Working Paper: CEPR ID: DP17410
Authors: Wanyu Chung; Duiyi Dai; Robert Elliott
Abstract: In this paper we develop a series of Brexit uncertainty indices (BUI) based on UK newspaper coverage. Using unsupervised machine learning (ML) methods to automatically select topics, our main contribution is to generate timely and cost-effective indicators of uncertainty. In further analysis we are able to distinguish Brexit related uncertainty from the uncertainly due to COVID-19. Our indices can be used to investigate Brexit-related uncertainties across different policy areas.
Keywords: Brexit; Uncertainty; Machine Learning
JEL Codes: D80; F50; E66
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Brexit uncertainty (F69) | investment (G31) |
| investment (G31) | productive capacity (E23) |
| Brexit uncertainty (F69) | delays in firm investment (D25) |
| COVID-19 (I15) | Brexit uncertainty (F69) |
| Brexit uncertainty indices (E32) | dynamics of uncertainty (D80) |
| Brexit uncertainty indices (E32) | strong correlations with established measures of uncertainty (D81) |