Bayesian inference for inflation volatility modeling in Ghana

Abstract

Purpose

The purpose of this paper is to emphasize the risks involved in modeling inflation volatility in the context of macroeconomic policy. For countries like Ghana that are always battling economic problems, accurate models are necessary in any modeling endeavor. We estimate volatility taking into account the heteroscedasticity of the model parameters.

Design/methodology/approach

The estimations considered the quasi-maximum likelihood-based GARCH, stochastic and Bayesian inference models in estimating the parameters of the inflation volatility.

Findings

A comparison of the stochastic volatility and Bayesian inference models reveals that the latter is better at tracking the evolution of month-on-month inflation volatility, thus following closely the data during the period under review.

Research limitations/implications

The paper looks at the effect of parameter uncertainty of inflation volatility alone while considering the effects of other key variables like interest and exchange rates that affect inflation.

Practical implications

Economists have battled with accurate modeling and tracking of inflation volatility in Ghana. Where the data is not well-behaved, for example, in developing economies, the stochastic nature of the parameter estimates should be incorporated in the model estimation.

Social implications

Estimating the parameters of inflation volatility models is not enough in a perpetually gyrating economy. The risks of these parameters are needed to completely describe the evolution of volatility especially in developing economies like Ghana.

Originality/value

This work is one of the first to draw the attention of policymakers in Ghana towards the nature of inflation data generated in the economy and the appropriate model for capturing the uncertainty of the model parameters.