Time-varying threshold regression model using the Kalman filter method

• Main Authors: Duangthip Sirikanchanarak, Worapon Yamaka, Chatchai Khiewgamdee, Songsak Sriboonchitta
• Format: Journal
• Published: 2018
• Subjects: Mathematics
• Online Access: https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85008352105&origin=inward; http://cmuir.cmu.ac.th/jspui/handle/6653943832/55955
• Institution: Chiang Mai University

© 2016 by the Mathematical Association of Thailand. All rights reserved. This paper explores a model, called the time-varying in threshold model with two regimes and which allows the regression coeffcients to change over time. This model take the advantage of the Kalman filter allowing the parameters to vary over time. We apply our model to analyze the effect of bank credit on GDP growth and ination because the financial time series data revealed strong signs of non-linearity and the context of the global economy has clearly changed in various dimensions. Note right away that the conventional threshold regression model appropriates when the relationship between dependent and independent variable seems constant, at least during the estimation period. Otherwise, a time-varying parameter non-linear model should be considered, especially in the context of structural change in the macroeconomics data. The main finding of this study reveals that there exists obvious important role the bank credit plays in the growth of the economy and inflation and there is a difference in behavior between regimes. However, after 2005 the effect from bank credit on GDP growth and iflnation are quite smooth partly due to change in the monetary policy is called inflation targeting and reform the credit regulations of the commercial bank to more caution.