Automatic Forecasting of Economic Times Series: An analisis of TRAMO/SEATS and X-13-ARIMA performance

This paper presents a comparative study of the forecasting performance of the most widely used ARIMA modelling and seasonal adjustment programs: TRAMO/SEATS (TSW) and X13-ARIMA, which are the most important software alternatives for automatically modelling economic time series. The newest versions of the programs (updated recently) are used to carry out the analysis. In particular, TSW was updated last May 2012 (R197 version) and X-13 is an update of the previous X-12, released on July 2012. Forecast and real values are compared and an analysis based on ROC-curves performed. Results will be of interest to statistical offices, central banks and corporate and academic users, and, more generally, to statisticians and economists involved with univariate modelling and forecasting of seasonal monthly or quarterly time series. To assess the performance of both programs when forecasting, thousands of European and Unites Stattes economic real time series are identified and adjusted using both programs’ modelling default options and one, four and twelve months ahead predictions obtained. Results are collected and an analysis using ROC-Curves is performed, from which it is possible to obtain an efficiency index, whatever the cut-off point selected for the acceptability of the model detection by the automatic on each software procedure.