A research submitted in partial fulfillment of the requirement for the degree of Master of Engineering

Research Studies Project Report (M.Eng.) - Asian Institute of Technology, 1993

The choice of the optimal functional form as an econometric problem in estimating energy-GDP relationship is investigated with the help of Box-Cox transformation using maximum likelihood estimation procedure. The functional relationship between aggregate energy and real gross domestic product (GDP) and between electricity consumption-GDP are examined for selected Asian countries using time series data from 1960 to 1990. In majority of the cases all three special viz. loglinear, linear and semilog functional specifications are rejected, suggesting no strong evidence in favor of any particular functional form; thus making the presumption of functional form very much arbitrary. In those cases, where the chosen functional form is among linear, loglinear or semilog; the stability of parameters using Chowtest is conducted with two data set by dividing the complete time series data into two regression data set from 1960-1973 and 1974-1990. It is found that estimated parameters are not stable over time and changed drastically. This test shows the mistake one makes by pooling together two set of data with different time periods without some modification in the functional relationship. Difference in selected optimal function and estimated results resulting due to difference in employed aggregation procedure is examined by two most common; BTU and Divisia; aggregation method using data after 1973 (where energy commodity price data are available). There is inconsistency in choosing the functional form. The result has suggested that values of estimated parameters depend on the method used for price and energy aggregation.