Forecasting Distribution of Inflation Rates: Functional Autoregressive Approach

Abstract

In line with the recent developments on the statistical analysis of functional data, we develop the semiparametric functional autoregressive (FAR) modeling approach to the density forecasting analysis of national inflation rates using sectoral inflation rates in the UK over the period January 1997-September 2013. The pseudo out-of-sample forecasting evaluation and test results provide an overall support to superior performance of our proposed models over the aggregate autoregressive models and their statistical validity. The fan-chart analysis and the probability event forecasting exercise provide a further support for our approach in a qualitative sense, revealing that the modified FAR models can provide a complementary tool for generating the density forecast of inflation, and analyse the performance of the central bank in achieving announced inflation target. As inflation targeting monetary policies are usually set with recourse to the medium-term forecasts, our proposed work may provide policymakers with an invaluably enriched information set.JEL Classification: C14, C53, E31.

On the Asymmetric U‐Shaped Relationship between Inflation, Inflation Uncertainty, and Relative Price Skewness in the UK

“…Our analysis of UK sectoral inflation draws heavily on Chaudhuri, Kim, and Shin (, henceforth CKS) . We utilize CKS's data set, which is based on subsector consumer price indices and the associated weights obtained from the Office for National Statistics.…”

Modelling the Density of Inflation Using Autoregressive Conditional Heteroscedasticity, Skewness, and Kurtosis Models

“…Additionally, Roger (2000) found evidence towards right skewness. In addition, Chaudhuri, Kim, and Shin (2011) found that the mean inflation is positively correlated with variance and skewness. These results suggest that a greater attention must be paid to increases than decreases of inflation rates.…”

“…Although ARCH family models are quite useful in modelling time-series variation in conditional volatility, these models assume that the conditional distribution are time-varying only in the first two moments and ignore the information content in higher-order moments (Chaudhuri, Kim, and Shin, 2011). To fill this gap, Harvey and Siddique (1999) developed a new approach to estimate nonconstant conditional skewness.…”

“…To allow for nonconstant conditional kurtosis, Leon et al (2005) developed the methodology of Harvey and Siddique (1999) by jointly modelling timevarying variance, skewness and kurtosis (GARCHSK model) assuming that the error term is derived by G-C series expansion of the normal density function which is easier to estimate than the noncentral t-distribution suggested by Harvey and Siddique (1999). Chaudhuri, Kim, and Shin (2011) introduce a semi-parametric Functional Autoregressive (FAR) model for forecasting a time-varying distribution of the sectoral inflation rates in the UK. Their approach employs the autoregressive operator to specify the time-dependence of the distribution function and thus allows all the moments to depend on all the past moments.…”