Robust FDI determinants: Bayesian Model Averaging in the presence of selection bias

Abstract

The literature on Foreign Direct Investment (FDI) determinants is remarkably diverse in terms of competing theories and empirical results. We utilize Bayesian Model Averaging (BMA) to resolve the model uncertainty that surrounds the validity of the competing FDI theories. Since the structure of existing FDI data is well known to induce selection bias, we extend BMA theory to HeckitBMA in order to address model uncertainty in the presence of selection bias. We show that more than half of the previously suggested FDI determinants are not robust and highlight theories that do receive robust support from the data. Our selection approach allows us to identify the determinants of the margins of FDI (intensive and extensive), which are shown to differ profoundly. Our results suggest a new emphasis in FDI theories that explicitly identify the dynamics of the intensive and extensive FDI margins.* We thank an anonymous referee, Christian Lorenczik, Monique Newiak, and Chris Papageorgiou for helpful suggestions. Assaf Razin and Hui Tong kindly shared their data. Lenkoski gratefully acknowledges support by the joint research project "Spatio/Temporal Graphical Models and Applications in Image Analysis" funded by the German Science Foundation (DFG), grant GRK 1653 as well as the MAThematics Centre Heidelberg (MATCH). Eicher thanks Max Soto Jimènez, the Instituto de Investigaciones en Ciencias Económicas, and the Department of Economics at the University of Costa Rica for their support and hospitality during the preparation of this paper. 1 IntroductionGlobal HeckitBMA reveals not only the determinants of the intensive and extensive margins of FDI ("the volume of investment flows" and "the decision to invest", respectively), it also permits us to estimate FDI determinants without having to constrain parameter estimates to be identical across both margins. There is no reason to suspect that the margins of FDI should feature identical determinants, nor that the same determinant has the identical impact for both margins. Our selection criterion is based on Razin,Rubinstein and Sadka (2004), who note that FDI involves fixed costs that give rise to two-part decisions: a marginal productivity condition that determines how much to invest, and a total profitability condition that indicates whether or not to invest abroad. Previous studies have confirmed the relevance of such FDI fixed costs. 4Our results show that the impact of model uncertainty on FDI estimates is substantial and that the Heckman selection methodology is necessary to obtain unbiased and consistent estimates. In the absence of explicit controls for model uncertainty, the conventional Heckit procedure suggests nearly three times as many FDI determinants asHeckitBMA at the extensive margin (32 vs 13) and a 50% more regressors at the intensive margin. This is not surprising, since Heckit is not designed to consider models associated with alternative theories. Instead, HeckitBMA discovers much more parsimonious models of FDI that score better as measured by the Bayesian I...

Want Export Diversification? Educate the Kids First

“…In particular, BMA has recently been applied to further identify the true determinants of economic growth—a literature that has been pioneered by the seminal papers of Fernandez, Ley, and Steel , Sala‐i Martin, Doppelhofer, and Miller (), and, most recently, Moral‐Benito in a panel setting. Other applications in the international macroeconomic context include those of Eicher, Henn, and Papageorgiou , who analyze the determinants of the gravity equation model, and Eicher, Helfman, and Lenkoski , who focus on foreign direct investment. In general, Moral‐Benito provides a detailed survey on the use of BMA methods in economics.…”

A multi-criteria decision-making model for evaluating priorities for foreign direct investment

“…Our study employs fifteen robust FDI criteria based on the study of Eicher et al [10] which constructs the determinants by considering two important constraints in an FDI database, i.e. uncertainty and selection bias.…”

“…Researchers have focused on several factors as determinants of FDI inflow-outflow, including bilateral distance [23,22,5], trade costs in the host and home countries, gross domestic product (GDP), GDP growth rate [35], market size of the host country relative to other countries [11,2,7,10], productivity [28], differences in factor endowments, exchange rate volatility [12], financial system development, economic policy activity [17], financial risk [28], host and source corporate tax rates [27], the existence of regional trade agreements [3], country-pair specific impacts, such as language [25], border and colonial history, legal protection, and the quality of institutions [6].…”

“…It is well worth mentioning that missing data jeopardizes the validity of FDI theories as well. Recently, [10] carried out a study to construct robust FDI determinants. They employed Heckit Bayesian Model Averaging (HBMA), which is concerned with model uncertainties regarding the validity of competing FDI theories and selection bias.…”

“…Using this methodology, which results in robust FDI determinants, they found that more than 50% of the suggested FDI determinants were not robust. We constructed our model based on fifteen robust FDI determinants, namely distance, market size, colony, common language, development, GDP growth, market potential, productivity, tax, LAIA, APEC, dollar, corruption risk, internal conflict risk, and religious tension risk as suggested in the study conducted by [10].…”