A SVAR model of the Thai economy

Submission note: A thesis submitted in total fulfilment of the requirements for the degree of Doctor of Philosophy to the Department of Economics and Finance, La Trobe Business School, La Trobe University, Bundoora.

This thesis comprises five related chapters on application of structured vector autoregression (structured VAR) model for the Thai economy. Chapter 1 is an introduction to the thesis. It describes the objectives of the thesis and expectations, and provides essential background of Thailand’s changing financial and economic landscape, recent domestic events and existing empirical Thai models. All together, they serve as necessary introduction to the subsequent chapters. Chapter 2 discusses background and motivation for the thesis. It presents macroeconomic models used in policy analysis at central banks, their key usage and strength. It then zooms on New Keynesian dynamic stochastic (DSGE) models, which is applied to our model building, and the structural vector autoregression (SVAR) framework, which our model is founded on, followed by a discussion on identification of the SVAR models and their limitation. The chapter then discusses the hybrid DSGE-VAR model and the structured VAR model, each combines strength of the DSGE and SVAR models in attempt to give a specification that can be used for policy analysis as well as forecasting. Chapter 3 provides the description and analysis of the Thai data. It investigates time series properties of the data necessary for model building in the next chapter. The framework for testing unit root and stationarity in the presence of structural breaks are discussed, and the tests of unit root, stationarity and normality test are performed. Chapter 4 applies the structured VAR model to the Thai economy to investigate the effects of shocks to the Thai economy as well as to examine the credit channel of monetary transmission. The theory-driven core model is set up first, then augmented so as to move it towards data-oriented one and to make it a better representation of the Thai economy. The chapter then discusses the Beveridge-Nelson decomposition. The limited information estimation approach is employed, and the structured VAR model is estimated with instrumental variable (IV) using the generalized method of moments (GMM) estimator. Chapter 5 concludes. It provides a brief summary of the thesis and summarizes empirical findings from this research. Policy implications drawn from the structured VAR model are discussed and ideas for further work are made.