Do media data help to predict German industrial production?

Arbeitspapier

Expectations form the basis of economic decisions of market participants in an uncertain world. Sentiment indicators reflect those expectations and thus have a proven track record for predicting economic variables. However, respondents of surveys perceive the world to a large extent with the help of media. So far, mainly very crude media information, such as word-count indices, has been used in the prediction of macroeconomic and financial variables. In this paper, we employ a rich data set provided by Media Tenor International, based on the sentiment analysis of all relevant media information in Germany from 2001 to 2014, whose results are transformed into several monthly indices. German industrial production is predicted in a real-time out-of-sample forecasting experiment using more than 17,000 models formed of all possible combinations with a maximum of 3 out of 48 macroeconomic, survey, and media indicators. It is demonstrated that media data are indispensable when it comes to the prediction of German industrial production both for individual models and as a part of combined forecasts. They increase reliability by improving accuracy and reducing instability of the forecasts, particularly during the recent global financial crisis.

Angaben zum Objekt

• Sprache: Englisch

Verweise und Beziehungen

• Erschienen in: Series: DIW Discussion Papers ; No. 1393

Klassifikation und Themen

• Klassifikation: Wirtschaft ; Econometric and Statistical Methods and Methodology: General ; Model Evaluation, Validation, and Selection ; Forecasting Models; Simulation Methods ; Business Fluctuations; Cycles
• Thema: forecast combination, ; media data ; German industrial production ; reliability index ; R-word

Beteiligte, Orts- und Zeitangaben

• Ereignis: Geistige Schöpfung
• (wer): Kholodilin, Konstantin A. ; Thomas, Tobias ; Ulbricht, Dirk
• Ereignis: Veröffentlichung
• (wer): Deutsches Institut für Wirtschaftsforschung (DIW)
• (wo): Berlin
• (wann): 2014

Weitere Informationen

• Handle: http://hdl.handle.net/10419/98409
• Letzte Aktualisierung: 10.03.2025, 11:44 MEZ