A new panel unit root test based on Bootstrap approach

Most econometric analysis methods are based on the assumption that variables are stationary. The use of non-stationary variables in econometric analyses may lead to spurious relationships. For this reason, it is essential to use unit root tests to determine whether the data are stationary before performing econometric analyses. This thesis proposes a new panel unit root test considering structural breaks and cross-sectional dependence. The nature of structural breaks and cross-sectional dependence is not always known. Therefore, we use Fourier functions for structural breaks and moving block bootstrap for cross-sectional dependence. Monte Carlo simulations investigate the size and power properties of this new panel unit root test considering different data generation processes. According to the simulation results, it is observed that the size of the test is close to the nominal level for data where errors are not autocorrelated. It is concluded that the power of the test approaches one with the increase in the number of observations, and the test is consistent. For the data where the errors are autocorrelated, the test shows size distortions, but the size distortions decrease as the time dimension and the number of unobserved factors increase. In the empirical part of the study, the newly developed panel unit root test is used to investigate whether the income convergence hypothesis is valid for 23 OECD countries between 1960 and 2022. We also present evidence from different panel unit root tests to compare the results. The results show that the income convergence hypothesis is invalid in OECD countries.