An examination of electricity consumption patterns in manufactured housing units

Abstract

For several decades, manufactured housing has been a crucial source of affordable housing, particularly for rural areas. However, electricity consumption per unit area and per capita are substantially higher for manufactured housing units relative to site built, single-family detached units. This article uses data from the federal Residential Energy Consumption Survey (RECS) to examine patterns of electricity consumption in manufactured housing units over time and to draw comparisons with single-family detached housing units. Regression analysis is used to model annual electricity consumption for manufactured housing units in 1990 and 2005. Temporal trends in key predictors are discussed and contrasted with those for single-family detached units. Findings suggest that the most important predictors of electricity consumption are comparable across the housing types considered and that while manufactured housing units may be gaining in energy efficiency over time, consumption per unit area and per capita are increasing faster than in single-family detached units.

Keywords:

• manufactured housing
• electricity consumption
• affordable housing
• energy efficiency

Notes

¹The Energy Independence and Security Act of 2007 mandates a revision of the HUD Code by December 2011.

²The other distinguishing factor is that the HUD Code requires a chassis.

³Hammad, Hastak, and Syal (2004) provide a concise and accessible overview of the manufactured housing production process.

⁴These are the regional designations established by the US Department of Commerce, Bureau of the Census.

⁵There are regional variations in HUD Code standards to account for climate (e.g. roof load) and safety (e.g. wind zone) differences.

⁶Members participate on a part-time basis and have other responsibilities within HUD (GAO 2008).

⁷Both surveys also include estimates of how much of the annual usage was devoted to space heating, space cooling, water heating, appliances, etc. These estimates are provided with relatively little transparency and therefore, were not used to further dissect electricity consumption. The annual usage data were instead selected as the dependent variable because they were provided by the household's utility company and are presumably more reliable.

⁸Information on dwelling location from the federal surveys was dichotomized in the interest of parsimony and to ensure balanced representation across the categories considered. The manufactured housing regression models were re-estimated using the disaggregate categories from the RECS and the adjusted R-squared was unchanged for both the 1990 and the 2005 models. Further, only the “City” dummy variable was significant and exclusively in the 2005 model. The implication here is that the details of how location along the urban–rural gradient is operationalized exert very little influence on the broader inference taken away from the regression analysis.

⁹Specifically, variation in the impact of predictors on the dependent variable in the tails of the distribution of the dependent variable.

¹⁰Weatherization studies are one potential exception.