Academic journal article Academy of Accounting and Financial Studies Journal

Investor Sentiment as Intervention of Stock Market Returns

Academic journal article Academy of Accounting and Financial Studies Journal

Investor Sentiment as Intervention of Stock Market Returns

Article excerpt


The possible effect of investor sentiment on equity returns remains one of the yet to be resolved issues in financial market analysis. As opposed to the conventional noise trader theoretical framework, the sentiment formation process has been approached from an unique event perspective by several recent studies such as Garner (2002), and Burch, Emery, and Fuerst (2003), and the event of September 11, 2001 has been used as a natural test of this hypothesis. Using intervention analysis, our study expands on this line of inquiry by statistically relating the formation of sentiment under this unique event scenario to its impacts on stock market returns.

The noise trader models as developed by De Long, Shleifer, Summers, and Waldmann (1990) contend that uninformed traders acting in concert of non-fundamental signals can introduce a systematic risk that will be priced by the market. To the extent that noisy traders based their trading decisions on investor sentiment, such measures will have additional predictive power beyond conventional fundamental factors. As an empirical matter, the statistical evidence so far are mixed at best. Studies such as Neal and Wheatley (1998), Wang (2001), Simon and Wiggin (2000), Lee, Jiang, and Indro (2002), Wang, Keswani, and Taylor (2003), Fisher and Statman (2003), Brown and Cliff (2005), Tetlock (2007), and Baker and Wurgler (2007), using various proxies for sentiment such as closed end fund discounts and directly observed consumer/investor surveys, find evidence that seem to suggest that some of the sentiment measures indeed predict equity returns, However, although the Brown and Cliff study attempted to control for the influences of fundamental factors explicitly, it still remains unclear in general whether the sentiment measures reflect additional independent explanatory power of their own.

The events of September 11, 2001 were sudden and unexpected, and represent a totally exogenous shock to the world financial markets, independent of any market fundamentals. Using September 11 as an unique event analysis, Burch, Emery, and Fuerst (2003) find that in the aftermath of the events small-investor sentiment shifted significantly into the negative immediately. It did not rebound to its pre-attack level until well after the market recovered. In the same vein, Garner (2002) suggests that sentiment indexes which normally contain relatively little information that is not available from other indicators might, however, under special circumstances such as this one, contain unique information that is not readily apparent from other sources. His results, however, show consumer sentiment indexes in this particular instance maintained a fairly normal relationship to other economic variables subsequent to the attack, and as such, did not contain much relevant new information. This line of reasoning is also consistent with the recent behavioral finance literature which places the emphasis on the effect of investor mood on asset prices. In their recent study, Edmans, Garcia and Norli (2007) use sport event outcomes to identify the sudden changes in the moods of investors and their impacts on market performance. Their results show strong negative stock market reaction to losses by national soccer teams. In this study, we take it as given that September 11 was an unique event that may contain relevant information additional to conventional market fundaments. Using interrupted time series analysis and ANCOVA, we test for the change in investor sentiment as a result of the event and its impact on stock returns.


Denoting the time series full impact intervention model as

[Y.sub.t] = f([I.sub.t]) + [N.sub.t]

where [Y.sub.t] represents the time series of interest, [I.sub.t] the intervention component or the transfer function, and [N.sub.t] the noise component of the stochastic process. The intervention function, [I. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.