9. Closing remarks
In this article, I introduce a novel approach for quantifying a firm’s flow of information and use the cross-firm similarity of this measure to predict future price comovement. Commonality in information flows is gauged by the textual similarity of firm-specific content appearing on the Reuters Integrated Data Network from 2003 to 2013. This measure of qualitative similarity predicts an economically meaningful portion of future return correlation after controlling for numerous alternative explanations of comovement that have been suggested in prior literature. Previous research shows that newswire text is informative about future stock returns. My paper is the first to show that this type of qualitative information also predicts price comovement.
The time series of a firm’s stock returns are the singledimensional output of a pricing function containing a broad range of inputs. Because this function evolves over time, the influence of inputs relevant to future prices may not be present in distant historical return series. The newswire text written about a firm describes these relevant inputs. Both in research and in practice, estimating the market correlation structure has sensibly relied on a lengthy historical times series. The depth of this qualitative information can amend the shortcomings of using distant historical prices for predicting comovement. Quantifying these inputs provides an opportunity to predict only the comovement that is implied by the contemporaneous pricing function. Thus, the approach introduced in my paper can produce estimates of future return correlation that do not require, or significantly benefit from, an abundant individual price history.
