Constructing and Characterising the Aggregate South African Financial Cycle: A Markov Regime-Switching Approach

The financial cycle has become a central point of consideration for policymakers since the 2007–2008 financial crisis. This study aims to construct and characterise the aggregate South African financial cycle. A dynamic factor model is employed to construct an aggregate South African financial cycle based on seven key South African financial components. Additionally, the Christiano-Fitzgerald bandpass filter is utilised to extract the aggregate South African financial cycle. Also, a spectral density analysis is implemented as a means of determining the cyclical duration of the financial cycle. Furthermore, a Markov regime-switching autoregressive model is employed to model and further characterise the aggregate South African financial cycle. Data are monthly and range from 01/01/1975 to 01/12/2017. Results show that South African credit conditions and South African equity market conditions are the strongest underlying drivers of the aggregate South African financial cycle, followed by South African bank balance sheet conditions. The most common duration of the financial cycle is eight years and five months, making it a Juglar cycle. The results also indicate that contractions in the aggregate South African financial cycle are typically shorter but harsher and deeper than cyclical expansions and that a level of cyclical persistence exists in the cycle. These findings have particular application for macroprudential policymakers.