As part of the Brown Bag Lunch series, the International Economics Department at the Graduate Institute is pleased to invite you to a public talk given by Caterina Rho, PhD student in International Economics.

A sentiment-based risk indicator for the Mexican financial sector

Abstract: We apply sentiment analysis to Twitter messages in Spanish to build a Sentiment Risk Index for the financial sector in Mexico. Using a sample of tweets that covers the period 2008-2019, we classify the tweets considering whether they reflect a positive or negative shock on Mexican banks, or whether they are merely informative. We use a voting classifier based on three different classifiers: the first based on word polarities from a pre-defined dictionary, the second on a Support Vector Machine classifier and the third on Neural Networks. We  find that the voting  classifier outperforms the other classifiers when taken alone. We also compare this proposed Sentiment Risk Index with existing indicators of financial stress based on quantitative variables. We find that this novel index captures the effect of sources of financial stress that are not explicitly reported in quantitative risk measures, such as financial frauds, fails in payment systems and money laundering. We also show that a shock in the Twitter Sentiment Index is positively correlated with an increase in financial market risk, stock market volatility, sovereign risk and foreign exchange rate volatility.