Measuring Investors' Opinion Divergence

ABSTRACT Numerous proxies for divergence of investors' opinions have been suggested in the empirical accounting and finance literatures. I offer a new proxy constructed from proprietary limit order and market order data. This allows me to capture additional information on investors' private valuations. Proxies from the extant literature, based on publicly available data, do not contain such information. Given my new measure, I ask which of the extant proxies correlates best with it. In my regression analysis, unexplained volume is the best proxy for opinion divergence. Conditioning on various firm‐specific and order‐specific characteristics generally does not change this conclusion. The main exception is the sample of firms without IBES forecast dispersion data, for which bid‐ask spread is the best proxy for opinion divergence. Factor analysis also suggests that unexplained volume is the preferred proxy for opinion divergence.