Business Publications
Document Type
Working Paper
Publication Date
11-2003
Abstract
Most financial markets allow investors to submit both limit and market orders but it is not always clear why agents choose one over the other. In this study we empirically investigate how several microstructure factors influence the choice and timing of submitting either limit or market orders using orders submitted to the Reuters 2000-2 system. Specifically we measure how these factors influence the expected time (duration) between successive orders using an autoregressive conditional duration (ACD) model. We find that the order submission process is not symmetric for market and limit orders. For example, we only find the lagged average volume, a proxy for market depth, affects the expected duration of market orders while the lagged market imbalance, quote intensity, average volume and bid-ask spread all shorten the expected duration of limit orders. We also find differences in the expected durations around the opening of different geographic markets on the expected duration, even after adjusting for the well-known time-of-day seasonalities. As a consequence our study provides empirical evidence in support of several market microstructure models of the order submission process.
