ABSTRACT
This study examines the very short, short, medium and long-term forecasting ability of different univariate GARCH models of United Kingdom (UK)'s interest rate volatility, using a long span monthly data from May 1836 to June 2018. The main results show the relevance of considering alternative error distributions to the normal distribution when estimating GARCH-type models. Thus, we obtain that the Asymmetric Power ARCH (A-PARCH) models with skew generalized error distribution are the most accurate models when forecasting UK interest rates, while for the short, medium and long-term term forecasting horizons, GARCH models with generalized error distribution for the error term are the most accurate models in forecasting UK's interest rates.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 Taylor [Citation27] showed in some financial time series, the sample autocorrelation of absolute returns was larger than that of squared returns.
2 The test is applied using the ‘fUnitRoots’ package in R.