Energy is consumed by various segments of the economy, including households, commercial establishments, manufacturing enterprises, and electric power generators. Only a portion of total energy demand is sensitive to temperature changes. Energy consumption of gasoline and jet fuel has strong seasonal components but is not very sensitive to heating and cooling degree days. In addition, the derived demand for primary fuels used by electric utilities, such as coal, natural gas fuel, and residual fuel oil, is also sensitive to weather because the demand for electricity changes with variations in temperature and other atmospheric conditions.
The challenge for understanding the importance of climate on energy demand is to identify and estimate the separate effects that the determinants of energy demand have on energy consumption. The demand for energy depends on a number of factors, including the price of energy relative to prices for other goods and services, household income and demographic features, output and technology adoption by businesses, and weather conditions. Using history as a guide, the challenge is to determine to what extent trends in energy consumption are influenced by trends in these basic drivers of energy demand.
Considine develops energy demand models that estimate the separate effects that these factors have on energy consumption trends. For example, businesses are assumed to minimize energy expenditures subject to exogenously determined constraints, such as output levels, fuel prices, and weather conditions.
This optimization model provides a basis to formulate a set of energy demand equations that relate energy consumption and these exogenous factors. Moreover, this approach allows a consistent representation of substitution possibilities among fuels. For instance, if the demand for natural gas increases with an increase in fuel oil prices, suggesting that natural gas and fuel oil are substitutes, then the demand for fuel oil would increase with higher prices for natural gas. This property, known as zero-degree homogeneity in demand, essentially means that only relative prices matter in determining energy demand.
Given these considerations regarding the role of prices in determining energy demand, identifying the effect of climate change on energy demand requires measurement of two components of the climate signal: the fixed seasonal effect and a random weather shock. Energy demand during other months of the year may have similar fixed monthly or seasonal effects. The second effect, associated with random weather shocks, represents that portion of energy consumption directly associated with departures from normal temperatures.
The demand models developed by Considine assume that monthly energy consumption depends on economic forces, technological factors, fixed seasonal effects, and random weather shocks. Energy demand systems are estimated for four sectors of the U.S. economy: residential, commercial, industrial, and electric utilities. In the first step, Considine specifies that aggregate energy demand is a simple log-linear function of real energy price, income or output, fixed seasonal effects, heating and cooling degree days, and a time trend as a proxy for exogenous technological change. The second step, determining the fuel mix, assumes that the fuels in each sector’s aggregate energy price index are weakly separable from other factors of production.
