U.S. Commercial Buildings Energy Consumption and Intensity Trends: A Decomposition Approach



Behjat Hojjati


U.S. Energy Information Agency (Washington, DC)


Steven H. Wade


U.S. Energy Information Administration (Washington, D.C.)



Increasing energy efficiency is a subject widely referred to as a potential, important contributor to national energy policy goals.  In this context, historical efficiency gains are also of interest, but in practice bottoms-up efficiency estimates are too data intensive to develop. Thus, aggregate energy intensity, defined as total energy use per building or square foot is often used inappropriately as a proxy for energy efficiency. Energy efficiency is the ratio of service provided to energy input. Thus, efficiency gains occur when either energy use is reduced for a given level of service or there are increased services provided from a given amount of energy input. Aggregate energy intensity encompasses all factors affecting energy consumption and includes the effects of efficiency in addition to other non-efficiency components, like conservation changes (reduced energy due to reduced services) and structural changes. Structural changes quantified in our analysis include changes in the distribution of buildings across building types,[1] changes in the regional distribution of buildings, and changes in the average size of buildings.[2]

According to the latest complete Commercial Buildings Energy Consumption Survey (CBECS), U.S. commercial buildings used more than 6.5 quadrillion Btu (quads) of site energy in 2003, up from 5.3 quads in 1989.[3] Between 1989 and 2003 commercial buildings site energy increased by 20.1 percent, or an average annual growth of 1.4 percent.[4],[5] Over this period, the number of buildings grew by more than 11.4 percent and average floorspace per building increased by about 6.8 percent, meaning total commercial floorspace increased by about 18.3 percent. Aggregate energy intensity increased by 8.7 percent on a per building basis and 1.9 percent on a square footage basis. To provide a better estimate of efficiency effects embedded in historical energy intensity, this study separates and removes some structural changes. [6]


The total change in buildings energy consumption between 1989 and 2003 was heavily influenced by identifiable structural effects. After adjusting for the changes associated with regional shifts, building type mix changes, and changes in building size across regions and building types, the remaining changes are referred to as the decomposed energy intensity. Over this period, decomposed energy intensity decreased by 1.1 percent compared with the aggregate intensity increase of 1.9 percent, a reversal of direction--indicating efficiency gains versus the aggregate estimate of increased intensity. Therefore the simple aggregate intensity, if mistaken as a proxy for efficiency, would have underestimated efficiency gains by 3.0 percent.[7] In the context of total energy consumption, if only energy intensities had changed while the underlying structural factors had remained stable energy consumption would have grown 3.0 percent less over the 1989 through 2003 period – 17.1 percent instead of 20.1 percent.


Electricity and natural gas were the two main sources of energy use in commercial buildings. A significant increase in electricity intensity influenced the magnitude of the decline in estimated decomposed energy intensity (Figure 1).




Figure 1: Percent Change in Energy Intensity, 1989 to 2003

Source: Hojjati and Wade (2012), USAEE/IAEE conference proceedings.


The analysis of sub-periods 1989-1995 and 1995-2003 found that an increase in the average size of buildings in the post-1995 sub-period had the largest increasing effect on energy consumption.  Except for total electricity consumption, energy intensity per square foot decreased faster in the second sub-period, which could be an indication of the increasing influence and effectiveness of energy efficiency standards.  The positive intensity effect of electricity was larger in the second sub-period, which has implications for any policy initiatives aimed at sustainability and reducing carbon dioxide.

Additional analysis showed that the building type attributions are sensitive to the number of building categories considered.  However, the decomposed intensity results are much more stable as well as significantly different from the calculated aggregate intensities.


[1] Building types consist of assembly, education, food service, food sales, health care (inpatient), lodging, large offices (>50,000 square feet), small offices (<=50,000 square feet), mercantile and service, warehouse, and others.

[2] Other structural effects, which our analysis is unable to quantify, include the effect of energy prices, weather, other economic and market conditions, changing importance of specific end uses, and the mix of energy sources used in buildings. These effects are not considered in our decomposition analysis.

[3] This analysis is based on the Btu value of energy at the point it enters the building. This is referred to as delivered or site energy. It does not include losses that occur in the generation, transmission, and distribution of electricity, which when added to delivered energy becomes primary energy.

[4] CBECS 1989 is the earliest survey year that has complete information for energy intensity analysis, so the current full CBECS interval used here is 1989 through 2003.  The reported estimates are for buildings using energy.

[5] The growth rates cited here are based on natural logarithms for consistency with the decomposition methodology.

[6] The reported results are based on Hojjati and Wade (2012), USAEE/IEAA conference proceedings, “U.S. Commercial Buildings Energy Consumption and Intensity: A Decomposition Approach.”

[7] Energy efficiency programs (e.g. federal equipment standards, state and local building energy codes, federal and state tax credits, and other voluntary programs such as ENERGY STAR buildings) likely affected the decline in decomposed intensity.

Click to view a printable version of this article.