Fuel-factor nesting structures in CGE models of China

Highlights: • KE-L is a more appropriate nesting structure than KL-E for China. • Previous exercises might have overestimated costs of energy conservation. • In the preferred KE-L bundle, energy saving is dominated by output effects. - Abstract: Fuel-factor nesting structures are critical features of CGE models but have rarely been chosen in a robust manner. This study thoroughly examines two popular fuel-factor nesting structures in CGE models of China, namely KL-E¹ and KE-L.² We use non-linear algorithms to estimate the underlying CES production functions of these two structures. We recommend the latter to be used in CGE models of China, as it exhibits 1) better goodness of fit, 2) better convergence, and 3) better compliance with economic meanings. Our estimations show that the ranges for the inner- and outer-CES parameters³ in our preferred KE-L structure are 0.6–0.8 and 0.7–0.9, respectively. Applying these in a CGE model, we find nesting strategies used in previous CGE exercises might have overestimated energy conservation costs in China. A decomposition shows that, using our KE-L, changes in energy use are dominate by output effects rather than substitution effects.