Statistics Project, Finance and Accounting event study.
Statistics Project, Finance and Accounting event study Project description Conduct an event study or test a trading strategy of your choice. Be creative! It is perfectly fine if the data does not support your hypothesis. This paper does not need to exceed 10 pages. When the first draft of the paper is due, you are expected to have selected an idea. Your idea needs to be approved first by me before you can proceed with data collection and empirical analysis. Some ideas for you to consider: 1. Within an industry, do returns of large-cap stocks lead those of small-cap stocks? (monthly data) 2. Do high volatility stocks have higher future returns? 3. Form portfolios based on some Compustat variable(s) and test for market efficiency. 4. Test the ˜Super Bowl Theory’ or some other such ˜event’. 5. Post-earnings announcement drift 6. Intermedium Momentum and long term reversal 7. Is there an illiquidity premium? (Do stocks with higher illiquidity earn higher returns?) The paper should follow the outline below A typical outline of an empirical paper: I. Introduction “ what is your paper about? II. Data Description “ generally describe your data, provide some summary statistics indicating number of stocks, time-period, distribution of key variables, etc. III. Empirical Tests “ describe the methodology and results of your empirical tests. Are they robust? V. Conclude “ pretty short summation Event Study An event study measures the impact of a specific event on the value of a security Event studies traditionally answer the question Does this event matter? (e.g. dividend initiation, index addition, stock splits, merger announcements, earnings announcement) But are increasingly used to answer the question: Is the relevant information impounded into prices quickly? (e.g. Post earnings announcement drift) What’s an Event Study? Short – horizon event studies examine a short time window, ranging from hours to weeks, surrounding the event in question. Since short run event expected returns are small in magnitude, this allows us to focus on the information being released and (largely) abstract from modeling expected returns. There is relatively little controversy about the methodology and statistical properties of short – horizon event studies. Short – horizon Event Studies Long – horizon event studies: Do IPO stocks under – perform in the three years after the IPO? Long – term reversal. Long – horizon event studies are problematic because the results are sensitive to the modeling assumption of expected returns. Long – horizon Event Studies Basic event study methodology is essentially unchanged since Ball and Brown (1968) and Fama et al. (1969). Abnormal return Anatomy of an event study: 1. Identify an event and the event window, security selection 2. Specification and estimation of the reference model, characterizing the œnormal returns (expected returns). 3. Computation and aggregation of œabnormal returns. 4. Hypothesis testing and interpretation. Basic Set up Let t index the event time. t = 0 is the time of the event. Event window: [T1+1, T2], [T0+1,T1] estimation window, [T2+1,T3] post – event window L1 = T1 – T0, L2 = T2 – T1, L3 =T3 – T2 Estimating the reference model using the estimation window. Compute normal returns in the event window and post – event window using the parameters estimated from the estimation window. Compute abnormal returns Analyze the abnormal returns: eg. test the significance Typical Time – line of an Event Study Common choices of reference models to characterize normal returns Constant expected return model ¢ ? ?? ?? ? ?? ?,? Market Model ¢ ? ?? ?? ? ?? ? ? ?,? ?? ?,? Linear factors models (e.g., three factors) ¢ ? ?? ?? ? ?? ?? ?,? ?? ?? ?,? ?? ?? ?,? ?? ?,? Short – horizon event study results tend to be insensitive to the choice of reference models. Reference Models Define abnormal return for the event window and post – event window. Suppose that we use the market model as the reference model, ? ? are estimated in the estimation window. The residual ? t ? t ? ? ? t for t in the event window and post – event window. Under H0: the event has no impact (on mean or volatility), then ? t ~N(0, ) Compute the abnormal return Abnormal returns need to be aggregated, typically through time and across securities (events) Aggregate through time for an individual security (event) the cumulative return from t 1 to t 2 is Under H0, Aggregate Abnormal Returns