What robustly determines FDI in emerging markets and developing countries? A sensitivity analysis

Abstract

What determines Foreign Direct Investment (FDI) inflows remains a primary concern of economists and policy makers; yet the uncertainty surrounding FDI theories and empirical approaches has created ambiguity regarding the determinants of FDI. This paper applies Leamer’s (1983, 1985) traditional Extreme Bounds Analysis (EBA) as well as the Sala-I-Martin modified form of EBA to identify the robust determinants of FDI using cross-sectional data (averaged for the period of 1985-2016) covering 103 emerging and developing countries. We consider 33 potential economic, political and institutional determinants. To address concerns about multicollinearity, the VIF restriction specifies that only estimates with VIF below five are reported. The results show that all variables of interest are fragile determinants of FDI when using traditional EBA. Yet, when applying the Sala-i-Martin EBA, nine robust variables are revealed. Some policy implications are discussed.

Highlights

• What determines Foreign Direct Investment (FDI) inflows remains a primary concern of economists and policy makers; yet the uncertainty surrounding FDI theories and empirical approaches has created ambiguity regarding the determinants of FDI.

• It is a common practice in the fields of economics and finance to report the most ‘appealing’ or convenient regression or regressions after an extensive search and data mining, given that the ‘true’ model is unknown.

• This paper applies traditional Extreme Bounds Analysis (EBA) as well as the Sala-i-Martin modified form of EBA to identify the robust determinants of FDI.

• We demonstrate that what is important is not the significance or otherwise according to one regression equation but rather the robustness or fragility of the variables in the sense that the sign and significance do not change over a large number of regressions selected according to a predetermined procedure.

• Policy makers would be able to understand the importance of the major determinants of FDI mentioned in the paper and take steps to formulate policies that encourage and attract more FDI. Such determinants could include market size, making regulations more international trade friendly and investing in the nation’s human capital. Further, research and development facilities could be developed to provide a basis for technological advancements which would attract more FDI inflows.

KEYWORDS:

• Foreign direct investment
• extreme bounds analysis
• emerging markets
• developing countries
• sensitivity analysis

JEL Classifications:

• F21
• F23
• F30
• C19

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The general-to-specific methodology is explained in detail by Hendry (1995). The proponents of the LSE approach include McAleer, Pagan and Volker (1985), Hoover and Perez (1999), and Campos, Ericsson and Hendry (2005). For more sceptical accounts, see Leamer (1978, 1983, 1985), Hansen (1996, 1999), and Faust and Whiteman (1995, 1997). Three common concerns with the general-to-specific approach are repeated testing, recursive procedures, and selection of variables.

2 Granger and Uhlig (1990) develop a restricted reasonable EBA. Reasonable EBA estimates the upper and lower extreme bounds of the coefficient of interest $\text{β}$ and eliminates models with poor goodness of fit (measured by R²). The R² decision rule specifies that the extreme bounds of $\text{β}$ should be selected from a distinct set of models that meet a predetermined level of goodness, in which case all models with low R² are deemed irrelevant for the calculation of EBA. This criterion is represented by the following equation: $R_{\mathrm{\varnothing }}^2\ge \left[\right(1-\mathrm{\varnothing }){\text{R}}_{max}^2+\mathrm{\varnothing }{\text{R}}_{min}^2]$