Energy consumption and economic growth in 12 Asian countries: panel data analysis

Abstract

This study examines ‘growth, conservation, neutrality or feedback’ hypotheses in 12 Asian countries for the period 1970 to 2010, using panel causality analysis, which accounts for dependency and heterogeneity across countries, supports evidence on the direction of causality and is consistent with the neutrality hypothesis in two-thirds of these 12 Asian countries. Growth hypothesis and conservation hypothesis hold for India and Philippines, respectively. However, a feedback was found for both Thailand and Vietnam. Thereby, the findings provide important policy implication for 12 Asian countries under study.

Keywords:

• energy consumption
• economic growth
• Asian countries
• dependency and heterogeneity
• panel causality test

JEL Classification:

• C23
• C33
• Q43
• Q53

Notes

¹ We refer to Ozturk (2010) and Payne (2010) for a recent survey on the energy consumption and economic growth nexus.

² In order to save space, we refer to Pesaran and Yamagata (2008) for the details of estimators and for Swamy's test.

³ Since country-specific bootstrap critical values are used, the variables in the system do not need to be stationary, implying that the variables are used in level form irrespective of their unit root and cointegration properties. Thereby, the bootstrap panel causality approach does not require any pretesting for panel unit root and cointegration analyses. Besides, by imposing country-specific restrictions, we can also identify for how many and for which countries there exists Granger causal relation.

⁴ It is important to note here that since the results from the causality test may be sensitive to the lag structure, determining the optimal lag length(s) is crucial for robustness of findings. As indicated by Kónya (2006), the selection of optimal lag structure is of importance because the causality test results may depend critically on the lag structure. In general, both too few and too many lags may cause problems. Too few lags mean that some important variables are omitted from the model and this specification error will usually cause bias in the retained regression coefficients, leading to incorrect conclusions. On the other hand, too many lags waste observations and this specification error will usually increase the SEs of the estimated coefficients, making the results less precise. For a relatively large panel, equation and variable with varying lag structure would lead to an increase in the computational burden substantially. In determining lag structure, we follow Kónya's approach that maximal lags are allowed to differ across variables, but must be same across equations. We estimate the system for each possible pair of , , , and , respectively, by assuming from one to four lags and then choose the combinations which minimize the Schwarz Bayesian Criterion.

⁵ We refer to Kónya (2006) for the bootstrap procedure on how the country-specific critical values are generated.