Many forecasts for energy use and carbon emissions assume that energy intensity will decline over time for reasons unrelated to energy prices, which are termed autonomous energy efficiency increases (AEEI). A cointegration analysis of a vector error correction model indicates that the types of fuels consumed, personal consumption expenditures spent on energy, and energy prices account for changes in the ratio of energy use to economic activity in the US between 1929 and 1999. Cointegration indicates that AEEI is associated with technical and/or structural changes which allow consumers to substitute oil, natural gas, and/or primary electricity for coal, and that shift energy use from final demand to intermediate sectors. Identifying the factors responsible for AEEI allows me to: (1) show that econometric efforts to measure technical change using a deterministic trend are inconsistent with economic theory and cannot be interpreted reliably; (2) show that modeling technical change with a deterministic trend may generate forecasts that overstate reductions in energy use and carbon emissions; and (3) test the observational record for the presence of price-induced technical change and its effect on economic growth. Together, the results indicate that current estimates for AEEI may overstate future reductions in energy use and that the economic impacts of policies to reduce energy use and slow emissions may have a greater effect on economic growth than anticipated currently.