Inferring Information from Trading

Abstract

Most theoretical as well as empirical models in market microstructure model information flow and insider trading for one stock at a time. Information consists of a market-wide, an industry-specific, and a firm-specific component. An as yet unexplored implication is that information for one stock may have consequences for other stocks as well. This suggests that market makers may be able to infer information for a specific stock not only from observing the order flow in that stock, but also from observing order flows in other stocks, particularly other stocks within the same industry. This paper models information arrival and the resulting order flows by developing a two-stock sequential trade model based on the one-stock model in Easley, Kiefer, O'Hara, and Paperman (1996). The paper uses a sample of NYSE listed S&P 500 stocks and estimates the model for stock pairs in the same industry and for control stock pairs in different industries. The control pairs consist of stocks in different industries, which are matched on market capitalization and average daily turnover. Using the information in trade data, the model determines the frequency of information events relevant to one or both of the stocks. The analysis shows that the probability of an information event relevant to both stocks is significantly higher for stock pairs in the same industry than for matched stock pairs in different industries. This suggests that market makers may not only infer information from the order flow in their assigned stock, but also from the order flow in other stocks within the same industry.