Capital inflows and the real exchange rate: An empirical study of sub-Saharan Africa

Abstract

This paper investigates the question of whether capital inflows, particularly Foreign Direct Investment (FDI), cause the real exchange rate to appreciate. It also examines whether different forms of captial inflow have variable effects on the real exchange rate. The paper estimates an empirical real exchange rate model specifying a set of capital inflow variables using dynamic panel techniques. Based on data for a sample of sub-Saharan African countries for the period 1980 – 2000, the study reveals FDI as the category of private capital inflow that causes the real exchange rate to appreciate. The results also show that an increase in official aid causes a real appreciation, the magnitude being greater compared to that associated with FDI.

Key words:

• Capital inflows
• Real exchange rate
• Dutch disease
• Sub-Saharan Africa

Acknowledgements

The author is indebted to Christopher Baum and Peter Ireland for invaluable comments. The author also thanks Fabio Ghironi, Luisa Lambertini, Todd Prono and two anonymous referees for their suggestions, and is grateful to participants at R@BC seminars and workshops at Boston College. The author is responsible for all errors.

Notes

¹ The World Bank (2002: 1).

² The aforementioned dynamics, which are symptomatic of the Dutch Disease, are similar to those in the version of the dependent economy model presented by Corden and Neary (1982), hence the reference to Dutch disease effects of capital inflows. The interested reader should see Corden and Neary (1982) for specific details of that version of the dependent economy model which focuses on Dutch disease effects of the resource boom and technological shocks.

³ A major area of concern with respect to poor macroeconomic management, mentioned in the literature, is increased government expenditure on unproductive activities in the face of rising official development assistance with its inflationary consequences. Also of concern is the issue of fungibility of aid. See for instance Falck (1997) for details.

⁴ There are different real exchange rate indexes just as there are different price indexes, and therefore which definition one uses is usually a matter of choice or data availability.

⁵ See for instance Cheung and Lai (1999).

⁶ See Figure 1 (Based on data from Global Development Finance, 2002).

⁷ Negative inflows of private loans refers to the situation where repayments of past loans and interest payments exceed new loans.

⁸ See USITC (2002: 36)

⁹ See Bhattachary, et al. (l997).

¹⁰ When FDI and other types of capital flows for that matter are expressed as a percentage of GDP, it provides a measure of the ‘real’ value of that capital inflow.

¹¹ It is noteworthy that the increase in the FDI – GDP ratio is due to the ovewhelming impact of higher FDI inflows, rather than any possible decline in GDP in some of the countries in the sub-region. Figure 2 shows that FDI inflows more than tripled between 1992 and 1997 and remained high thereafter.

¹² See USITC (2002; 36 – 37).

¹³ The differences between the three groups of countries in terms of attracting FDI lies in the stability of the political and economic environments, the extent of regulation and openness of markets, and the presence and sizes of mining and oil industries, among other factors.

¹⁴ A baseline model featuring a single explanatory variable in either FDI or OCF, each serving as a proxy for private capital inflows is estimated as a first step before the fully specified model is estimated.

¹⁵ Arguably, it is unlikely that FDI undertaken in a given year will generate much output in the same year, therefore the one-period lag of FDI is included as an explanatory variable.

¹⁶ The fixed effects estimator transforms the data by subtracting the time series mean of each variable, thereby eliminating the country specific effects.

¹⁷ Under these assumptions, the GMM difference estimator is based on the following moment conditions:

¹⁸ The additional moment conditions for the regression in levels are

• See Arellano and Bond (1991) and Blundell and Bond (1998) for technical details on the GMM difference and system estimators.

¹⁹ The countries are selected based on data availability. They are Burundi, Cameroon, Central African Republic, Cote d'Ivoire, Equatorial Guinea, Gabon, The Gambia, Ghana, Lesotho, Malawi, Nigeria, Sierra Leone, South Africa, Togo, Uganda, Zambia. It should be noted that due to missing observations for some variables, the panel is unbalanced with the minimum number of time periods being 11 years and the maximum, 21 years.

²⁰ For estimation purposes, all data series are converted into logarithms. Detailed definitions of all variables are provided in the appendix.

²¹ I use two period lags of the covariates to form the instrumental variable matrix for each of the GMM estimators. In addition, both the levels and one year lags of the growth rate of GDP for OECD countries and gross domestic capital formation are used as additional instruments.

²² An examination of the time series properties of the series shows that all the capital inflow variables together with the real exchange rate are I(1). Results of the tests are given in Table 7. Furthermore, Table 8 presents a comparison of the GMM estimates to OLS and Fixed Effects estimates, an ad-hoc method for ruling out small sample biases. As the table shows, the GMM estimates lie between the OLS (higher) and the Fixed Effect estimates (lower).

²³ See Table 1.

²⁴ The results assume the same slope coefficients across the different countries, and hence conclusions from this model should be drawn with caution. Ideally, a model that allows for heterogeneous responses should be estimated, which is not feasible here due data constraints.