Doğrusal Olmayan Panel Veri Modellerinde Eşbütünleşme Testleri

Abstract

Nonlinear cointegration tests are proposed in nonlinear panel data framework. Cross section dependency problem that is faced in the long (in terms of time series data) panel data is achieved via implementing sieve bootstrap methodology. The tests developed in the thesis are used in testing the validity of Fisher effect in OECD countries. Countries are divided into two groups whose are named with G7 and Developed countries. Panel cointegration tests are applied for each group and the results distinguished between groups are tried to explain for various periods. Empirical results serve the conclusion that Fisher effect is observed in Developed countries whereas it does not exist in G7 countries for either linear or nonlinear models.