The negative consequences of financial instability for the world economy during the recent financial crisis have highlighted the need for a better understanding of financial conditions by policy-makers and decision-makers all over the world, and more importantly, their impact on the real economy. It is for this reason that I conduct a study of South Africa‟s financial conditions and their impact on and implications for the real macroeconomy.
In order to meet this objective, I construct a financial conditions index (FCI) for the South African economy so as to ascertain whether: (1) financial conditions in South Africa have long-term effects on the macroeconomy; (2) South Africa‟s FCI can be regarded as an early warning system, and; (3) the nature of the impact of the FCI on the macroeconomy is linear or nonlinear.
This thesis begins with the compilation of an FCI for South Africa using a number of different approaches. The best FCI is chosen from these alternatives, namely a rolling-window principal components analysis (PCA) approach. The FCI is then purged of endogenous macroeconomic feedback effects emanating from output, interest rates and inflation. The performance of this FCI is evaluated by assessing its ability to pick up turning points in the South African business cycle, and by running in-sample causality (forecast) tests against the major macroeconomic variables of output, inflation and an interest rate. It is found that the FCI does a good job of reflecting recessionary periods in South Africa, and causality tests indicate that this FCI is a good in-sample predictor of industrial production growth and the Treasury Bill yield, but a weak predictor of inflation.
I then go on to ascertain whether this FCI has good out-of-sample forecasting ability with respect to the major macroeconomic variables, as compared to the 16 individual financial time series which make up the FCI. A host of forecast encompassing tests are conducted, and their results are adjusted for data-mining. It is found that the estimated FCI has good out-of-sample forecasting ability with respect to manufacturing output growth at the one, three and six month horizons, while it has no predictive power for inflation and the Treasury bill yield. Therefore, the FCI can be regarded as a leading indicator of manufacturing output growth.
Finally, the FCI is inserted into a nonlinear vector autoregressive (VAR) framework, so as to test for asymmetry in the effects that financial conditions may have on the macroeconomic variables of output, interest rates and inflation in South Africa. I make use of a nonlinear logistic smooth transition vector autoregression (LSTVAR), which allows for the transition of a chosen switching variable between upper and lower regimes. I estimate two such models: one with inflation as a switching variable; and one which allocates a different switching variable to each equation within the LSTVAR. I find that the South African economy is strongly nonlinear in its responses to financial shocks, and that manufacturing output growth and interest rates are more affected by financial shocks during upswings, while inflation responds more during downswings.