Modeling stock market return volatility in the presence of structural breaks

Abstract

This study sought to model the stock market return volatility at the Nairobi Securities Exchange (NSE) in the presence of structural breaks. Using daily NSE 20 share index for the period 04/01/2010 to 29/12/2017, the market return volatility was modeled using different GARCH type models and taking into account four endogenously identified structural breaks. The market exhibited a non-normal distribution that was leptokurtic and negatively skewed and also showed evidence for ARCH effects, volatility clustering, and volatility persistence. We found that by considering structural breaks, volatility persistence was reduced, while leverage effects were found to lead to explosive volatility. In addition, investors were not rewarded for taking up additional risk since the risk premium was insignificant for the full period. However, during explosive volatility, investors were rewarded for taking up more risk. Moreover, we found that risk premium, leverage effects, and volatility persistence were significantly correlated. The GARCH (1,1) and TGARCH(1,1) models were found to be the best fit models to test for symmetric and asymmetric effects respectively. While the GARCH models were able to provide evidence for the stylized facts in the NSE, we conclude that the presence or absence of these features is period specific. This especially relates to volatility persistence, leverage effects, and risk premium effects. Caution should, therefore, be taken in using a specific GARCH model to forecast market return volatility in Kenya. It is thus imperative to pretest the data before any return volatility forecasting is done.

Review on volatility and return analysis including emerging developments: evidence from stock market empirics

“…Though a lot of studies have been conducted to understand this abrupt behavior, empirics have tried to generalize the short-term price adjustments (Malpezzi, 1999). Most of the researchers have applied the benchmark models to capture the price changes, but in the current scenario, these economic models do not deliver a suitable solution to capture the un-sudden movements (Ndei et al , 2019). These uncertain variations create outbursts in the form of volatility spillovers (Natarajan et al , 2014).…”

Modelling Stock Returns Volatility and Asymmetric News Effect: A Global Perspective

“…In the developing markets of Africa, there have been research efforts towards modelling volatility of stock market returns and asymmetries in Nigeria, South Africa, Kenya, Morocco, Egypt, etc. (Bello, 2020;Jebari & Hakmaoui, 2017;Kuhe, 2018;Ndei, Muchina, & Wawure, 2019). These empirical studies, though with varying findings, used country specific market indices (such as, the Dow Jones, S&P 500, FTSE 100, Nikkei index, NIFTY index, NSE index, etc.)…”

“…and applied in recent empirical studiesBello (2020);Ndei et al (2019);Banumathy and Azhagaiah (2015);Jegageevan (2015) the GARCH model was adopted to unravel stock returns volatility from a global perspective. In the GARCH model, the conditional variance is a function of its previous own lags.…”

“…In Malaysia,Lai et al (2019) used daily time series data spanning from 1996 to 2016 with GARCH (1,1) model to show that stock returns of oil & gas sector were most volatile. In Kenya,Ndei et al (2019) analyzed the Nairobi Stock Market from 2010 to 2017 using GARCH(1,1) and TGARCH(1,1) and revealed persistent returns volatility, leverage effects, and absence of riskreturn trade-off. Using daily time series from 1997 to 2018,Caporale et al (2019) showed that volatility persistence in the Korean stock market was driven by buy and sell trades depending on the type of investor trading and phase of business cycle.…”