The multiple regression technique was applied
using three functional forms namely, linear, semilog
and Cobb-Douglas. The best fit was selected
after considering the levels of estimated error,
magnitude of R2, number and signs of estimated
regression coefficients. Explicitly the model is
specified as follows