The impact of corruption on FDI: is MENA an exception?

Abstract

The eruption of the Arab Spring in Tunisia and Egypt was ensued by deterioration in FDI inflows. Whether a new Middle East free of corruption accompanying previous dictatorships will offset the negative ramifications of the uprisings and enhance FDI in the long run remains debatable. Since the evidence on the causal relationship between corruption and FDI is inconclusive, this study attempts to take another step. The paper investigates the link between corruption and FDI flows to the Middle East and North Africa (MENA) and assesses whether or not corruption has more importance than other FDI determinants. By employing several panel settings with various econometric specifications on 21 MENA countries over the period 2003 to 2009, it is demonstrated that FDI varies positively with corruption. Additionally, FDI in MENA was found to vary positively with per capita income, openness, freedom and security of investments and negatively with the tax and homicide rates. Since corruption was not found to hinder FDI inflows, treating corruption should be based on sound legal procedures that infringe neither on the rights, freedom and security of FDI nor on the degree of openness and freedom of the economy, which are the real stimulants of FDI in MENA.

Keywords:

• foreign direct investment
• corruption
• determinants
• panel data analysis
• Middle East and North Africa (MENA)
• Arab Spring
• fixed effects
• GMM

Notes

1. There is no strict definition of MENA. For example, the World Bank definition differs from the definition of the Economic Research Forum for Arab Countries, Turkey and Iran (ERF) in that it includes Algeria, Djibouti, Egypt, Iran, Iraq, Jordan, Lebanon, Libya, Morocco, Syria, Tunisia, West Bank and Gaza, Yemen and leaves out the rest of the Arab countries (including all the GCC countries) in addition to Turkey. In our study we include Arab countries, Turkey, Iran, Israel and Pakistan, since the latter is sometimes considered as part of the Greater Middle East.

2. The World Bank classification of the MENA region is more confusing as it does not include Turkey and the GCC countries; we therefore preferred not to represent its data.

3. For example, in the UK the share of public expense in GDP (public expense is cash payments for operating activities of the government in providing goods and services) was 46.4% in 2009 while its CPI records a high rate of 7.7. Similarly, the respective percentage for Israel in 2009 is 40.6% while its CPI rate was 6.1 (World Bank data online and Transparency International online).

4. According to Transparency International, corruption is defined as the ‘abuse of entrusted power for private gain’ (Transparency International online).

5. See for example the various definitions of openness cited in Yanikkaya (2003).

6. The measurement of openness as [Exports + Imports]/GDP is sometimes criticized as being biased in favor of small-sized economies by incorrectly implying their higher degree of openness relative to larger-sized economies; for example, when using this measure a small country such as Singapore appears to be more open than the US, which is the world’s largest trading partner. However, the simplicity and availability of this measure makes it the most widely used by researchers.

7. It is worth noting that Eviews excludes any observation from the total sum of observations when the data on any variable in any year on any country is missing.

8. Initial estimation of the model using panel least squares produced a model with a low explanatory power and an insignificant F-statistic.

9. Random effects estimation, which is used when the unobserved effects are assumed to be uncorrelated with the explanatory variables in each time period (Wooldridge 2002), was also conducted. However, when the model was first estimated using panel, fixed effects and random effects estimators, the explanatory power proved to be stronger in the fixed effects estimations. Additionally, the Hausman test for correlated random effects supported the fixed effects estimator. Accordingly, we will estimate the models using panel GLS and fixed effects estimation.

10. The intercept that appears in the fixed effects estimation represents the average of all individual intercepts, so that the sum of the fixed effects estimates adds to zero. Individual fixed intercepts can be interpreted as deviations from the overall mean. The recorded R-squared and F-statistics reflect the difference between the residuals sum of squares from the estimated model and the sum of squares from a specification based on a single constant; and not the fixed effects only specification. These statistics thus interpret the explanatory power of the whole specification including the fixed effects estimation.