Essays on estimation and testing in nonstationary panel data

Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)




Chihwa Kao


Estimation, Testing, Nonstationary panel data

Subject Categories



In the first chapter, the limiting distributions for ordinary least squares, fixed effects, first difference, and generalized least squares estimators are studied. A simple linear time trend with a one-way error component model is considered. It is shown that when the disturbances are stationary, the fixed effects estimator is asymptotically equivalent to the generalized least squares estimator. However, when the disturbances are nonstationary, the first difference estimator is asymptotically equivalent to the generalized least squares estimator. The asymptotic distributions for the estimators are also presented for the case when the disturbances are nearly nonstationary. Monte Carlo simulations are employed to compare the performance of the four estimators in finite samples.

In the second chapter, two classes of test statistics for detecting a parameter break at an unknown date are proposed. The model is similar to the one considered in chapter one. The first test is based on the fluctuation test of Ploberger-Kramer-Kontrus, extended to panel data. The second test is based on the mean and exponential Wald statistics of Andrews and Ploberger and the maximum Wald statistic of Andrews, extended to panel data. The limiting distributions of the proposed tests are derived when the disturbances are stationary, nonstationary, and nearly nonstationary. These limiting distributions are then used to simulate the critical values for the tests. Monte Carlo simulations are performed to examine the size and power of the proposed tests in finite samples.

In the third chapter, the following panels of countries are considered: (1) panels based on individual growth rates, (2) OPEC countries, (3) industrialized countries, (4) G7 countries, and (5) panels based on geographic location. The theory from chapters one and two is used to estimate and test for structural change in the annual growth rates of real per capita GDP, real per capita consumption, and productivity in each of the different panels. It is found that industrial countries experienced slowdowns in growth in the early 1970s, whereas less developed countries also experienced slowdowns in growth, but the timing of the slowdowns is in the mid to late 1970s.


Surface provides description only. Full text is available to ProQuest subscribers. Ask your Librarian for assistance.