Exchange market pressure, stock prices, and commodity prices in West Africa

Abstract

As Africa continues its decade of rapid economic growth, the continent also faces the risk of becoming more susceptible to financial ‘contagion.’ Capital flows and trade linkages might cause one country’s currency market to influence those of its neighbors. Likewise, shocks to global commodity or asset markets might induce a crisis in one or more countries in the region. This study generates monthly measures of exchange market pressure (EMP) for four individual West African countries, as well as for the WAEMU franc zone, from 2002 to 2012. Vector Autoregressive (VAR) methods are then used to test for linkages among them, as well as to analyze the effects of various external price shocks. A number of spillovers are uncovered. More importantly, local connections dominate global ones in the case of stock- and commodity-price declines. Ghana, for example, is shown to be a ‘commodity currency’ when West African commodity prices are included in the VAR, but not when a global index is used.

Keywords:

• exchange market pressure
• commodity prices
• contagion
• vector autoregression
• Africa

JEL Classifications:

• F31
• F41

Notes

1. Using the Census X-12 procedure.

2. The relevant rates include the T-bill rate for the Gambia, Nigeria, and Sierra Leone; the discount rate for Ghana; and the money market rate for WAEMU and South Africa. All represent the best available short-term rate, given data limitations.

3. Using principal components analysis to combine UK, French, and US stock indices resulted in a measure that reflected the US index almost entirely. In addition, the results below show that the separate results are unique to each country, so that combining more than one measure in a seven-or-more variable VAR is not likely to eliminate the results that were found in the six-variable VARs.

4. Because of criticism of the EMP series’ weighting scheme, an alternative measure is constructed as the first principal component of the three EMP components listed in equation (1). Of the five West African countries, three (Ghana, Nigeria and the WAEMU) have the ‘proper’ signs on the three components (negative for reserve changes and positive for the other two). This PCA measure is highly correlated with the one generated in equation (1), but it shows less extreme ‘spikes’ and therefore is less sensitive in measuring currency crises. However, since it can only be used in certain cases, it is not applied here.

5. We also calculate Generalized Forecast Error Variance Decompositions (FEVDs) for all specifications. These support our findings. In the interest of brevity, they are available upon request from the author.

6. This index was chosen arbitrarily. As shown in Table 5 and it does not minimize the AIC or SIC; but the inter-country responses are robust to the choice of index.