Open Interest and Consensus Among Investors

Abstract

Recent studies (e.g., Verrecchia [1981]) have investigated conditions (if any) under which volume of security transactions can be used to infer consensus, or lack of consensus, of investors' beliefs. One conclusion is that volume cannot be used to infer the extent of consensus. This finding has motivated suggestions to examine other market variables (e.g., Morse [1981] and Morse and Stephan [1984]), which might be used for this purpose. Consistent with this suggestion, this paper reports the results of examining a new variable, open interest. Open interest is a market variable for assets with zero net supply (e.g., options and futures).1 It represents the total number of investors' long (or short) positions in one of these assets. Changes in open interest therefore reflect changes in the extent to which investors employ these assets in their portfolios. Open interest in an asset is clearly related to volume in that asset. Changes in open interest underlie changes in volume, in some sense. However, even casual observation suggests that there is no isomorphic (i.e., one-to-one) relation between the two. More formally, it is easy to show, using the triangle inequality, that volume is always at least as great as the change in open interest, and the relation need not be stable if investors can change sides in the market. (A proof will be provided upon request.) As a result, whatever information can be extracted from a scrutiny of open interest will be transmitted imperfectly to volume.