Economic Forecasting Based on Copula Quantile Curves and Beliefs

Jianxu Liu, Songsak Sriboonchitta, Thierry Denoeux

Abstract

This paper applies belief functions-based copula quantile curvesmodel to capture dependence structure between crude oil and corn returns, andquantify uncertainty of the corn returns at one step period. We employ the time-varying copulas, including Gaussian, T and Clayton, which can be used to capturedynamic correlations between variables. We forecast their correlation ahead of oneperiod, and the uncertainty of corn returns ahead of one period is measured underp-th copula quantile curves. The empirical results show the range of corn returnsand its uncertainties under 5% and 95% copula quantile curves. In addition, thetime-varying T copula describes the dependence structure between crude oil andcorn returns quite well.