The Importance of Location and Housing Type with Respect to Future Residential Sector Energy Use

The Importance of Location and Housing Type
with Respect to Future Residential Sector Energy Use

by
John H. Cymbalsky

Households use energy to provide a wide variety of necessary services, such as space heating and cooling, water heating, and lighting, and to power a number of other appliances. The amount of energy consumed depends on such factors as the type, size, and location of the house; race and income level of the household; and the efficiency of both the equipment and the building shell. The purpose of this paper is to examine the importance of projected changes in housing patterns—in terms of location and housing type—with respect to residential sector energy consumption in the Annual Energy Outlook 1998 (AEO98) reference case.

Introduction
Background
Case 1: Housing Stock with Fixed Regional Shares
Case 2: Housing Stock with Fixed Building Types
Case 3: Housing Stock with Fixed Regional and Building Type Shares
Conclusions

Introduction

By 1995, the residential sector in the United States consisted of almost 100 million households, which are considered to be the primary residences for a population of more than 260 million people. Households use energy to provide a wide variety of services, from necessities such as heating and refrigeration to convenience items such as garage door openers and microwave ovens. The amount of energy consumed in households depends, in part, on how many and how often appliances are used, which in turn depends on the location, occupancy level, and size and type of residential structure.

This paper examines two important factors affecting energy consumption in households, namely, the type of future additions to the housing stock and their location. The reference case developed for the Annual Energy Outlook 1998 (AEO98) projects a number of changes in housing trends and, as a result, residential energy consumption between 1995 and 2020. In order to gauge the importance of the projected shifts in U.S. housing trends, this analysis examines projections that were derived by fixing future household additions in proportion to their historical shares in the stock. The analysis examines the importance of changes in housing type and location separately and together. Because the focus of the analysis is energy consumed by households, all energy use is stated in terms of delivered energy, to remove the effects of efficiency changes in the electric utility sector; however, to illustrate the importance of other fuels in the generation of electricity, electricity-related losses are included in the graphs shown in the “Analysis Results” section.

Background

Of all the factors affecting energy demand in the U.S. residential sector, the types and locations of houses are the most important. Figure 1 shows the number of U.S. households in 1995 by type (single-family, multifamily, and mobile homes) and Census region. The South Census region, with more than 36 percent of the nearly 100 million households in the Nation, is the largest in terms of households.

Single-family homes are most prevalent, accounting for more than two-thirds of the stock and consuming more than three-quarters of the delivered energy used in the residential sector (Figure 2). These homes tend to be, on average, larger than the other types in terms of both physical size and number of occupants, requiring more energy for cooling, lighting, and space and water heating. Multifamily units, on the other hand, account for 25 percent of all housing units but consume only 17 percent of the delivered energy used in the sector. Small dwelling size, fewer occupants per unit, and a higher percentage of units heated with electricity, which is more efficient on a delivered basis, all contribute to energy consumption that is less than its share of the housing stock. Mobile homes, which use liquefied petroleum gas (LPG) more often than other homes and are concentrated in the South Census region, account for 6 percent of the housing stock and consume 5 percent of all residential energy.

Figure 1. Households by Type and Census Region, 1995

Source: Energy Information Administration,
Office of Integrated Analysis and Forecasting,
Housing characteristics data for 1997 are available
from web site www.eia.doe.gov.

The climate, particularly as it relates to demand for space heating, has a significant influence on energy consumption. Although natural gas has far fewer uses within the home than does electricity, its consumption—47 percent of delivered energy in the residential sector—is considerably higher, primarily because of the high level of demand for space heating, which makes up 54 percent of all delivered energy use in the sector (Figure 2). Energy consumption in the Northeast and Midwest Census regions, which hold slightly less than half (44 percent) of all U.S. households (Figure 1), accounted for 54 percent of the U.S. total in 1995 (Figures 2 and 3), further demonstrating the importance of space heating in the residential sector.

Figure 2. Residential Sector Delivered Energy Consumption, 1995 Figure 3. Residential Sector Delivered Energy
(Quadrillion Btu) Consumption by Major Fuel and Region, 1995

                                                                                                                                                      Source: Energy Information Administration,
                                                                                                                                                      Office of Integrated Analysis and Forecasting

Note: Totals may not be equal due to independent rounding.
Source: Energy Information Administration,
Office of Integrated Analysis and Forecasting.

Because climate plays such a significant role in determining the types and amounts of energy consumed from year to year, it is important to relate “normal” (30-year average) weather conditions to those experienced in 1995 in the four Census regions. Table 1 lists the 1995 and 30-year average heating and cooling degree-days¹ for each Census region. The number of heating and cooling degree-days has a direct effect on the amount of fuel consumed for space heating and cooling. For example, if the trend in housing is away from the northern climates toward the southern climates, then the amount of fuel needed for heating decreases and the amount of fuel needed for cooling increases. Table 1 shows that a shift in housing to the South from the Midwest yields half as many heating degree-days, but two and a half times as many cooling degree-days.

Table 1. 1995 and 30-Year Average Heating and Cooling Degree-Days per Year by Census Region

Region	Heating Degree-days		Cooling Degree-Days
Region	1995	30-Year Average	1995	30-Year Average
Northeast	6,021	6,061	726	609
Midwest	6,706	6,499	940	809
South	2,838	2,852	2,118	2,021
West	3,374	3,830	884	831
Source: Energy Information Administration, Office of Integrated Analysis and Forecasting.

National Energy Modeling System (NEMS) Analytic Approach

Given the importance of the factors described above, the NEMS residential module was developed to account for changes in fuel types, end-use efficiency, regionality, and construction patterns for the different housing types. Accordingly, the NEMS residential module represents 7 fuel types, 13 end uses, 9 Census divisions, and 3 building types.²

Demand for Energy Services

The energy required for end-use services can vary widely, depending on the type and location of household. Therefore, it is imperative that the energy intensities (the amount of delivered energy used per household) associated with the different housing types and regions be accounted for. The base year (1993) intensities for energy services and fuel types by Census region and housing type are based on the Energy Information Administration’s (EIA’s) Residential Energy Consumption Survey (RECS). RECS provides NEMS with data on the number of households, number of appliances, and energy use associated with specific appliances. Figures 4 and 5 show 1995 energy use per household for the major fuels both by Census region and by housing type. By averaging the energy consumption over all households, regardless of whether they use the fuel or not, the relative importance of each fuel can be determined for the different regions and building types.

Figure 4. Household Delivered Energy Use by Fuel                         Figure 5. Household Delivered Energy Use by Fuel
and Region, 1995                                                                                     and Housing Type, 1995



Source: Energy Information Administration,                                    Source: Energy Information Administration,
Office of Integrated Analysis and Forecasting.                                 Office of Integrated Analysis and Forecasting.

To estimate future energy demand, the NEMS residential module employs a stock/vintage approach, which projects the numbers and efficiency of major household appliances over time. As older appliances in the stock are replaced by newer, more efficient models, energy use per appliance decreases, all else being equal.

The stock of energy-using equipment is a function of the saturation levels for the various end-use services. Of the major end uses represented in NEMS, only central air conditioning and clothes drying are assumed not to be fully saturated. In other words, ownership of these appliances has been increasing and is projected to continue to increase over time. All other major end uses (heating, water heating, cooking, refrigerators, and freezers) are assumed to be fully saturated at their respective 1993 levels for new and existing housing, since their respective ownership levels have been stabilized.

Once the amount of equipment needed to meet the demand for the entire housing stock is known, estimates of energy consumption can be calculated. Some services, such as space and water heating, can be furnished by more than one fuel type. In these cases, decisions about fuel type must be made before energy use can be estimated. Future energy prices, which are determined by the interaction of all the NEMS supply and demand modules, will affect fuel types, energy efficiency, and the intensity at which fuel is used in future years. It is assumed that rising real energy prices over time will lead to decreasing energy intensity through improved equipment and building shell efficiency and changes in behavior, such as adjusting thermostat levels.

Efficiency of Energy Services

Several factors contribute to the efficiency of the appliance stock over time, including energy prices, Federal efficiency standards, turnover rates, the relative intensities at which appliances are used, and the purchase costs of competing technologies. Because market characteristics cause investments in energy efficiency to be evaluated at high implicit discount rates,³ energy prices tend to have a relatively small impact on consumer choice with regard to the efficiency of purchased appliances. Many barriers in the residential market contribute to high implicit discount rates, including short occupancy periods, renter-occupied units (currently around 35 percent of the stock), emergency equipment replacements, and general inertia regarding equipment purchases. Generally speaking, appliance efficiency is higher in owner-occupied single-family households, for which total energy requirements are greater. For example, a large single-family owner-occupied home in the upper Midwest would be more likely to invest in an energy-efficient gas furnace than would a rental unit in the deep South.

Analysis Results

There are many factors that influence the forecast for future residential energy consumption in the United States, including:

Housing and population elements

Technology characteristics and availability

Market forces.

The number of occupied households plays a key role in determining residential sector energy use. In the forecast, occupied households are a function of housing starts, which are a function of economic activity and population trends. This paper examines the energy consequences of several cases in which housing starts are altered to control for the effects of changing housing patterns by both Census region and structure type, using the AEO98 reference case as a point of comparison.

Housing and Population Elements

To better understand the effects of housing and population changes on residential energy use in the AEO98 reference case, it is necessary to examine household formation. The NEMS residential module bases its estimate of occupied households on data from EIA’s 1993 RECS. As the levels of economic activity (i.e., income) and population increase over time, housing starts—the key economic indicator in the housing sector—increase as well. This variable serves as the key driver in the NEMS residential module.

To isolate the effects of shifts in location and type of house, a control case was examined that adds future households to the stock in the same proportion that they represent in the existing stock, leaving the level of total additions unchanged from that in the AEO98 reference case. To establish the relative importance of regional shares and housing types in terms of future U.S. residential energy consumption, the regional shares and housing types were fixed at their 1999 reference case levels. Three cases were examined:

Case 1: Fix housing starts by region to their 1999 reference case shares,
allowing the number of starts and the building types to vary as they do
in the reference case

Case 2: Fix housing starts by type at their 1999 reference case shares,
allowing the number of starts and their regional distribution to vary as
they do in the reference case

Case 3: Fix housing starts by type and by region at their 1999 reference
case shares, allowing the number of starts to vary as they do in the reference case.

For all three cases, 1999 served as the base year, with 2000-2020 serving as the analysis period.

Case 1: Housing Stock with Fixed Regional Shares

In this sensitivity case, all housing starts from 2000 to 2020 were set at the AEO98 reference case levels, but the regions in which the houses are built were representative of the stock as it existed in 1999. This case serves to establish the importance of the regional migration of the population assumed in the AEO98 reference case forecast.

Figure 6 shows additions of new households in the reference case and in the sensitivity case with static regional housing shares. The regional trend in housing is clear; the shift is away from the northern climate regions (Northeast and Midwest Census regions) and toward the south and western climate regions (South and West Census regions). In this sensitivity case, the number of households added in the Northeast Census region through 2020 would be more than double the number projected in the AEO98 reference case. The South Census region, which is projected to show the strongest growth in the AEO98 reference case, would be the most adversely affected.

Figure 6. Household Additions by Region in the
Reference and Fixed Regional Shares Cases, 2000-2020

Source: Energy Information Administration,
Office of Integrated Analysis and Forecasting.

Given the shift away from the colder regions of the country in the AEO98 reference case, it is intuitive that space heating, and the fuels associated with it, would be most affected in the case with the housing stock at fixed regional shares. Figure 7 shows changes in delivered energy consumption by fuel and end use in the two cases, as well as the change in energy losses associated with electricity generation, transmission, and distribution (i.e., electricity-related losses). All the major space heating fuels—natural gas, distillate, and LPG—would increase in importance if housing were constructed according to the regional share of households in 1999. Electricity, which is used in virtually every home, regardless of region, shows little variation between the two cases.

Figure 7. Change in Residential Delivered Energy
Consumption by Fuel and end Use and Electricity-
Related Loses in the Reference and Fixed Regional
Shares Cases, 2000-2020

Note: Electricity-related losses were calculated
at 2.20 Btu lost per Btu of electricity delivered in 2000.
Source: Energy Information Administration,
Office of Integrated Analysis and Forecasting.

Case 2: Housing Stock with Fixed Building Types

For this sensitivity case, as in case 1, housing starts in 2000-2020 were set at the levels projected in the AEO98 reference case, but the types of homes built were set at the proportions that existed in the 1999 stock. This case serves to isolate the energy effects of the projected shift in housing types in the AEO98 reference case.

Figure 8 shows additions of new households in the reference case and in the sensitivity case with fixed building types. The figure shows the projected increase in importance of mobile homes in the housing stock, relative to the other types of housing, in the AEO98 reference case. The number of mobile homes added through 2020 in the reference case is nearly six times that in the sensitivity case. Additions of single- and multifamily households are correspondingly higher in the sensitivity case.

Figure 8. Household Additions by Type in the
Reference and Fixed Building Type Cases, 2000-2020

Source: Energy Information Administration,
Office of Integrated Analysis and Forecasting.

The type of house built has a direct affect on fuel consumption, because the different housing types use fuels in different proportions. For instance, mobile homes tend to use LPG more frequently for space heating than do either single-family or multifamily structures. Given the variation of fuel shares among the three housing types, fuel consumption in the housing stock with fixed building types case should vary from that projected in the AEO98 reference case.

Figure 9 shows the change in residential energy consumption from 2000 to 2020 by fuel and end use, as well as the change in electricity-related losses, in the reference and sensitivity cases. Given the shift toward more mobile homes in the reference case, LPG consumption is lower in the fixed building types sensitivity case, whereas natural gas and electricity consumption is higher. Distillate use changes little from the AEO98 reference case, because its use is dependent on region rather than building type. In terms of end-use consumption, space conditioning increases in this case, because the housing stock, on average, is larger in terms of physical size, requiring more fuel to heat and cool the larger floor space.

Figure 9. change in Residential Delivered Energy
Consumption by Fuel and End Use and
Electricity-related Losses in the Reference and
Fixed Buildig Type Cases, 2000-2020

Note: Electricity-related losses were calculated
at 2.20 Btu lost per Btu of electricity delivered in 2000
and 1.88 Btu lost per Btu of electricity delivered in
2020.
Source: Energy Information Administration,
Office of Integrated Analysis and Forecasting.

Case 3: Housing Stock with Fixed Regional and Building Type Shares

The third sensitivity case combined the assumptions of the first two cases. This case, therefore, factors out all the regional and housing type shifts that affect residential sector energy consumption in the AEO98 reference case.

Figure 10 shows additions of new households from 2000 to 2020 by Census region and housing type in the reference and sensitivity cases. The largest source of positive change in terms of household additions in the AEO98 reference case is in the South Census region and, in particular, in multifamily and mobile homes. The numbers of single- and multifamily homes in the Northeast are much smaller in the AEO98 reference case than in the fixed shares sensitivity case.

Figure 10. Household Additions by Type and
Region in the Reference and Fixed Shares Cases,
2000-2020

Source: Energy Information Administration,
Office of Integrated Analysis and Forecasting.

The energy implications of this sensitivity case basically combine those of the first two cases. Electricity consumption in this case is identical to that in the AEO98 reference case (Figure 11), indicating that in the AEO98 reference case, the decrease in electricity consumption related to shifts in housing types is offset by the increase related to shifts in the regional distribution of housing starts. Natural gas consumption, on the other hand, shows a relatively large increase in this case relative to the AEO98 reference case. With the shares of housing starts by both housing type and Census region fixed at 1999 stock levels, natural gas consumption increases as more homes with relatively high natural gas intensities are added in cold climates than are projected in the AEO98 reference case.

Figure 11. Change in Residential Delivered Energy
Consumption by Fuel and Census Region and
Electricity-Related Losses in the Reference and
Fixed Shares Cases, 2000-2020

Conclusions

By examining trends in the major driver of residential energy consumption—housing starts—the importance of both location and type of household with respect to projected residential energy consumption can be quantified. This analysis has shown that for the AEO98 reference case, the location (i.e., climate) of the housing stock has a larger impact on residential sector delivered energy consumption than does the type of house. The projected shift in housing starts from the Northeast and Midwest Census regions to the South Census region in the AEO98 reference case has the greatest effect on the fuels used for space heating—specifically, natural gas and distillate. The type of house built, while having less impact on delivered energy consumption than location, still affects fuel choice. The shift away from single-family homes toward mobile homes in the AEO98 reference case dampens the potential for natural gas, because mobile homes tend to use LPG and electricity as a space heating fuel more often than do single-family homes.

To:
Measures of Oil Import Dependence

Back To:
Issues in Midterm Analysis and Forecasting 1998 Homepage

File last modified: July 22, 1998
URL: http://www.eia.doe.gov/oiaf/issues98/house.html

Contact the National Energy Information Center for questions at (202) 586-8800

If you having technical problems with this site, please contact the EIA Webmaster at wmaster@eia.doe.gov