/Forecast

Forecasting monthly Economic Policy Uncertainty (EPU) using various time series techniques

Primary LanguageR

Forecast

Abstract

Markets are characterized by volatility. Volatility can be caused by different factors in- cluding uncertainty of future development. The aim of this paper is to examine a poten- tial relationship between market development and economic policy uncertainty (EPU) by quantifying movements in policy-related economic uncertainty. After carrying out a descriptive analysis of the EPU time series, other predictors were identified in order to carry out in-sample analyses and out-of-sample prediction of market performance em- ploying past EPU (and possibly other predictors) and vice versa. A number of linear models were used to forecast the EPU, including ARMA models and models that com- bine AR components with external covariates. In the out-of-sample analysis the latter strongly outperformed the former and indicated the necessity of accounting for external variables for predicting the EPU. Models specified to capture the underlying dynamics of industrial production growth lead to the conclusion, that the additional explanatory value of EPU growth for industrial production growth is limited. For the linkage of EPU growth and stock market volatility we find statistically significant effects which we consider to be interesting and with potential for deeper exploration.

Repository Structure

  • data: Time series data for EPU and all predictors
  • images: Plots generated in the R scripts that are used in the paper
  • lib: some helper functions
  • literature: some useful references
  • source: additional scripts that were used
  • Predicting_Monthly_EPU_Final_Paper: the paper
  • regressions.rmd: regressions and tests