Economic Growth Analysis of Singapore
We find very useful to begin the introduction by taking a look at the beginnings of simultaneous equations system. As it is known in the literature of econometrics, when we use the simultaneous equations system, we decide to deal with several linear or dynamic regressions basing to the macro-economic theory. Thus, a simultaneous equation model (SEM) will be available provided they has been chosen in the light of the economic theory allowing a correct diagnosis of the system and reflecting the real interactions between variables which their use helps in prediction and in proper planning. For this the importance of looking to the interactions between the variables, on the one hand, to realize a correct estimate of the equations, and on the other hand, to have for the ability to interpret them. The proposed equations, which are known as structural equations, must comport with the economic theory.
This paper has carried out an in-depth study based on the simultaneous equations model by estimating three structural equations associated to the three components of the Real Gross Domestic Product per Capita (gdp) in Singapore over the period (1991-2017), that is, the Real Gross Domestic Saving per Capita (gds), the Household Final Consumption Expenditure per Capita (hfce), the Government Final Consumption Expenditure per Capita (gfce). The primary nominal data were divided by the product of the consumer price index and the annual population for leading to real data per capita taking into account both inflation and population. The fourth equation represented the income identity expressed by equality (gdpt=gdst+hfcet+gfcet). Seven instruments variables are used to accomplish the study: a constant, three predetermined variables characterized by gdst-1, hfcdt-1 and gfcet-1, three exogenous variables as real interest rate (rirt), the real foreign direct investment per capita (fdit) and the real money supply per capita (m1t). The study shows that the three structural equations are over-identified and by consequence; each equation is estimated using the following methods: Two-Stage Least Square estimator (2SLS), HeteroscedasticTwo-Stage Least Squares (H2SLS), Limited Information Maximum Likelihood (LIML) and the Three-Stage Least Squares (3SLS) which is often more efficient than other methods and promoted by Hausman test. Finally, the performance of the estimated equations is measured comparing the fitted values with the observed values by the Mean Relative Error (MRE). The findings have shown that the MRE values are 2.46%, 1.37%, 4.9% and 1.37% for the variables gdst, hfcet gfcet and gdpt respectively.
Finally, to measure the performance of the estimated equations, the fitted values are calculated and a comparison with the observed values is performed. This is measured by the Mean Relative Error (MRE) expressed as a percent recalling that the (MRE) expresses how large the absolute errors are compared with the observed values we are measuring. The findings in Appendix 3 shown the MRE values 2.46%, 1.37%, 4.9% and 1.37% for the variables gdst, hfcet, gfcet and gdpt respectively.
Journal of Business & Financial Affairs