Estimating the Adverse Selection Cost in Markets with Multiple Informed Traders

We investigate the relation between the number of informed traders in a financial asset and the estimated adverse selection cost of trading in that asset, lambda, after controlling for the effects of previously identified determinants of market liquidity. As a proxy for informed traders, we use dual traders in futures markets - i.e., floor traders who trade both for customers and their own accounts on the same day. We show, theoretically, that it is optimal for dual traders to mimic both the size and direction of their informed customers' orders. Using data from four selected futures contracts we show that the number of dual traders in a contract is indeed a significant determinant of the adverse selection cost of trading in that contract. We also examine how the adverse selection cost of trading changes with the number of competing dual traders, m, in an asset. Our model demonstrates the existence of a non-monotonic relationship between lambda and m. Specifically, for securities with relatively small numbers of dual traders, there is a positive relationship between lambda and m. For securities with relatively large number of dual traders, there is a negative relationship between lambda and m. Our data indicates a strong support for the non-monotonic relationship between lambda and m. A practical implication of our results is that an observable variable such as the number of broker-dealers in the market can provide an accurate estimation of the adverse selection cost of trading in that market