英语论文网

the following manner.
1. T test is used to review if the means of the estimated and real values are equal.
2. Wilcoxon signed-ranks test is used to review if the medians of the estimated and real values are equal.
Also, as a result of using panel data method, tests special to this analysis method is applied which will be explained later.
Zaranejad and Anvari (2005) state that panel data are referred to a set of data based on which the observations are studied by a large number of cross-section variables (N), mostly selected by random, during the specified periods of time (T). In this case, N*T statistical data are called panel data.
Since the panel data include both approaches of time-series and cross sectional data, employing appropriate statistical explaining models to describe the features of those variables is more complex than models used in time-series or cross sectional data. In recent years, the panel data method has been used in many applied researches. The advantage of using this method is increased statistical power of coefficients compared to separated analysis of statistical data as cross-section or time-series. In this method, with a common consideration on the changes of variables in any section and at any time, all available data are used and, therefore, the error of observations becomes lower. Although the range of statistics is wide during the analyzing of statistical cross-sectional data, much more information is used to survey panel data. As a result, with an increase in the range of statistics and information, the degree of freedom is increased and the estimations made on statistical population will have much more efficiency (as quoted by Gujarati, 2003). The other advantage is that using this method would lead to
the observation of some effects not being simply seen in time-series and cross-sectional data.
Therefore, panel data are better and more suitable for the dynamic study of changes.
Of course, it cannot be said that there is no problem while modeling with panel data. They state
that there are different methods to test panel data and special tests to select each of these methods.
These methods are Pooled Data Method, Fixed Effect Method, Random Effect Method, and SUR
Method; the selection between Random Effect Method and Fixed Effect Method is made according to
Hausman Test, the selection between Pooled Data Method and Fixed Effect Method according to
Chow Test, and the selection between Pooled Data Method and Random Effect Method according to
Beusch-Pagan LM Test, respectively.

67 International Research Journal of Finance and Economics - Issue 36 (2010)
According to Zaranejad and Anvari (2005), the first condition to use a Random Effect Model is that the variables should have been selected randomly. Thus, since in this research the variables have not been selected randomly from different variables this method can be of no use. The Fixed Effect Model also incorporates the effects of other variables in the intercept of regression equation so that the relevant regression equation should have intercept, and, since the forecasting model under test in present research, according to Ohlson (1995) and Feltham-Ohlson (1995), lacks intercept, this method cannot be applied as well. Moreover, SUR Method may be used in cases where the length of timeseries is more than the number of cross sections and as 本论文由英语论文网提供整理，提供论文代写，英语论文代写，代写论文，代写英语论文，代写留学生论文，代写英文论文，留学生论文代写相关核心关键词搜索。