This section describes the conceptual framework of the methodology, the hypothesis of the paper, the data, the estimation of SAR, the reliability of SAR estimates and testing for the heteroskedasticity of political structural break.
Conceptual framework and hypothesis of the paper
Certain studies in the literature conclude that the Egyptian stock market is a weak-form efficient [4, 5]. Nevertheless, Mlambo and Biekpe [44] claim that there are no rules that can be used in detecting stock prices so that they can be used in predicting future prices in the Egyptian stock market. In this regard, the event study methodology is quite relevant as a further examination of that issue. Therefore, the framework of this paper can be illustrated as follows.
Hypothesis
This paper tests the hypothesis that follows:
“The structural political break in Egypt on 23rd of June 2013 has a significant effect on stocks abnormal returns.”
Data
The data are obtained from Egypt stock exchange for the stocks listed in the index (EGX30). The data cover the period June 3, 2012– March 4, 2014. Although the literature does not specify certain time interval in event studies [33], the authors consider 25 days before the event (estimation period) and 25 days after the event (event period). The event day is excluded.
Estimation of standardized abnormal returns in event study methodology
Event study methodology offers a way for avoiding the bias of accounting-based measures of returns [43]. This methodology is well cited in the literature for variety of applications [12, 26, 29, 34, 37, 46, 48, 49]. The aim of an event study is to quantify and statistically test the presence of any abnormal or excess returns that are associated with certain events. The abnormal or excess returns are calculated as the difference between the observed returns and expected returns which is the expected return in case of non-occurrence of the event using a specific return calculation model [46].
The literature includes three main methods to estimate the expected return. There methods are as follows [16, 46]:
-
1.
Market model (MM).
-
2.
Mean-adjusted return model (MAR).
-
3.
Market-adjusted or index model (IM).
Estimation of standardized abnormal returns
In this paper, the event time is identified as the speech of the Egyptian Minister of defense on June 23, 2013. The expected returns of listed firms in the index (EGX30) before and after the event are estimated using the three above-mentioned methods that follow.
-
A.
Market model [29].
$$R_{it} = \alpha_{i} + \beta_{i} R_{mt} + \varepsilon_{it}$$
(1)
where \({\text{R}}_{\text{it}}\) is return on security i for period t (observed return), \(\alpha_{i}\) is constant intercept, \(\beta_{i}\) is slope coefficient, \(R_{mt}\) is return on market index m for period t, \(\varepsilon_{it}\) is error term (abnormal return) for security i in period t. In this case, the observed minus predicted return equal abnormal return, which is the disturbance term.
-
B.
Mean-adjusted return model [14, 15].
$$R_{it} = \mu_{i} + \varepsilon_{it}$$
(2)
where \({\text{R}}_{\text{it}}\) is return on security i for period t (observed return), \(\mu_{i}\) is the mean return on security i over t periods within estimation period normal (expected or predicted) return, \(\varepsilon_{it}\) is disturbance term (assumed to have mean = zero, and SD = 1). In this case, the observed minus mean return equal abnormal return, which is the disturbance term.
-
C.
Market-adjusted or index model [14, 15].
$$R_{it} = R_{mt} + \varepsilon_{it}$$
(3)
where \({\text{R}}_{\text{it}}\) is return on security i for period t (observed return), \(R_{mt}\) is return on market index m in period t, \(\varepsilon_{it}\) is disturbance term (white noise) assumed to have mean = zero, and SD = 1. Each period’s observation in white noise time series is a complete surprise. In this case, the observed minus index returns equal abnormal return (or disturbance term).
The authors in this paper compute the standardizing abnormal returns (SAR) to ensure the robustness of abnormal returns estimation [24, 41]. The standard error of the estimate is calculated as follows [46], taking into consideration that this equation applies to the market model and the market-adjusted model only.
$$S_{ie} = \sqrt {\frac{{\sum\nolimits_{j = 1}^{T} {(R_{ij} - R_{ij}^{*} )^{2} } }}{T - 2}}$$
(4)
The standardized abnormal return (SAR) for security i in period t is calculated as follows:
$$SAR_{it} = \frac{{AR_{it} }}{{S_{it} }}$$
(5)
$$SAR_{it} = \frac{{AR_{it} }}{{S_{i} }}$$
(6)
In the mean-adjusted model, the standardized abnormal return for security i in period t is calculated as follows:
$$S_{i} = \frac{1}{\sqrt T }\sqrt {\sum\limits_{j = 1}^{T} {R_{ij}^{2} - \left[ {\frac{{\sum\nolimits_{j = 1}^{T} {R_{ij} } }}{T}} \right]}^{2} }$$
(7)
The standard error of the estimate that can be used in the case of mean-adjusted returns is as follows [46].