Published online by Cambridge University Press: 14 December 2017
This article provides a new methodology to predict armed conflict by using newspaper text. Through machine learning, vast quantities of newspaper text are reduced to interpretable topics. These topics are then used in panel regressions to predict the onset of conflict. We propose the use of the within-country variation of these topics to predict the timing of conflict. This allows us to avoid the tendency of predicting conflict only in countries where it occurred before. We show that the within-country variation of topics is a good predictor of conflict and becomes particularly useful when risk in previously peaceful countries arises. Two aspects seem to be responsible for these features. Topics provide depth because they consist of changing, long lists of terms that make them able to capture the changing context of conflict. At the same time, topics provide width because they are summaries of the full text, including stabilizing factors.
We thank Tim Besley, Melissa Dell, Vincenzo Galasso, Hector Galindo, Matt Gentzkow, Stephen Hansen, Ethan Kapstein, Daniel Ohayon, Akash Raja, Bernhard Reinsberg, Anand Shrivastava, Ron Smith, Jack Willis, Stephane Wolton, and the participants of the workshops and conferences ENCoRe Barcelona, Political Economy Cambridge (internal), EPCS Freiburg, ESOC in Washington, Barcelona GSE Calvo-Armengol, NBER SI Economics of National Security, Conflict at IGIER, and the seminars PSPE at LSE, BBE at WZB, and Macro Lunch Cambridge for valuable feedback. We are grateful to Alex Angelini, Lavinia Piemontese, and Bruno Conte Leite for excellent research assistance. We thank the Barcelona GSE under the Severo Ochoa Programme for financial assistance. All errors are ours.
Full text views reflects PDF downloads, PDFs sent to Google Drive, Dropbox and Kindle and HTML full text views.