Energy consumption–economic growth nexus for Pakistan: Taming the untamed

A recent survey of energy-growth literature has highlighted the potential trade-off between bivariate models that suffer from omitted variable bias, and the danger of over-parameterization of multivariate models in the individual country setting (Narayan and Smyth [2]). This is a serious limitation when the interest is in drawing policy implications for specific countries with short times series of available data. The maximum entropy bootstrap approach was used to re-examine the nature of causal relationship between energy consumption and economic growth for Pakistan where the available time series data was only from 1971 to 2011. Unlike the techniques used in much of the earlier literature, this approach does not rely on asymptotic methods and, therefore, leads to robust inference even in small samples. Moreover, the approach can be applied in the presence of non-stationarity of any type, and structural breaks, without requiring data transformation for to achieving stationarity, and is not sensitive to specification errors such as those in lag length selection. The empirical findings, based on both the bivariate as well as the multivariate frameworks, supported the conservation hypothesis, implying the existence of a unidirectional causality from economic growth to energy consumption.