New nonlinear unit root tests based on fractional frequency Fourier functions


İmre Bıyıklı S., Hepsağ A.

Communications in Statistics - Simulation and Computation, cilt.54, ss.1-22, 2025 (SCI-Expanded)

Özet

The study develops a nonlinear unit root test based on fractional frequency Fourier functions, taking into account the Fourier-Sollis (FSOLLIS) test developed by Omay et al. and the Fourier-Kruse (FKRUSE) unit root test developed by Güriş. Through both Monte Carlo simulations and empirical applications, the developed tests were found to be more powerful than alternative tests. In this study, where nonlinear unit root tests based on fractional frequency Fourier functions are developed, two different tests have been introduced: the Fractional Frequency Fourier-Kruse (FFKRUSE) and the Fractional Frequency Fourier-Sollis (FFSOLLIS) unit root tests. The FKRUSE and FSOLLIS tests have been compared in terms of their size and power characteristics. As a result, the fractional frequency unit root tests were found to be more powerful than alternative tests in most cases. It was also found that the new tests do not suffer from size bias. Finally, the validity of Purchasing Power Parity (PPP) theories for 29 middle-income countries has been tested using both the unit root tests developed in this study and alternative tests. The empirical results show that the fractional frequency unit root tests provide stronger results than the alternatives.