Persistence of announcement effects on the intraday volatility of stock returns: Evidence from individual data
Published date | 01 November 2017 |
DOI | http://doi.org/10.1016/j.rfe.2017.03.001 |
Date | 01 November 2017 |
Persistence of announcement effects on the intraday volatility of stock
returns: Evidence from individual data
Remzi Uctum
a,
⁎, Patricia Renou-Maissant
b
, Georges Prat
a,c
, Sylvie Lecarpentier-Moyal
d
a
EconomiX,CNRS/Universitéde Paris Ouest Nanterre laDéfense, 200 av. de la République,92001 Nanterre Cedex,France
b
CREM,CNRS/Université de Caen Normandie,UFR SEGGAT, Esplanadede la Paix, 14032 Caen Cedex, France
c
IPAG BusinessSchool, 184 bd Saint-Germain,75006 Paris, France
d
ERUDITE,Université Paris EstCréteil, Faculté de SciencesEconomiques et de Gestion,Mail des mèches, 61 av.du Général de Gaulle, 94010Créteil, France
abstractarticle info
Articlehistory:
Received13 May 2016
Receivedin revised form 18 January 2017
Accepted1 March 2017
Availableonline 7 March 2017
We analyze the persistence effects in the empirical relationship between announcement releases and return
volatilities of four majo r companies of the French stock market using hig h frequency data over the period
1995–1999.Besides its institutional stability,this sample period avoids the econometricdifficulties inherent to
simultaneous news arrivals.Our approach contributes to the relevant literature in that we focuson individual
stock volatilitiesrather than indices, we distinguishfirm-specific and macroeconomicannouncements, and we
endogenize both the durations of announcement effects and the response patterns of equity prices. We find
thatour individual volatilitiesare affected bya systematic marketeffect, calendareffects, announcementsrelated
to the firms’macroeconomicenvironmentand announcements relatedto the firms’and their competitors’stra-
tegic dealingsand commercial outcomes.We find evidence that all volatility responsesare gradual with persis-
tence horizonsranging from oneto three hours, revealinga significant degree of inefficiencyof the French stock
market over the period. This inefficiencycan be viewed as a breeding ground for the implementation of more
performantinformational and tradingsystems that allowed marketsto move towards more efficiency.
© 2017 Elsevier Inc. All rights reserved.
Keywords:
Intradayvolatility
High frequencymodelling
Persistenceof announcement effects
Firm-specificstock returns
1. Introduction
The expected future volatility of financialmarket returns is a main
element in assessing asset or portfolio ri sk and plays a key role in
derivativespricing models,trading and hedging strategiesand portfolio
allocation problems.As such, accurate measures and good forecasts of
volatility arenecessary for the implementation andevaluation of asset
pricing models. Thus, not surprisingly, much effort has been devoted
to modelingreturn volatility dynamics. In spite oftheir good in-sample
performance, standard volati lity models (GARCH-type or standard
volatility models) have not provento provide accurate volatility fore-
casts at higher than daily freque ncies.
1
However, Andersen and
Bollerslev (1998b)have documented how improved ex-post volatility
measures can be constructed when high frequency(intraday) data are
used.
Modeling high frequency financialdata requires consideringspecific
statistical properties, such as heteroscedasticity,leptokurticity, asym-
metry. Andersen and Bollerslev (1998a) provide a robust econometric
methodology for capturing the volatility components that contribute
to explaining thesevolatility patterns: the traditional ARCHeffect that
reflects the inter-day volatility, calendar effects that represent the intra-
day and intra-month structure of volatilityand the announcement ef-
fects that reflect the impact of public inf ormation on volatility.
Intraday effects are consistentwith the implications of microstructure
models (Goodhart and O'Hara, 1997; Andersen and Bollerslev, 1997,
O'Hara,2015). They involve periodicpatterns, such as marketopenings
and closingsor other significantphases withina day,which Andersen et
al. (2000)have shown that theyare adequately approximated by a Fou-
rier flexible form (Gallant,1981). As for the announcement effects,they
can be responsible of erratic jumps or irre gular patterns inherent to
asset return volatility. These announcement effects represent inves tors'
surprise to event shocks which they were n ot able to predict on the
basis of their available information.
2
Andersen and Bollerslev (1998a)
show that the calendar effect dominatesthe German Mark-Dollar spot
exchange rate volatility while the announcement effect is strong but
Reviewof Financial Economics 35 (2017)43–56
⁎Correspondingauthor.
E-mailaddresses: uctum@u-paris10.fr (R. Uctum), patricia.renou@unicaen.fr
(P. Renou-Maissant), prat@u-paris10.fr (G.Prat), sylvie.lecarpentier-moyal@u-pec.fr
(S. Lecarpentier-Moyal).
1
Foran overview on thesemodels see Bollerslevet al. (1992),Degiannakisand Xekalaki
(2004)or Engle, Focardi and Fabozzi(2008).
2
Sucha prediction failurecan be attributed to the factthat the cost of informationwas
higher than information yield in terms of a reduction in the forecast error, leading the
operatorto ignore a subset of (useful) information.An announcement effect is therefore
significant when the announced event was too costly to forecast. This interpretationis
consistent with the economically rational expe ctations theory introduced by Feige and
Pearce (1976),where the optimal information collected by the forecasterresults from a
cost-and-advantage analysisof information.
http://dx.doi.org/10.1016/j.rfe.2017.03.001
1058-3300/©2017 Elsevier Inc. All rightsreserved.
Contents listsavailable at ScienceDirect
Review of Financial Economics
journal homepage: www.elsevier.com/locate/rfe
To continue reading
Request your trial