Emotional news : how emotional content of news and financial markets are related

Abstract

We present here a first step towards developing a quantitative model that relates investor emotions to financial markets. We used Wall Street Journal articles as a proxy of investor emotions on a "macro" level. We measured the emotional characteristic of the article texts quantitatively through content analysis to arrive at a daily set of emotional and subject category scores. After establishing the statistical and informational validity of these scores, we ran correlations and regressions between the daily category scores and broad market indices variables such as return, volume, and volatility to determine whether there is a relationship. We found that negative emotions are more strongly correlated with market variables than positive emotions. We also found that markets are a better predictor of emotions than emotions of markets. There also appears to be a stronger relationship between emotions and market volatility than with market returns. In investigating the source of the correlations, we found that the most extreme category scores are responsible for driving the bulk of the correlations. Event study results suggest that there is a stronger relationship between negative events and negative emotions than between positive events and positive emotions. A challenge we encountered that remains to be fully addressed is how to integrate our interpretation of the analysis results into our understanding of the link between emotions and financial markets from a causal and psychological perspective.