A Regime-Level Empirical Model of the Specialist Quote Revision Process

NYSE trading is a continuous auction process distinguished by order flow imbalances and non-coincident revision of the bid and the ask. To deal with the aggregation problem presented by non-coincident revision of the quotes, we propose a regime-level empirical model and use it to test the Brock and Kleidon queuing theory of a continuous auction. Using transactions data for IBM for calendar year 1988, Harris, McInish, and Chakravarty (1995) performed a Hausman-type specification test that confirmed the exogeneity of order flow volumes at the bid and the ask. Extending their work, this paper estimates two simultaneous autoregressive ask and bid equations for 50 randomly selected stocks and contrasts thinly traded and volatile stocks. The results support Brock and Kleidon's distinguishing implication--namely, exogenous increases in trading volume raise the ask and lower the bid. Although some queuing system characteristics prove endogenous in thinly traded stocks and in especially volatile stocks, exogenous order flow volume continues to increase spreads across the 50 stock sample. The paper draws conclusions about the appropriate specification of endogeneity, cross-equation disturbances from the bid to the ask, and cross-equation queuing information flows. Copyright 2000 by Kluwer Academic Publishers