South African Historical Interest Rate Volatility - Evidence of Regime- Switching

Abstract Accurate estimates of volatility are important for the valuation and risk management of financial assets. The benchmark interest rate for many assets in South Africa, the 3-month Jibar, is found to exhibit a high volatility persistence when modelled with a standard generalised autoregressive conditional heteroskedasticity (GARCH) process. The literature suggests that unobserved regime-switching in interest rate data may lead to an overestimation of volatility persistence. In this study 120 GARCH-type volatility models are tested to determine which model, conditional distribution and number of regimes best fit the data in order to extract the most accurate in-sample estimation of historical volatility for asset pricing. The data analysed consists of 877 weekly observations of 3-month Jibar in total, spanning from September 2001 to July 2018. The switching between regimes is governed by a Markov chain process which produces state-dependent transition probabilities for the unobserved regimes. The study finds that a standard single regime GARCH model may not capture the underlying volatility dynamics of the interest rate. The best performing model in this study is the 4 State Threshold-GARCH indicating that in addition to evidence of regime-switching in the data, there is an asymmetric reaction to negative information.