Contributing to better energy and environmental analyses: how accurate are decomposition analysis results?

This paper attempts to contribute to the improvement of decomposition analyses for better policy-making. This is achieved under Shapley (1953)/Sun (1998)’s approach, by taking into account the net entry effect of products. Indeed, we propose to estimate the standard errors of contributions using bootstrapped normal-approximation confidence interval in order to investigate whether or not effects are significantly different from zero at standard significance levels. Therefore, our work introduces a new criterion for the choice of decomposition approaches, so-called accuracy criterion. The application is based on a decomposition of CO2 emissions embodied in China’s bilateral trade (EEBT). The results show that omitting the net entry effect can lead to under- or over-estimates of the contributions, wrong signs and even a wrong order of magnitude of the contributions, and incorrect estimation of the effects’ accuracy. Also, the analyses reveal that the effects may have different accuracy levels and some effects (even those with large magnitude contributions) may be non-significantly different from zero. This implies that there are effects that are not relevant to the explanation of China’s EEBT. Our results therefore suggest that not considering the net entry effect or not being aware of the effects’ accuracy may lead to incorrect economic interpretations and misguided policy-making. Another interesting point arising from the methodology is that the complexity of S/S’s approach, as the number of effects increases, should go hand in hand with the results’ accuracy. Hence, from this standpoint, the S/S’s approach should be preferred when the number of effects is high, contrary to what is stated in the literature.