Revisiting the Long-Run Relationship Between Inward/Outward FDI and Income Inequality: New Evidence from the OECD

Abstract

The relatively small panel cointegration literature on the dynamics between FDI and income inequality predominantly finds that FDI will reduce income inequality in the long-run in developed countries. However, we point out an important technical oversight in the literature. Not accounting for cross-section dependence in panel data methodologies may yield unreliable results. Expanding on the work of Herzer and Nunnenkamp [(2013). Inward and outward FDI and income inequality: Evidence from Europe. Review of World Economics, 149(2), 395–422. https://doi.org/10.1007/s10290-013-0148-3], who pioneered the use of panel cointegration in the European context, we obtain different results when we account for cross-section dependence and employ economic procedures robust to it. Using a panel containing 16 OECD countries (1979–2017), 2 income inequality measures, and 4 FDI measures, we begin by showing strong evidence for the existence of cross-section dependence. Then, using second-generation econometric procedures, we do not find any evidence for a cointegrating relationship between inward FDI and income inequality. We do find evidence that outward FDI is cointegrated with income inequality; however, contrary to the main results of the literature, we find that it widens the income gap in the long-run. Additionally, our results support the view that fiscal policy is an important tool to reduce income inequality.

Keywords:

• Foreign direct investment
• income inequality
• panel cointegration
• cross-section dependence

JEL Classifications:

• F02
• F15
• F62

Disclosure Statement

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

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Notes

1 The models using FDI as a flow variable span from 1979–2017 and the models using FDI as a stock variable span from 1980 to 2017.

2 The test results were computed using EViews 12.

3 The test results were computed using the R statistical language.

4 It is important to note that the tests can be augmented with common time dummies to address cross-section dependence. The results reported here; however, do not include the time dummies.

5 The test results were computed using the R statistical language.

6 The test results were computed using Gauss 19.

7 See Stock (1987) and Phillips (1991) for details.

8 The test results were computed using Gauss 19.

9 The test results were computed using Gauss 19.

10 This is corroborated by the literature. See (Botta et al., 2022; Doytch & Uctum, 2011), and Golden and Wallerstein (2011)

Additional information

Notes on contributors

Mert Akyuz

Mert Akyuz is a doctoral student in economics in Ankara Yildirim Beyazit University (Turkey). His research interests are macroeconomics and international economics.

Ghislain Nono Gueye

Ghislain Nono Gueye is an assistant professor of economics at Louisiana Tech University (USA). His research interests are macroeconomics and international economics.

Cagin Karul

Cagin Karul is a doctoral student in econometrics in Pamukkale University (Turkey). His research interests are macroeconomics and panel data econometrics.