Threshold Regression Model for Taylor Rule: The Case of Turkey


  • Pinar Deniz Marmara University
  • Thanasis Stengos University of Guelph
  • Ege Yazgan Istambul Bilgi University



Nonlinearities, Taylor rule, Threshold regression models


This paper employs the structural threshold approach of Kourtellos et al. (2016) to examine various specifications of the Taylor rule model. Contrary to the previous work on the Taylor rule, this methodology allows for endogeneity of the threshold variable in addition to the right-hand-side variables suggesting a fully comprehensive flexible framework that does not rely on restrictive linearity and/or exogeneity assumptions. In order to examine the model, Turkey is selected as an inflation targeting developing economy, since its central bank (the Central Bank of Turkey) as argued by Dincer and Eichengreen (2014) has been one of the fastest improving central banks in terms of its transparency score. We use monthly data for the period of 2004-2018 that includes a number of historical episodes such as the global financial crisis as well as various internal political developments that may have had an impact on the fluctuations of the relevant macroeconomic variables as well as on the functional form of the inflation targeting Taylor rule specification. Empirical findings highlight the different reactions of the central bank in determining policy rate under different regimes.