An Assessment of Alternative Models of Interest Rate Volatility in the Bond Market in Kenya

Abstract

This research project focuses on estimating volatility in interest rates in the bond market in Kenya. It assesses the linear and non linear models of estimating volatility. Data comprising of redemption yields for all bonds issued since January 1995 was obtained from NSE. In the analysis six different models were assessed in estimating volatility. These six models were the random walk, moving average (MA), autoregressive (AR), autoregressive moving average (ARMA), autoregressive integrated moving average (ARIMA), autoregressive conditional heteroskedasticity (ARCH) and generalised autoregressive conditional heteroskedasticity (GARCH). The Akaike Information Criterion (AIC) were used to rank the models. This study found out that the bond market in Kenya mainly comprised of treasury bonds and twelve listed corporate bonds. It was also observed from the data that the number of participants and trading frequency still remain low as compared to other developed bond markets. The non-linear models ie ARCH and GARCH models scored the lowest AIC values. From the findings of this study it was concluded that nonlinear models better estimate volatility as compared to linear models.