Template-Type: ReDIF-Article 1.0

Author-Name: José Daniel Aromí
Author-Workplace-Name: Universidad de Buenos Aires, Facultad de Ciencias Económicas, IIEP-Baires

Title: Measuring uncertainty through word vector representations

Abstract: Uncertainty is approximated processing economic press content from 1900 through 2017. The indicator exploits word vector representations that are trained to identify terms that are closely related to uncertainty. The resulting index co-moves with alternative proxies for uncertainty and spikes around crisis episodes. In-sample and out-of-sample forecasting exercises indicate that the proposed metric provides valuable information on future levels of expected stock market volatility (VIX). This informational gain is not observed when simpler text processing techniques are implemented.

Classification-JEL: C5, G1.
 
Keywords: uncertainty, forecast, volatility.
Journal: Económica

Pages: 135-156
Volume: 63
Year: 2017
Month: January-December
File-URL: https://revistas.unlp.edu.ar/Economica/article/view/5074/4248
File-Format: Application/pdf
Handle: RePEc:akh:journl:610