Estimating liquidity using information on the multivariate trading process

In this paper we model the dynamic multivariate density of discrete bid and ask quote changes and their associated depths. We account for the contemporaneous relationship between these trading marks by exploiting the concept of copula functions. Thereby we show how to model truncations of the multivariate density in an easy way. A Metropolized-Independence Sampler is applied to draw from the dynamic multivariate density. The samples drawn serve to construct the dynamic density function of the quote slope liquidity measure, which enables us to quantify time varying liquidity risk. We analyze the influence of the decimalization at the NYSE on liquidity.