AN INVESTIGATION OF BAYESIAN QUADRATIC AND EXPONENTIAL GARCH MODELS UNDER SYSTEMATIC AND ASYMMETRIC ERROR INNOVATIONS
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Department of Statistics, Nasarawa State University Keffi
DOI
Abstract
This research work focused on Investigation of Bayesian, Quadratic and Exponential GARCH
Models under Systematic and Asymmetric Error Innovations. The Bayesian approach offers an
attractive alternative which enables small sample results, robust estimation, model
discrimination, model combination, and probabilistic statements on (possibly nonlinear)
functions of the model parameters. However, the Quadratic GARCH models are used to describe
the negative skewness in stock market indices. In fact, the distribution of return can be skewed.
Naturally the symmetric GARCH models cannot cope with skewness which may result to the
forecast and forecast error variances thereby making the GARCH model to be biased for skewed
time series. Therefore, the Quadratic GARCH models is the best, when the sample does not
contain extreme observation such as the stock market crash. Consequently, the EG ARCH model
overcome some weaknesses of the GARCH models in handling financial time series for
asymmetric effects between positive and negative asset returns. The live data was gotten from
the Nigeria insurance Stocks from 1961 - 2019. The data used for the research were analyzed
using Bayesian, Quadratic, and Exponential GARCH Models. Descriptive statistics of prices of
live data as well as the returns were conducted. The result from the three selected models shows
that the quadratic and exponential GARCH model performed better than the Bayesian GARCH
model for both prices of insurance stock and their long returns. It becomes pertinent that in
embarking in insurance stock, quadratic as well as exponential GARCH model should be
adopted in carrying out the modelling of insurance stock in other to obtain optimum result.
Description
Keywords
Citation
Aruna, U.S. (2022) AN INVESTIGATION OF BAYESIAN QUADRATIC AND EXPONENTIAL GARCH MODELS UNDER SYSTEMATIC AND ASYMMETRIC ERROR INNOVATIONS
