Modelling the business cycle of South Africa: linear vs non-linear methods.

The purpose of this study is twofold. Firstly, business cycle theories have been developed as early as 1911 (Shumpeter). These theories are well researched and well documented, and all of these theories concentrate on the real sector. South Africa is an emerging market and since 1994 the country has liberalized its market, a process that holds advantages and disadvantages. This emerging market status as well as the relative size of imports and exports to GDP in South Africa, makes the country very vulnerable to changes in the world economy. Examples of this are the contagion from Asia in 1997, the Russian crisis in 1998, and the impact of September 11 in the US on the South African economy. Business cycles also have changed over the years; they are less volatile and more synchronized over the world and the financial markets play a more important role. This is another reason why it might be useful to identify a financial cycle and investigate its relationship with the real cycle. The SARB (South African Reserve Bank) has some financial indicators in its leading indicator but the latter is mainly driven by real indicators. The financial cycle identified uses the equity market, the capital market and the domestic financial market as components. All of the determinants of these three components are available at a higher frequency than the GDP growth (our proxy for the business cycle); therefore the financial cycle can be used as a leading indicator incorporating international and domestic financial events. Secondly, an ongoing debate in business cycle research is the question of a stable economy (business cycle) influenced by exogenous shocks or an unstable economy with an endogenous business cycle (Classical vs. Keynesian view). This issue will be addressed by modelling the business cycle with a linear as well as a non-linear model. Linear models are usually used to demonstrate exogenous shocks on the business cycle, whereas nonlinear models have more of an endogenous assumption regarding the business cycle. Non-linear models learn over time and adjust to the new level of peaks and troughs and can therefore predict turning points more accurately. This suggests that business cycles have changed since 1960: they became less volatile, more synchronized across the world and the amplitude of peaks and troughs is lower. Because of these characteristics it would be useful to fit a non-linear model to the business cycle. However, exogenous shocks cannot be totally ignored – especially in an emerging market such as South Africa. The STAR (smooth transition autoregressive) model makes room for a linear and a non-linear component, and can over time determine if there is only a linear or non-linear component or sometimes both. The results of this study support the structural or institutional view. They believe economic fluctuations are caused by various structural or institutional changes. Adherents to this view do not believe that the market system is inherently stable or systematically unstable (Classical vs. Keynesian view). They focus on structural changes and unpredictable events. They do not have set ideas on economic policy. According to them the appropriate policy will vary from time to time as circumstances change.Prof. L. Greylin