Market Making with Costly Monitoring: An Analysis of the SOES Controversy

We develop a model of price formation in a dealership market where monitoring of the information flow requires costly effort. The result is imperfect monitoring, which creates profit opportunities for speculators who pick off "stale quotes". Externalities associated with monitoring give rise to multiple equilibria in which dealers earn strictly positive expected profits. We obtain various policy implications. A switch to automatic execution can improve or worsen spreads and price discovery depending on the specific equilibrium. A reduction in the minimum quoted depth tightens the spread but it reduces price efficiency. Our analysis is relevant for the SOES controversy given that speculators in our model behave as the real world SOES "bandits". Our model predicts that SOES bandits should trade in stocks with small spreads and that SOES bandit activity should widen the spread. We provide empirical evidence consistent with these predictions.