Assumption to the Annual Energy Outlook

Household Expenditures Module

The Household Expenditures Module (HEM) constructs household energy expenditure profiles using historical survey data on household income, population and demographic characteristics, and consumption and expenditures for fuels for various end-uses.  These data are combined with NEMS forecasts of household disposable income, fuel consumption, and fuel expenditures by end-use and household type.  The HEM disaggregation algorithm uses these combined results to forecast household fuel consumption and expenditures by income quintile and Census Division (see Figure 5). 

Key Assumptions 

The historical input data used to develop the HEM projections for AEO2004 consists of recent household survey responses, aggregated to the desired level of detail.  Two surveys performed by the Energy Information Administration are included in the AEO2004 HEM database, and together these input data are used to develop a set of baseline household consumption profiles for the direct fuel expenditure analysis.  These surveys are the 2001 Residential Energy Consumption Survey (RECS) and the 1991 Residential Transportation Energy Consumption Survey (RTECS). 

HEM uses the consumption forecast by NEMS for the residential and transportation sectors as inputs to the disaggregation algorithm that results in the direct fuel expenditure analysis.  Household end-use and personal transportation service consumption are obtained by HEM from the NEMS Residential and Transportation Demand Modules.  Household disposable income is adjusted with forecasts of total disposable income from the NEMS Macroeconomic Activity Module. 

The fundamental assumptions underlying HEM’s processing of the historical and NEMS forecast data to obtain its results are: 

• Individual households are assumed not to migrate between income quintiles throughout the analysis period. 
• All households within a household segment are assumed to consume the average quantity of fuel for that segment.  Distributions about, or deviations from, the average are not explicitly modeled. 
• The change in average household consumption between forecast year and survey base year is captured from the NEMS run at the finest available level of detail, and the same proportional change is assumed to occur in each HEM subsegment of the analysis. 

Application of the HEM algorithm produces a direct household fuel expenditure forecast at the finest level of disaggregation; namely, by fuel, end-use service, housing type and vintage, ethnicity, disposable income quintile, Census Division, and year.  Results obtained are summed across end-uses to yield total direct fuel expenditures as a function of disposable income for each household segment.  The consolidation of these high-resolution results into national average household expenditure results requires a weighted averaging in order to obtain the desired aggregations.  The weighing scheme used requires the proportions of households of each type and vintage headed by householders of each ethnicity and income quintile.  The survey data provides these historical subsegment proportions, and for the AEO2004 they are assumed to remain constant throughout the forecast period.

Released: February 2004