|
Title:
A macroeconometric model for Fiji
Untitled
Author:
Singh, Rup.
Institution:
University of the South Pacific.
Award:
M.A.
Subject:
Date:
2005.
Call No.:
Pac HC 685 .5 .S46 2005
BRN:
1019108
Copyright:
10-20% of this thesis may be copied without the authors written permission
Abstract:
Undoubtedly, the modern macroeconomy is large, difficult to define, measure and control. Therefore, economists attempt to build macroeconometric models to approximate it. Models may be used to conduct experiments such as policy analysis, forecasting or testing of theoretical relationships. But a single model cannot fulfill all the three objectives simultaneously. Far and large these models are the tools for economists that help them interpret the reality and therefore, the search for better specifications and estimation methods of model building is an ongoing process. Models are built for specific purposes and therefore should be evaluated on the basis for which they are built. The parameters of a model are estimated using relevant historical data together with reference to economic theory. Based on empirical estimates, inferences are made regarding the dynamic structure of the economy. Generally, for structural models, Two/Three Stage Least Squares (2/3SLS) or Full Information Maximum Likelihood Methods (FIMLE) are appropriate. Nonetheless, single equation estimation of such models based Ordinary Least Squares (OLS) or Partial Adjustment Methods (PAM) are also common. This is largely due to data limitations. However, these methods have come under serious criticisms in light of the recent developments in time series econometrics particularly following the unit roots literature. Consequently, in this thesis, we attempt to build a small scale structural macroeconometric model for Fiji using two time series methods, viz, General to Specific (GETS) and Johansen Maximum Likelihood (JML) approaches by paying full attention to time series properties of variables. The complete model is then simulated over the historical period and standard simulation evaluation tests are performed.
|
|
|
