Investor sentiment and stock returns : using newspaper and Google search sentiments to predict returns on Oslo stock exchange

Abstract

We hypothesize that current sentiments can predict future stock returns, and we construct sentiment indexes based on Norwegian newspapers and Norwegian Google searches respectively. The indexes measure changes in the occurrence of economic terms with a negative sentiment, like refinancing, recession and fraud. We investigate if the indexes predict future returns on Oslo Stock Exchange. Our first finding is that an increase in a weekly newspaper index predicts negative return two weeks later. A one standard deviation increase in the index is associated with 0.4% lower return for the 10% largest stocks. The effect is only apparent for large stocks. Our finding suggests that the index explains 0.6% of returns in week two. The second finding is that an increase in a monthly Google search index predicts positive return the next month. A one standard deviation increase in the index is associated with a 1.2% higher return the subsequent month. The effect is strongest for large stocks. Our finding suggests that the index explains 5.9% of next month’s return.