(Derwall et al., 2005), but investors do not seem to pay attention to aggregated scores.
However, to our knowledge, there is no study on the relationship between CSR rating
announcement and trading volume.
This question is of great importance, since trading volume is one of the best metrics of
information content (Cready and Hurtt, 2002, show that volume metrics are more
powerful than return metrics to detect investor responses to an event). Thus, this paper
underlines the information content (or lack thereof) of CSR ratings. Moreover, trading
volume can be analysed in terms of stock market liquidity. If we follow Hong and
Kacperczyk (2009), a high CSR could improve stock market liquidity, since more
investors (including responsible ones) may invest in the ﬁrm, whereas low CSR would
deter responsible investors from choosing the ﬁrm.
In addition, we use new evidence on CSR from throughout Europe, whereas most
empirical studies on SRI performance are built on either US or UK samples with extra-
ﬁnancial information from KLD or Innovest (Renneboog et al., 2008a). In fact, the
European context of extra-ﬁnancial information from Vigeo is speciﬁc on this matter.
Month by month, Vigeo rates the ﬁrms of an ICB (Industry Classiﬁcation Benchmark)
industry following a calendar known in advance, while other agencies (such as KLD) rate
all ﬁrms annually. The mean time between two ratings in the same industry, and therefore
of the same ﬁrms, is about 18 months. The extra-ﬁnancial ratings are privately given to
subscribers, who can be either professional (institutional investors, asset managers) or
non-governmental organisations. Since dissemination is not wide (about 100
professionals are paying subscribers), we do not follow the standard approach of
considering this information as public (Kempf and Osthoff, 2007). Indeed, we actually
take into account the true nature of this event: a private, costly signal. Consequently, we
face two speciﬁc aspects of market reaction to information disclosure: ﬁrst, diffusion of
extra-ﬁnancial information versus the usual ﬁnancial one; second, reinforcement of
information asymmetry among investors, due to the private signal on CSR rating.
Further, Vigeo ratings contain a high degree of precision, with several aggregation
levels, and they follow a metric scale. Thus, we can study the effects, on investor
behaviour, of both aggregation level and score value.
The monthly frequency of Vige o ratings enables us to use an event study to measure
Abnormal Volume (AV) associated w ith announcements. This choice is motiv ated by
several arguments. As for event stu dies on stock returns, we use this methodolo gy to
identify any information co ntent of an announcement. An event st udy on trading
volume could reveal how investors change their beliefs at the time of announcement.
Moreover, the event-study me thodology, generally used to show t he information
content of an announcement, can also be considered a causality test (see for exam ple
Morck and Yeung, 2011). Th is methodological choice has two beneﬁts . First, the
event study yields evidence o f causality from CSR to stock market, controlling for
endogeneity. Second, it is ve ry unlikely that volume suffers fr om a reverse causality
problem. Thus, this study yields clear conclusions on the information contribution of
CSR ratings to the ﬁnancial markets.
In the absence of any study on the relationship between CSR and trading volume, we
consider several arguments that suggest that CSR rating announcements induce a change
in trading volume, mainly based on informational motives. First, rating announcements
can mitigate incomplete information (Merton, 1987). In this framework, spreading (even
already known) information across the market induces a discount rate reduction, due to
less incomplete information, and in turn increases ﬁrm value and trading volume.
© 2015 John Wiley & Sons, Ltd.
486 Alexis Cellier, Pierre Chollet and Jean-FranScois Gajewski