Individual Effects Tests For Unbalanced Panel Error Components Models With Time Invariants

In the era of big data,more data can be collected and more information can be obtained which is more conducive for scholars to analyze and solve practical problems.Panel data includes two dimensions of individual and time which can better explore the role of each influencing factor and the relationship between each factor.So it is widely used in many fields,such as economy,biology,medicine and so on.Most of the theoretical and empirical studies on panel data are based on complete panel data sets.However,in the process of data collection,researchers often get incomplete panel data sets.Therefore,in order to solve more extensive problems,the incomplete panel data model is established and studied.This dissertation studies the test of individual effects in the incomplete panel error component model with time-invariant regressors.Firstly,the estimators of the coefficients of time-invariant explanatory variables are given in two different cases.When the time-invariant explanatory variables are exogenous variables,the FEF estimators are given.When the existence time-invariant explanatory variables are endogenous variables,the FEF-? estimators are given.The asymptotic properties of FEF and FEF-? estimators are also proved.Secondly,under different robustness levels,the moment estimation of the variance of random disturbance term is obtained,and the Hausman test idea is used to construct test statistics to test the existence of individual effects in the model.Under the null hypothesis,it can be proved that the test statistics constructed obeys the asymptotic normal distribution.In the simulation experiments,the test statistics obtained in this dissertation are compared with the common test statistics,and found that the statistics proposed in this dissertation are more robust and have good properties.In the last chapter,give an example analysis and draw the corresponding conclusion,in order to prove the feasibility of test method.