Incomplete Information, Trading Costs and Cross‐autocorrelations in Stock Returns

Publisher Website
Google Scholar
Abstract
This paper provides an economic rationale for the cross‐autocorrelation patterns in stock returns in the context of a microstructure model in which investors have incomplete information. The paper shows that in a market in which investors are informed about only a sub‐set of stocks, the emergence of lead‐lag, cross‐autocorrelations is a function of the cost of trading in other stocks based on information about the sub‐set of stocks. If cross‐trading costs are high, informed investors will trade only in the sub‐set of stocks they are informed about; if cross‐trading costs are moderate, informed investors will randomize between trading and not trading in other stocks; and if cross‐trading costs are low, they will trade in all stocks. When informed investors trade only in a sub‐set of stocks, prices of stocks with more informed trading will adjust to common factor information faster than the prices of stocks with less informed trading giving rise to asymmetric lead‐lag cross‐autocorrelations. When informed investors trade in all stocks, asymmetric lead‐lag cross‐autocorrelations will disappear as a result of their cross‐market arbitrage trading. These results provide a number of testable implications for lead‐lag cross‐autocorrelation patterns. The data is consistent with the empirical predictions. (J.E.L.G12, G14).

This publication has 21 references indexed in Scilit: