Türkiye'de satın alma gücü paritesine yakınsama: Şehir düzeyinde veri ile kanıt

This thesis examines the price dispersion in Turkey over the period of January 2003 ? June 2011 by using consumer price indices for twenty-six Turkish city groups. Preliminary analysis of the time series properties of price levels and deviations from purchasing power parity (PPP) with base cities Ankara, Istanbul, and Diyarbakir shows that deviations can be considerably persistent. The thesis examines convergence to PPP more formally by using two unit root tests (i.e., tests where the null is that the PPP deviations have a unit root) and a stationarity test (i.e., test where the null is that the PPP deviations are stationary). Unit root and stationarity tests show that deviations from PPP have a unit root for many of the cities. Specifically, the stationarity test rejects the null of stationarity in deviations for all cities when Istanbul is the base city and rejects all but one when Ankara and Diyarbakir are the base cities. Although unit root tests provide relatively more promising results in favor of PPP in Turkey, overall findings are not supportive of a weak form of PPP. Moreover, unit root tests show some variation across different base cities. For example, the Augmented Dickey-Fuller test provides evidence in favor of PPP in three, eight, and twelve city groups with Ankara, Istanbul, and Diyarbakir, respectively, as the base cities. Given the rather bleak evidence against PPP within Turkey, the thesis attempts to quantify the degree of persistence in deviations by using half-life measures. Despite some shortcomings of the measure used, the thesis provides some preliminary results that show relatively short half-lives for PPP deviations. Findings reveal that deviations from PPP on average halves within less than one year and the speed of convergence to PPP changes with the base city used. Deviations from PPP with Ankara and Istanbul as base cities tend to converge slower than Diyarbakir as the base city. In order to provide some insights into the variations in the persistence of PPP deviations across cities, the thesis regresses half-life measures on a proxy of transportation cost and estimates linear and nonlinear probability models to check to what degree outcomes of unit root tests can be explained by the transportation costs. Results show that transportation costs can plausibly explain both the variability in the persistence of deviations and the likelihood of having a unit root in the deviations.