Forecasting Inflation in Emerging Markets: An Evaluation of Alternative Models

This paper carries out a comprehensive forecasting exercise to assess out-of-sample forecasting performance of various econometric models for inflation across three dimensions; time, emerging market countries and models. The competing forecasting models include univariate and multivariate, fixed and time varying parameter, constant and stochastic volatility, small and large dataset, with and without bayesian variable selection models. Results indicate that the forecasting performance of different models change notably both across time and countries. Similar to some of the recent findings of the literature that focus on developed countries, models that account for stochastic volatility and time-varying parameters provide more accurate forecasts for inflation than alternatives in emerging markets.