activités d’investissement direct bilatéral à l’étranger ont des spécifications substantiellement différentes et peu d’accord sur les variables co-reliées incluses. On utilise des techniques statistiques bayesiennes qui permettent de balayer un vaste ensemble de variables à la recherche de celles qui sont davantage susceptibles d’être des déterminants des activités d’investissement direct à l’étranger. Les variables qui se retrouvent de manière régulière dans la liste de haute probabilité d’impact sont les variables reliées à la gravité, les facteurs liés à la distance culturelle, les dotations relatives en facteur travail, et les accords commerciaux. Il y a peu de support pour des variables comme l’ouverture au commerce multilatéral, la plupart des coûts d’affaires, les infrastructures et les institutions dans les pays hôtes. Ces résultats suggèrent que plusieurs co-variations qu’on a jugées significatives dans les études antérieures ne sont pas robustes.

JEL Classification: F21, F23, C52

1.Introduction

Empirical analyses of the factors determining foreign direct investment (FDI) across countries have employed a variety of econometric specifications. Many previous studies of cross-country FDI activity have used a gravity equation, which controls mainly for the economic size of the parent and host countries, the geographic distance separating the countries and proxies for certain economic frictions. Like trade flows, this specification does a reasonably good job of fitting the observed data, but leaves one wondering if such a parsimonious specification captures all relevant factors.

Recent papers by Carr, Markusen and Maskus (2001; CMM) and Bergstrand and Egger (2007) have developed theoretical models of multinational enterprise’s (MNE’s) foreign investment decisions that suggest additional possible factors that determine FDI patterns. These studies point out a number of modifications to a standard gravity model that may be necessary to accurately explain FDI patterns. First, while gravity variables may adequately capture “horizontal” motivations for FDI, where firms look to replicate their operations in other countries to be more proximate to consumers in those markets, additional controls are necessary to allow for “vertical” motivations of FDI, where firms look for low-cost locations for labour-intensive production. For example, these studies introduce measures of relative labour endowments in the host country with the expectation that countries with relatively high shares of unskilled labour will be attractive locations for MNEs due to lower wages. In addition, these studies show that FDI decisions by MNEs are complex enough that interactions between key variables (e.g., GDP and skilled labour endowments) may be necessary to account for nonlinear effects of these variables on FDI patterns. Head and Ries (2008) differs from these previous studies by modelling FDI as arising from decisions by firms to acquire and control foreign assets (i.e., cross-border mergers and acquisitions), rather than development of new (or greenfield) plants. Their analysis of FDI patterns highlights the potential role of common culture and language between countries.

While these prior studies have been important in deepening our understanding of the factors that determine cross-country FDI patterns, they have generally focused on regression models involving specific sets of covariates determined by the researcher and the particular theoretical framework for FDI they chose to examine. By conditioning on a particular regression model specification, this practice ignores uncertainty regarding the model specification itself, which can have dramatic consequences on inference.¹ Most notably, inference regarding the effects of included covariates can depend critically on what other covariates are included versus excluded.

In this paper, we take a Bayesian approach to confront uncertainty regarding the appropriate set of covariates to include in a regression model explaining FDI activity. From a Bayesian perspective, incorporating such uncertainty is conceptually straightforward. The choice of covariates, or “model,” is treated as an additional parameter that lies in the space of potential models, which allows us to compute the posterior probability that each potential model is the true model that generated the data. Posterior distributions for objects of interest, such as the effect of a particular covariate, are then averaged across alternative models, using the posterior model probabilities as weights. This procedure, known as Bayesian Model Averaging (BMA), produces inferences that are not conditioned on a particular model.

To be clear, we are taking a purely empirical approach to determine the correlates with observed FDI patterns. As we discuss in the next section, there is very little consistency in the empirical FDI literature about the covariates one should use when empirically modeling cross-country FDI. We view this paper as a first step in pointing out these inconsistencies and providing evidence of the empirically robust determinants of FDI.

Although conceptually straightforward, BMA is practically difficult when the set of possible models is large, as direct calculation of posterior probabilities for all models becomes infeasible. In our application, we have a large set of potential covariates, which yields an extremely large set of possible models (>7 × 10¹⁶). To sidestep this difficulty, we use techniques designed to obtain random draws of models from the probability distribution defined by the posterior model probabilities. Such draws are made possible even when the posterior model probabilities are unknown by using the MC³ algorithm of Madigan and York (1995). These random model draws are then used to construct estimates of the posterior model probabilities.²

Our set of potential FDI determinants is meant to be comprehensive and includes a combination of covariates proposed by the previously mentioned studies, as well as other prior literature on FDI. We examine mainly cross-sectional patterns for the year 2000.² We examine both levels and log-linear regressions, placing more weight on our results for the log-linear regressions because most previous studies have used a logarithmic transformation to address skewness in the FDI variable. We also examine three measures of FDI — FDI stock, affiliate sales and cross-border mergers and acquisitions activity — in order to better compare with a broader set of prior studies. At the end, we also explore a specification that first differences observations across the years 1990 and 2000 to control for bilateral country-pair fixed effects as well as a negative binomial specification to better model the nature of our dependent variable.

Our analysis indicates that many of the covariates used in prior FDI studies (and often found statistically significant) do not have a high probability of inclusion in the true FDI determinants model once we consider a comprehensive set of potential determinants using BMA. A fairly parsimonious set of covariates is suggested by our analysis. The covariates with consistently high inclusion probabilities include traditional gravity variables, cultural distance factors, relative labour endowments and trade agreements. Variables with little support for inclusion are multilateral trade openness, most host-country business costs, host-country infrastructure (including credit markets) and host-country institutions. A few variables that have rarely been included in prior FDI studies, namely host-country remoteness, parent-country real GDP per capita and host is an oil-exporting country, have surprisingly high inclusion probabilities.

The remainder of the paper proceeds as follows. The next section reviews previous empirical literature on the determinants of FDI and makes the case that the appropriate model specification for explaining FDI patterns is far from settled. Section 3 then lays out the BMA methodology we use to assess model uncertainty. Section 4 describes the data and its sources, while section 5 reports the results and compares to the existing literature. Section 6 concludes.

2.Prior FDI Literature

There is little consensus on how to empirically model bilateral FDI patterns, with many past empirical FDI papers using a base model consisting of gravity-type covariates (country-level GDP and distance) because of its popularity for explaining trade flows. As mentioned in the introduction, there have been a few recent efforts to develop specifications based on theoretical models — namely the knowledge-capital (K-K) model developed by James Markusen and co-authors, which was brought to data in CMM (2001), Bergstrand and Egger’s (2007) model incorporating physical capital and Head and Ries’ (2008) model of acquisition FDI.

There is little consistency in the covariates that are postulated to explain worldwide FDI patterns across these three papers. To see this, the first three columns of table 1 lists the covariates used in each of these papers. Distance between countries is the only covariate common to all three studies. There are 22 different covariates between...

Table of contents

• Cover
• Halftitle
• Series Editors
• Title
• Copyright
• About the Authors
• Contents
• Introduction
• Part I. Determinants of FDI
• Part II. Effects of FDI on Home and Host Countries