Display omitted •Granger causality tests performed for GDP and energy consumption (EC) in US states.•Panel Granger causality is tested for the US and for regions within the US.•Wide variation between US states and regions.•Including capital and labor changes direction of Granger causality in 35/50 states.•Energy conservation policy design should accommodate within-country variation. Does energy consumption lead to economic growth or does economic growth cause energy consumption? Arguments can be made either way and empirical studies have been inconclusive. Most of the existing studies relating Gross Domestic Product (GDP) and Energy Consumption (EC) use countries as the unit of observation which complicates the interpretation and generalization of results because countries differ greatly in their stage of economic development, culture, technology and so forth. This study focuses on states within the U.S.A. which avoids many of these complications. Specifically, the relationship between state energy consumption and state GDP for the country is analyzed. Empirical results using panel cointegration and panel causality tests which allows for heterogeneity and structural breaks are applied to the country as a whole and regions within the country. There are significant regional differences within the U.S. especially for two regions; in the Rocky mountain region energy consumption Granger causes state GDP and in the Southwest it is opposite, GDP Granger causes energy consumption. The full results suggest that federal energy policy needs to be flexible to be most beneficial to the different regions.