Are Shocks to Energy Consumption Persistent? Evidence from Subsampling Confidence Intervals.

This paper analyzes the persistence property of energy use in 107 countries around the world during 1971-2011 using different subsampling confidence intervals introduced by Romano and Wolf (2001). These confidence intervals are much more informative than the unit root tests and are more robust to misspecification errors as they require fewer assumptions on the nature of data generating process. While providing evidence about the stationarity or non-stationarity of the variables, they also show the degree of persistence and consequently are very informative for the role of government intervention in environ- mental oriented policies. The findings show that there are three classes of countries in terms of energy use: with explosive behavior (highly populated with high growth economies- 4 countries); non-stationary (developing and highly oil dependent economies- 64 countries); and stationary (generally developed and energy-rich countries- 39 countries). An explosive behavior of energy use would make government environmental related policies improbable as they require strict enforcement rules for the policy to be effective. For the nonstationary cases, government interventions can be effective for energy conservation and other environmental oriented policies, while for the stationary cases the effect of government intervention, energy conservation or environmental-oriented demand-management policies would be tem- porary, and their effects will not last long.