Persistence in Intertrade Durations

This paper examines long-term dependence in times between trades on financial markets. The autocorrelation functions of several intertrade duration series show a slow, hyperbolic rate of decay typical for long memory processes. For example, a shock to times between trades of the Alcatel stock on the Paris Stock Exchange (SBF Paris Bourse) may persist in the transactions time for a long period of 1,000 or 2,000 ticks. With an average duration of 52 seconds between transactions, this may amount to sixteen or thirty-two hours in calendar time. This paper introduces a fractionally integrated autoregressive conditional duration (FIACD) model for intertrade duration series. It also examines transformed duration processes representing times between consecutive returns to states of null, positive or negative returns. This approach captures the relationship between the duration persistence and return dynamics. The times elapsed between returns to various states feature very similar autocorrelation patterns and do not possess the long memory property. The persistence in durations is also determined by the times spent within specific states of returns. The average visiting time is state dependent, features intraday variation and may be considered as an instantaneous measure of state persistance. The long memory patterns are examined in data on the Alcatel and IBM stocks traded on the Paris Bourse and NYSE, respectively.

Cet article s'intéresse à la dépendance à long terme des durées entre transactions sur les marchés financiers. Les fonctions d'autocorrélation de telles durées présentent en effet des décroissances lentes typiques des processus à mémoire longue. Pour les traiter, nous introduisons des modèles de durée autorégressifs fractionnaires (FIACD). Ils sont appliqués à l'étude des durées entre les changements de sens d'évolution de prix (rendements positifs, négatifs ou nuls), et aux durées passées dans les divers états. Ces phénomènes de mémoire longue sont étudiés empiriquement sur des données des titres Alcaltel et IBM, échangés respectivement sur les bourses de Paris et de New York.