The event study methodology

Estimating the impact of an event on financial markets has been the subject of various researches. On the surface, this seems a hard task, but a measure can be constructed easily using the event study methodology. The usefulness of such a study comes from the efficient market hypothesis (EMH) which implies that any information must be directly incorporated in prices (MacKinlay, 1997; Harju and Hussain, 2011). Thus a test can be constructed by measuring of the event’s economic impact over a relatively short time period around the announcement. How the event study can be built? What are the problems faced when implementing the event study? To what extent MarketScience can use it?

 1-    Procedure for event study: In the literature, there are two ways to implement the event study. The first consists in regressing surprises on asset return or volatility around the announcement. This is a standard fashion and commonly used to test ex-post the significance of the impact of scheduled macroeconomic announcements. Bernanke and Kuttner (2005) among others use this method to estimate the average response of asset markets to monetary policy decisions. The second event study consists in first determining the period over which the security prices involved in this event will be examined-the event window, and second, estimating the significance of the difference between the real change around the announcement and the “theoretical” change, generally calculated using the market model or any other model that could reflect the “normal” return if there was no announcement. The difference between the real and the theoretical change is called the abnormal return and the sum of abnormal returns over the event window is called the cumulative abnormal return.

2-    Limits of the event study:  Though this methodology is easily implemented and it allows to test for the economic impact of news, especially the unanticipated one, it could be criticized for many reasons. The first problem is that the event window as well as the estimation window are randomly determined. Since market impact is changing over time, one must care about the window used to estimate the average response and must take into account business conditions. Second, the hard task before using the event study is to choose the right model for “normal” returns. Last but not least, one must care about the arrival of other news at the same interval, which may biases the results.

3-    Is it interesting for Beyond Market Noise? The event study is a way to test the impact of macroeconomic announcements and to determine if such movements around the announcements are fundamentally driven or speculative. But when we look at the disavantages of the methodology, one must care about the choice of the parameters (window, model…) and a combination with other indicators such as relevance, number of analysts or data mining must be necessary to corroborate the findings.