Replication and Robustness Analysis of 'Energy and Economic Growth in the USA: a Multivariate Approach'

We replicate Stern (1993, Energy Economics), who argues and empirically demonstrates
that it is necessary (i) to use quality-adjusted energy use and (ii) to include capital and
labor as control variables in order to find Granger causality from energy use to GDP.
Though we could not access the original dataset, we can verify the main original
inferences using data that are as close as possible to the original. We analyze the
robustness of the original findings to alternative definitions of variables, model
specifications, and estimation approach for both the (almost) original time span (1949-
1990) and an extended time span (1949-2015). p-values tend to be substantially smaller
if energy use is quality adjusted rather than measured by total joules and if capital is
included. Including labor has mixed results. These findings tend to largely support
Stern’s (1993) two main conclusions and emphasize the importance of accounting for
changes in the energy mix in time series modeling of the energy-GDP relationship and
controlling for other factors of production. We also discuss how the inclusion of the
original author in designing the replication study using a pre-analysis plan can help to
counterbalance the incentive of replicating authors to disconfirm major findings of the
original article to increase the probability of getting published.