News Concentration

57 Pages Posted: 7 Apr 2021 Last revised: 12 Jun 2023

See all articles by Nikolas Kuhlen

Nikolas Kuhlen

University of Cambridge - Faculty of Economics; The Alan Turing Institute

Andrew Preston

University College London - Department of Economics

Date Written: June 12, 2023

Abstract

We develop a quantitative measure of one economically relevant dimension of news coverage: the degree of news concentration. Intuitively, news coverage is likely to become highly concentrated around particularly newsworthy events, meaning it potentially conveys information about the economy's latent state variables. We observe that news concentration does indeed exhibit clear spikes around important economic, financial, and political events, and find that these increases are associated with two key features: an immediate rise in uncertainty, consistent with theory, and a future macroeconomic contraction. These increases are not found to be driven by a range of typical macroeconomic shocks. The variable is a robust forecaster of recessions and is priced in the cross-section of returns. We rationalise the latter result through an ICAPM framework by showing that innovations in news concentration represent volatility news.

Keywords: text analysis, news, local projections, asset pricing, volatility, entropy, machine learning

JEL Classification: C43, D80, E32, G12

Suggested Citation

Kuhlen, Nikolas and Preston, Andrew, News Concentration (June 12, 2023). Available at SSRN: https://ssrn.com/abstract=3820449 or http://dx.doi.org/10.2139/ssrn.3820449

Nikolas Kuhlen (Contact Author)

University of Cambridge - Faculty of Economics ( email )

Sidgwick Avenue
Cambridge, CB3 9DD
United Kingdom

The Alan Turing Institute ( email )

British Library
96 Euston Road
London, NW1 2DB
United Kingdom

Andrew Preston

University College London - Department of Economics ( email )

Drayton House, 30 Gordon Street
30 Gordon Street
London, WC1H 0AX
United Kingdom

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
570
Abstract Views
8,652
Rank
87,015
PlumX Metrics