How to Conduct Long-Run Event Studies Essay - 2750 Words

How to Conduct Long-Run Event Studies (Dissertation Review Sample) Content: Literature Review on how to conduct long-run event studies (examining performance more than 12 months following an event)NameCourseProfessorà ¢Ã¢â€š ¬s nameInstitutionHow to conduct long-run event studiesEvent studies are an important element that gives a helpful verification on the response of the stock prices to the information (Dutta Pynnonen, 2002). According to Gur-Gershgoren, Zender, Hughson (2004), the majority of the studies have focused on short window returns usually of few days on a plainly dated event. The importance of short window approach is that since the daily returns expected are almost zero, the expected return model doesnà ¢Ã¢â€š ¬t have a huge consequence on inferences regarding the abnormal returns. Dutta Pynnonen (2002) argues that the short return window studiesà ¢Ã¢â€š ¬ supposition lags because price response to a particular event is short-term. According to Gur-Gershgoren, Zender, Hughson (2004), this assumption has been challenged by several pieces of literature on the basis that there is a slow adjustment of stock prices to information. Therefore, there is a need for the expected returns to be examined over long horizons of more than 12 months following an event to have a full understanding of the inefficiency of the market. According to Dutta Pynnonen (2002), also, several articles in the finance literature have indicated that there are abnormal returns earnings by the firms over a long period (more than 12 months following an event). Gur-Gershgoren, Zender, Hughson (2004) argues that usually, there is a contradiction between the evidence of long-horizon abnormal returns and the hypothesis of the efficient market that there is an adjustment of the full stock information within a short window. It is important to note that long horizon eventsà ¢Ã¢â€š ¬ statistical inferences depend on the methodological choice (Atkinson, 2009). Thus, it is important to have a good understanding of the limitations as well as the properties of the approaches available before making a decision on the methodology to be used in the long-horizon event studies (Atkinson, 2009).According to Atkinson (2009), there are two tasks at the center of long-horizon event studies. The first task involves the measurement of the event-interrelated horizon abnormal returns whereas the second task involves null hypothesis testing that this long horizon abnormal returns distribution focuses around zero. Atkinson (2009) further argues that an efficient testing procedure regarding the long-horizon event studies involves appropriateness of the two tasks. Or else, there is a possibility of the occurrence of two types of error as well as a result of inferences that are inaccurate. According to Carlson, Fisher, Giammarino (2000), the occurrence of the first error is based on the rejection of the null hypothesis, not because true abnormal returns have been generated by an event, but rather there is an application of a prejudiced b enchmark in the measurement of the abnormal returns. There is shifting of the focus of the abnormal returns by the prejudiced benchmark away from zero as well as results to numerous null hypothesesà ¢Ã¢â€š ¬ false rejection. According to Atkinson (2009), the occurrence of the second error is based on the acceptance of the null hypothesis, not because an event does not have an impact, but rather the test doesnà ¢Ã¢â€š ¬t have sufficient power to discriminate statistically the mean abnormal return to zero. Thus, there is undesirability regarding the test with low power as it makes researchers arrive at a false conclusion that there is a statistical insignificance regarding the long-term effect. Therefore, the researchers require a procedure that reduces the error sources, or at least, select a balance between the two errors (Zender Hughson, 2004)In the recent literature regarding the measurement as well as testing of the long-term abnormal returns, there has been the application of two approaches (Carlson, Fisher, Giammarino, 2000). In the first approach, benchmark is used in the measurement of the abnormal return of buy-and-hold for each firm in a sample as well as testing is done to determine if there is a zero mean to the abnormal returns. According to Atkinson (2009), a portfolio to every calendar month is formed by the second approach comprising of the firms that an event has occurred in a particular time period preceding the month, as well the null hypothesis is tested that there is zero intercept in the regression regarding the portfolio income of the monthly time calendar against replica asset-pricing factors.In application of either the approaches, there is a need for the researchers to make choices on the appropriate option. Regarding the approach to the calendar-time portfolio, to fit the model, researchers have to make a choice on the technique of estimation as well as the model of pricing the assets (Bell, Brooks, Prokopczuk, 2013). The commonl y used models of pricing the assets are the three-factor model and its extension of four factors including additional factors connected to the momentum. Two systems are applied primarily to aid in the fitting of the price model i.e. the technique of the weighted least squares (WLS) as well as the techniques of the ordinary least square (OLS) (Bell, Brooks, Prokopczuk, 2013). Contrary, on the application of the approach of the buy-and-hold benchmark, researchers can make a choice on the portfolio reference or a single control firm as the abnormal returnsà ¢Ã¢â€š ¬ measurement benchmark as well as choose either parametric or nonparametric statistics regarding null hypothesis testing of the zero abnormal return (Dutta Pynnonen, 2002)According to Wooldridge (2009), a large number of potential procedures of testing can be generated by the permutation of the both the approaches under theses choices applied in the study of long-horizon events. Atkinson (2009) reiterates that in a pragma tic study of a financial event, it is neither sensible nor reasonable to employ all the procedures of testing. Thus, it is important to give direction regarding the proceduresà ¢Ã¢â€š ¬ strengths as well as weaknesses based on the outcomes of simulation. Zender Hughson (2004) says that a large numbers of replications are generated by the study of simulation under diverse conditions for each procedure of testing. This enables the tabulation of the types of errors for the comparison purposes.Critical issues in the Long-horizon event studiesThe approach of buy-and-hold benchmarkAccording to MacDonald Murphy, 2002), numerous studies indicate that there is the sensitivity of the long-term abnormal returns to the benchmark choices. The application of an incorrect benchmark in the measurement of the long-term abnormal returns, there would be errors regarding the conclusion on the particular eventsà ¢Ã¢â€š ¬ significance. Atkinson (2009) argues that most of the existing investigations a re based on the lone matched firm or a matched portfolio of reference as the benchmark. New listing prejudices are eradicated by the approach of the control firm. Also, the rebalancing prejudice, as well as the setback of skewness, is also eliminated by this approach. Furthermore, in all the conditions considered, it yields well-particular test statistics. According to Dionysiou (2012), there is an advocating of a firmà ¢Ã¢â€š ¬s reference portfolio matching on size as well as the BE/ME. Practically, the choice aspect regarding the benchmark remains unsolved. Another critical problem connected with the event not being representative in significant facets of the relevant harmonized portfolio in the approach of reference portfolio is overcomed by the method of control firm (Zender Hughson, 2004). This result to a portfolio return that is matched creating a prejudiced approximation in regard to the firmà ¢Ã¢â€š ¬s expected return. This predicament is predominantly stern with small organizations.According to Zender Hughson (2004), using a benchmark matching the BE/ME and event firm on size is a universal practice in the computation of long-term abnormal return of an event firm. This practice is frequently warranted through the understanding that there is a combination of the size as well as the BE/ME to aid in the capture of the cross-sectional deviation in a monthly average stock return and that there is no additional power of market beta in providing an explanation regarding the diversity of the cross-sectional difference. Dutta McMillan (2015) argues that there are aspects connected with the expected monthly stock return i.e. aspects connected to a book-to-market ratio (BE/ME), aspects connected to size as well as market aspects. In solving this problem, beta-based matching in addition to BE/ME as well as the size doesnà ¢Ã¢â€š ¬t improve the approachà ¢Ã¢â€š ¬s performance. Zender Hughson (2004) reiterates that a recent trend involves the application of the tests based on the computation-rigorous bootstrapping, including the skewed-accustomed t-statistics of bootstrapped Johnson as well as the p-values of the replicated empirical. According to Dutta McMillan (2015), the actions depend on the repetitive random sampling to aid in the measurement of the relevant test statisticsà ¢Ã¢â€š ¬ significance. Because of the random sampling nature, there is variation regarding the significance of the resultant measurement every time this procedure is applied. Therefore, diverse researchers came up with conflicting conclusions applying the same procedure in the on the event firmsà ¢Ã¢â€š ¬ equivalent sample. Zender Hughson (2004) argues that contrary, the test of the simple nonparametric tests including the test of Wilcoxon signed-rank is free from the variation based on the random sampling. On a large scale simul...