Persistence in Intertrade Durations

This paper examines long memory in durations between trades of stocks on financial markets. The approach is based on semi-parametric methods as well as on the fractionally integrated autoregressive conditional duration model (FIACD). The persistence is also investigated in endogenously aggregated duration processes, representing times between consecutive returns to the states of null, positive or negative returns on stocks. This analysis captures the relationship between the duration persistence and stock return dynamics in varying market regimes. The paper also explores the sojourn times measuring durations of various states of stock returns.The long memory patterns are examined in data on the Alcatel and IBM stocks traded on the SBF Paris Bourse and NYSE