The Analysis Of Panel Data Model In Econometrics

A panel, or longitudinal, data set is one that follows a given sample of individuals over time, and thus provides multiple observations on each individual in the sample. In the classic econometrics the sample observations are cross sectional data or time series data. As the development of theory and method in econometrics and its more and more extensive application we should make use of both cross sectional data and time series data. For example, when we analysis the configurable unemployment every regional industry structure will impact this problem, at the same time our national macro policy in every period can impact it. If one only imposes cross sectional data, namely the same time series data but the different region as the sample observations, one can analyze the structural unemployment's effect because of the different industry structure in various regions, however, we can't analyze the structural unemployment's effect because of the national macro policies. If one only utilizes the time series data, namely the different regional data in the different time as the sample observations, one can analyze the structural unemployment's effect because of the national macro policies, but can't analyze the problem because of the different industry structure. However, if we make use of panel data, namely the different regional and time's data as the observations, we can analyze the structural unemployment's effect because of both the different industry structure in various regions and the national macro policies. From the above we can learn that the practical economical problem, which is resolved by utilizing panel data, is very prevalent.In the text we mainly introduce the character of panel data, the simple linear statistic model and dynamic panel data model are considered, we give the estimations of the parameters in these models. In the linear panel data model we consider the parameter property in the incomplete information panel data model with two-way error and give the test statistic of the individual and time effect. Finally we prove that these test statistics are normal approximation and we do a Monte Carlo experiment to prove these results. As to dynamic model we also receive the consistent estimation of the parameters and have an application with our national traffic passenger transport characters. Moreover, we model a kind non-stationary panel data and get the consistent estimators. At last the errors-in- variables panel data models are considered, we obtain consistent estimates through the general method of moment and extend this method to the dynamic panel data model with error of measurement, the moment conditions are obtained, we can obtain the consistent GMM estimators through these moment conditions.