Finance Seminar: Giang Nguyen

Joint with Brown Bag Seminar

A new Class of Empirical Dynamic Order Book Models with an Application to the U.S. Treasury Market

Giang Nguyen, PSU

We propose a new class of empirical dynamic order book models to study the joint dynamics of liquidity and volatility. The multiplicative error structure of the models permits a more flexible probability distribution and helps avoid the possibility of forecasting negative values for liquidity and volatility variables. We apply our models to the U.S. Treasury securities market and show that intraday liquidity and volatility are highly persistent, and that they are negatively interrelated at the inside bid and ask. However, beyond the first tier, this feedback relationship is not present: instead, market depth can predict subsequent price volatility but not vice versa. In addition, the dynamics of the limit order book change during the recent financial crisis. While the persistence of both depth and volatility does not change significantly, there is a composition shift toward lesser impact of depth shocks, but greater impact of volatility shocks. Our analysis also documents a sharp reduction in market liquidity concurrently with a surge in volatility at the height of the crisis in the fall of 2008.

(with R. Engle, M. Fleming, E. Ghysels)