The Logarithmic ACD Model: An Application to Market Microstructure and NASDAQ.

This paper introduces the logarithmic autoregressive conditional duration model (log-ACD model). The logarithmic version allows for more flexibilitythan the ACD model of Engel and Russel (1995), when additional variables are included in the model. We apply the log-ACD model to bid/ask prices relative to securities listed in the NASDAQ and investigate the way market makers revise their beliefs on these bid and ask prices.The updating beahviour of the market makers is closely linked to the evolution of the liquidity of the market over the trading day.