EVALUATION OF THE 1984-1992 IMPACTS OF BONNEVILLE'S 1993 INTERIM RESIDENTIAL WEATHERIZATION PROGRAM FINAL REPORT

---

PREPARED BY SYNERGIC RESOURCES CORPORATION (SRC)

This report was prepared by Synergic Resources Corporation (SRC) for the Bonneville Power Administration (Bonneville). The report summarizes an analysis of the persistence over time of energy savings attributable to Bonneville's Residential Weatherization Program.

Past evaluations of Bonneville's Residential Weatherization Program have indicated that the net annual energy savings attributable to the program decline over time. In the most recent evaluation of the long-term impacts of the l983 program (i.e., the cohort of households participating in 1993) savings were estimated to decline from 1,982 kWh in 1984 to 1,439 kWh in 1989.

The purpose of this study was to determine if the observed decline in estimated savings for the 1983 cohort continued through 1992, the latest year for which data are available, and to examine several hypotheses concerning the observed decline in estimated savings. To achieve this purpose, the weather-normalized billing data for the years 1983 - 1989 for samples of 1983 program participants and a comparison group, which were used in past studies of the long-term impacts of the program, were combined with similar data for 1991 and 1992 for the same samples of households. These data were analyzed statistically to determine the extent to which the savings decline continued beyond 1989, and to test the statistical significance of the decline in estimated savings. In addition, characteristics data collected in 1983 and 1992 from sub-samples of the samples for whom billing data are available were analyzed statistically, in an attempt to identify potential "causes" of the observed decline in estimated savings.

The study yielded the following results:

• Statistical analysis of the weather-normalized billing data produced the following point estimates of annual energy savings (in kWh), with standard errors in parentheses:
  • 1984 -- 2.075 (313)
  • 1985 -- 1,963 (380)
  • 1986 -- 1,440 (409)
  • 1987 -- 1,699 (420)
  • 1988 -- 1,583 (427)
  • 1989 -- 1,104 (477)
  • 1991 -- 1,301 (522)
  • 1992 -- 1,708 (569)
• A variety of tests were conducted of the differences in savings between years. None of the tests supported rejection at the 90 percent confidence level of the null hypothesis that the savings have been constant over time. In other words, there is no statistical evidence of changes (i.e., either degradation or improvement) over time in the savings attributable to the 1983 program.
• Survey data for 83 of the 390 households were analyzed to determine if the proportions of participants and non-participants who took energy conservation actions between 1983 and 1992 differed. Of the twelve actions considered, a greater proportion of non-participants than participants installed one measure (wall insulation), while greater proportions of participants than non-participants installed two measures (water heater wraps and automatic setback thermostats). The differences for the other nine actions were not statistically significant. Engineering estimates of the savings associated with the actions were combined with the estimated proportions of participants and non-participants who took the actions between 1983 and 1992. These calculations showed that participants should have saved more than non-participants as a result of actions taken between 1983 and 1992, although the difference was, again not statistically different.
• The characteristics data for these 83 households were analyzed to determine if differences between participants and non-participants in 1983 - 1992 changes in characteristics not affected by the program could be spuriously causing estimated savings to decline. Of the nine characteristics considered, the differential rate of change for only one was statistically significant; the number of occupants in participant households decreased more than in non-participant households. Because this phenomenon should cause savings estimates to increase, not decrease, over time, there is no statistical evidence that differential rates of change in important characteristics are causing estimated savings to decline over time.

These results lead to the conclusion that there have been no changes (either degradation or improvement) in the annual energy savings attributable to the 1983 Residential Weatherization Program.