Essays on exchange rate behaviour in South Africa

Khomo, Melvin Muziwakhe (2018-03)

Thesis (PhD)--Stellenbosch University, 2018.


ENGLISH SUMMARY: The dissertation presents four essays on exchange rate behaviour in South Africa with a focus on misalignment and volatility of the real effective exchange rate (REER), and the concomitant influence of exchange rate uncertainties on both economic growth and exports. Motivating the study is the proposition in recent literature that proactive exchange rate policies centred on exchange rate disequilibria and deliberate currency undervaluation can foster economic growth through exports. Despite exchange rate behaviour receiving much attention in the literature, several questions regarding the pass-through effects to the economy remain unanswered in the South African context where the country faces the challenges of low economic growth, high unemployment and significant inequality. The major question therefore is whether exchange rate policy can be used in South Africa to promote exports, thus aiding economic growth and employment creation. To answer this question, I model exchange rate misalignment and volatility, and then assess the influence of both factors on growth and exports with an emphasis on the possible asymmetries in the reaction to undervaluation and overvaluation episodes. The results are presented in four essays. In this context, the first essay applies cointegration techniques in the behavioural equilibrium exchange rate (BEER) framework of Clark and MacDonald (1998) to estimate the equilibrium value of the rand consistent with economic fundamentals, and interpret the deviation of the observed exchange rate from this level as exchange rate misalignment. A Markov regime switching method is then applied to quantify whether the exchange rate’s departure from the equilibrium level is meaningful enough to be considered as either over- or undervalued. The results indicate that a long-run equilibrium relationship exists between the rand’s REER and economic variables that include the terms of trade, external openness, external capital flows and government expenditure. Frequent deviations of the observed exchange rate from the estimated equilibrium level are found over the period studied (1985-2014) and the Markov regime-switching model correctly captures the exchange rate misalignment as distinct episodes of exchange rate overvaluation and undervaluation. It is important to note that the observed misalignments have been mainly in response to economic shocks emanating from either the South African economy or global factors and not in response to deliberate policy action. In the second essay, univariate symmetric and asymmetric generalised autoregressive conditional heteroscedasticity (GARCH) approaches are used to model the volatility of the South African rand’s REER. The objective is to explore the relevance and compare the performance of the GARCH family of models in terms of their ability to capture the stylised facts of the rand’s volatility and thus identify the best model to apply for volatility modelling and forecasting. The findings from the essay show that exchange rate volatility in South Africa is quite persistent and exhibits volatility clustering and asymmetric effects. In terms of sample-fit, the results confirm that the GJR-GARCH (1,1) with the normal error distribution is the best fitting model for the rand’s REER volatility, when compared to the ARCH, GARCH (1,1) and EGARCH (1,1) models with in-mean and different error distribution assumptions. The GJR-GARCH (1,1) model is able to accurately capture the significant increases in exchange rate volatility experienced in South Africa over the sample period, with such episodes of high volatility linked to the historical exchange rate depreciation experiences (e.g. at the height of the political crisis in 1985, the emerging market crisis of 1997-1998, speculative attacks on the rand in 2001 and lastly the global financial crisis of 2007-2008). Asymmetric autoregressive distributed lag (ARDL) cointegration methods are used in the third essay to explore the effects of exchange rate misalignment on economic performance. Specifically, I investigate the response of economic performance to exchange rate misalignment depending on the direction and size of misalignment such that the reaction would be contingent on whether the exchange rate is undervalued or overvalued, and if the magnitude of the misalignment (small or large) is important in explaining the influence on the economy. Given the fact that both exchange rate misalignment and volatility represent uncertainty, the essay also seeks to ascertain whether a combination of both uncertainty indicators could be important in explaining the exchange rate’s influence on economic activity. Although not robust, the results confirm asymmetry in the reaction of economic performance in South Africa to exchange rate misalignment, with exchange rate undervaluation of approximately 10% being positively correlated with economic performance. Although not statistically significant, the results indicate that exchange rate overvaluation has a negative influence on economic performance in South Africa. Finally, the essay concludes that exchange rate volatility, as specified through a GJR-GARCH (1,1) model, does not have a significant influence on real GDP. The last essay empirically assesses the reaction of South Africa’s exports to exchange rate misalignment occurrences and volatility with the major focus being possible asymmetries in such a relationship. This study brings in a new dimension to the literature by making a comparison as to whether it is only exchange rate misalignment or volatility, or a combination of both, that influences South African exports at both aggregate and sectoral levels. At the aggregate level, the findings confirm asymmetry as exports appear to benefit less from exchange rate undervaluation than they suffer from an overvalued exchange rate. The same observation is true for manufactured exports, while the results confirm no meaningful relationship between exchange rate misalignment and agricultural and mining exports. On volatility, it is found that this variable on its own does not have an influence on exports. When considered together with misalignment (overvaluation and undervaluation), volatility exerts a negative and statistically significant influence on both gross and manufactured exports. Regarding the size of misalignment (for both overvaluation and undervaluation), based on the sample data, the study fails to confirm evidence of hysteresis in the reaction of exports to exchange rate misalignment for both aggregate and manufactured exports. This result is probably a confirmation of the weak influence of exchange rate misalignment on exports in South Africa. In summary, the thesis contributes to the literature on modelling exchange rate misalignment and volatility in South Africa. The main contribution of the study is through testing for asymmetries in the reaction of economic performance and export performance to exchange rate developments in the country. The results from the study provide credence to the view that maintaining the exchange rate at an appropriate competitive level is desirable as a measure to boost manufactured exports and growth, although such a policy should be secondary to labour productivity, and a good supportive infrastructure that allows manufacturers to produce at full capacity, together with macroeconomic stability. From a policy perspective, efforts to avoid exchange rate overvaluation and smooth out excessive currency volatility are desirable as a measure to support South Africa’s economic performance. In episodes of exchange rate appreciation and concomitant overvaluation, the Reserve Bank could use such opportunities to intervene in the foreign exchange market as a measure to boost foreign exchange reserves. Further research in this area could focus on the application of different methods to estimate the equilibrium exchange rate (e.g. FEER, NATREX) or the influence of exchange rate misalignment on other economic variables such as investment, imports, the balance of payments or employment.

AFRIKAANSE OPSOMMING : Geen opsomming beskikbaar.

Please refer to this item in SUNScholar by using the following persistent URL:
This item appears in the following collections: