The energy use impacts presented here reflect the difference in usage changes between the matched R-PACE and comparison group. In aggregate, the average change in electricity usage for comparison group households was less than 1.5% for each measure category. Separate estimates of energy impacts for R-PACE and comparison households are available upon request.
3.1. Usage Impacts by Utility
Table 3 shows our estimates of the average reduction in grid electricity and gas use for R-PACE households in each measure category for each California IOU. (We use the term “reduction in grid electricity use” here because it includes both electricity savings from energy efficiency measures and avoided grid electricity consumption due to on-site PV generation. We study the change in household load net of PV generation, and (for PV homes) cannot separate avoided grid consumption from efficiency savings, since both reduce net load.) We also present confidence intervals in
Table 3. To generate confidence intervals by utility and measure category, we first calculate the confidence interval for each R-PACE household using the method specified in ASHRAE Guideline 14 [
22]. Next, we calculate a confidence interval for the difference in savings between the R-PACE and comparison household, assuming independence in the errors from each household [
23]. Finally, we aggregate the confidence intervals for these differences to the desired level (e.g., energy efficiency projects in PG&E households) again assuming independence of errors.
For R-PACE households that only installed efficiency measures, electricity savings ranged between 3 and 5% for households located in SCE and PG&E service territory while average electricity usage increased slightly for SDG&E R-PACE customers (0.4%). In analyzing savings for households that installed energy efficiency measures, it is important to note that the vast majority of R-PACE energy efficiency projects affect only space conditioning energy usage, and that the majority of California households are heated by natural gas. (86% of California single-family homes used natural gas as the main source of space heating and only 2% used electricity, 8% used other fuels, and 4% were not heated in 2009, the most recent data available [
24].) Moreover, many homes in coastal areas of California enjoy very mild summers and do not have air conditioning. (In 2009 43% of California households were not air-conditioned [
25]. Air conditioning saturation varied dramatically by climate zone, from less than 20% in the San Francisco Bay Area, to around 50% in coastal urban areas in southern California, to over 90% in many inland areas [
24]. In particular, SDG&E’s service territory is largely coastal, making it difficult to save electricity via upgrades to space conditioning equipment. The negative electricity savings in SDG&E territory may also reflect installation of air conditioning equipment in some households that previously did not have air conditioning.
Average gas usage declined by 2%, 3% and 7% in R-PACE households that are located in SoCal Gas, SDG&E and PG&E territories, respectively. PG&E’s service territory is in northern and central California, where winters—while still mild by national standards—are cooler. Consistent with this, R-PACE energy efficiency projects save more gas usage on average in PG&E territory.
Our metered data analysis shows much larger reductions in grid electricity use from solar PV than from energy efficiency projects. For R-PACE households that installed solar PV, reductions in grid electricity use ranged between 66 and 83% of electricity usage across the three IOUs. Not surprisingly, average reductions in grid electricity use were quite high for these households (7300–8800 kWh per year). SDG&E solar projects generate more electricity on a household percentage basis than those in the other utilities, likely due to a combination of high insolation and lower average pre-project household usage due to low cooling loads. For these homes that installed solar PV, gas usage increased modestly (3%) for homes located in SoCal Gas and SDG&E service territory after installation and declined modestly (2%) for homes located in PG&E service territory. In general, solar projects have little impact on gas consumption, though the impacts vary across utilities in a manner for which we have no immediate explanation.
In PG&E and SDG&E service territories, R-PACE households that installed both solar PV and energy efficiency measures saw average reductions in grid electricity use of similar magnitude to homes that just installed solar PV. However, in SCE territory, reductions in grid electricity use were approximately 10% lower in homes that installed solar PV and efficiency measures compared to PV-only households (57% vs. 66% savings). This result is counter-intuitive and surprising. Given that we find this result only in SCE territory, it may be explained by particular characteristics of these projects. All the joint EE/PV projects in SCE territory were installed by a single R-PACE program. While we cannot test this given our data, it is possible that many of these 368 projects were done by a single R-PACE contractor that may have installed smaller-than-typical PV systems. For this group of homes that installed both solar PV and efficiency measures, we observe reductions in gas use at households in all three service territories, ranging from 2.6% at PG&E to 3% at SDG&E and 5.3% at SoCal Gas.
Figure 2 presents California’s climate zones, and
Figure 3 shows our estimates of the average R-PACE project impacts by California climate zone for households that installed various types of energy efficiency measures. Percentage electric savings from energy efficiency projects are highest (4 to 13%) in non-coastal climate zones (4, 12, 13, 14, and 15–Central Valley, Sierra Nevada region, and inland desert areas) which experience warmer summers and therefore higher cooling loads.
Note that the savings for projects in various climate zones include some projects that installed new central AC in households that had previously lacked such units, or installed a gas furnace in a home that had previously used an electric space heater or wood stove. We might expect that a greater share of R-PACE projects would include installation of new HVAC technologies, rather than replacement of existing units, in climates where heating or cooling demand is low and therefore many homes would not have previously had these technologies. Such effects likely explain the negative savings for electricity in climate zone 7 (a temperate coastal Southern California climate zone where air conditioning penetration is low) and for gas in climate zone 15 (a hot desert climate where minimal heating is necessary).
Gas savings are highest in more northern or more inland zones (3, 4, 12, 13, and 14—the greater Bay Area, northern Central Valley, and Sierras). Zone 3 is coastal but far enough north to require more significant space heating in the winter (4% savings); zones 4, 12, 13, and 14 are far enough inland that they also have significant heating demand (6 to 11% average gas savings). Zone 15 is essentially a desert environment, warm enough to not require much space heating despite being inland.
3.3. HVAC Project Savings
The measure data we received from PACE providers generally do not establish whether the HVAC projects in our data installed new heating or air conditioning equipment, both, or neither (e.g., household installed new duct sealing or new whole house fans). Moreover, an unknown fraction of projects that installed HVAC equipment represent households that did not have space cooling equipment (A/C) or central heating systems prior to installation. This may be a larger issue in California’s mild climate than in most parts of the U.S. (As of 2009, approximately 40% of California single-family households did not have air conditioning and approximately 14% did not have heating systems, leaving room for penetration of new systems. Nationwide, only 4% of households are not heated and only 17% lack air conditioning [
24,
25]. In the case of heating, some R-PACE households may be installing new gas- or electric- heating equipment that is replacing or supplementing wood or propane heating.)
As such, the average result for each fuel is based on an agglomeration of projects expected to reduce usage of that fuel (e.g., a furnace replacement’s impact on gas usage); projects expected to have little impact on that fuel (e.g., an air conditioning project’s impact on gas usage); and projects expected to increase usage of that fuel (e.g., a new gas-fired furnace installed in a house that did not have gas heating equipment). Thus, our HVAC results should not be interpreted to mean that the average air conditioning replacement reduced electricity use by only 4%, or that the average gas furnace replacement reduced gas usage by only 2% (see
Figure 4).
Figure 5 and
Figure 6 show distributions of impact estimates for electricity and gas usage for R-PACE project and comparison households that installed HVAC only measures. Note that households with negative savings in
Figure 5 or
Figure 6 increased electricity or gas usage, respectively, after the R-PACE project was installed. Households with positive savings decreased their electricity or gas usage. The distribution of “impacts” for comparison households (shown in gray shading)—which did not implement projects in the time window analyzed—gives a sense of the natural variability of year-on-year household energy usage due to non-project factors, even after we have normalized for the effects of weather.
Figure 5 and
Figure 6 show that a greater share of comparison HVAC households than R-PACE HVAC households had energy impacts within 10% of baseline usage than project households. A total of 67% of comparison households had changes in usage between 𢄬10% and 10% of baseline usage, whereas only 50% of R-PACE households had usage changes in that range. A greater share of R-PACE HVAC households than comparison households decreased their consumption. A total of 27% of R-PACE households saved more than 10% of baseline electricity usage as compared to 18% of comparison households. However, a higher percentage of R-PACE households also increased usage after the R-PACE project. A total of 22% of R-PACE households increased electricity usage by 15% or more relative to their baseline usage (“negative savings”) compared to only 15% of comparison households. We find similar patterns among HVAC-only households in our gas analysis.
The greater share of R-PACE households on the far left side of the distributions suggests that these projects installed a new furnace or air conditioning equipment and did not have this type of equipment prior to the project. We do not find higher shares of project households on the left-hand side of the distribution for other measure categories. This finding is consistent with R-PACE program rules in California, which allow for installations of new HVAC equipment irrespective of prior HVAC equipment practices.
To gain greater clarity on HVAC impacts, we use an algorithmic method to infer which projects may be installing new HVAC equipment and did not have such equipment prior to the project. We focus on seasons and times of day that are most likely to reveal the actual context and baseline conditions underlying projects that installed HVAC equipment. Specifically, we examine each HVAC-only project’s electricity usage impact from noon to midnight in the summer months of July through September (where we would expect air conditioning to be used). If weather-normalized electricity use increases by more than 15% during these hours, we infer that the project includes a new air conditioning installation. If weather-normalized electricity use decreases by more than 10% during these hours, we infer that the project includes a replacement of existing air conditioning equipment. Projects with usage changes that fall between these values are ambiguous or may not have included a cooling-related efficiency measure at all (e.g., only measures affecting gas usage were installed). For natural gas, we apply similar logic to all usage during the winter months of December to February. An increase in gas consumption of 25% after the R-PACE project was completed is our threshold for inferring installation of a new gas furnace in a household that previously did not have gas central heating equipment. If gas usage for HVAC-only projects decreased by at least 10%, then we assume that a new gas furnace replaced existing gas-fired equipment. We define these cut-off points from the data underlying
Figure 3 and
Figure 4. We choose cut-off points near the parts of the savings distributions where the share of project households begins to be greater than the share of comparison households, suggesting that values at and beyond these cut-off points are more likely to reflect project effects than other factors.
Figure 3 and
Figure 4 show full-year results. By focusing on seasonal time periods where HVAC effects ought to be more pronounced, we are more confident that the observed effects are related to installed measures in projects. For example, we guard against mis-identifying electric heat impacts as air conditioning impacts. Note that we restrict our inquiry to air conditioning and gas heat, so are not considering electric heat (such as via an air-source heat pump).
Figure 7 shows annual savings estimates for HVAC-only projects categorized in this manner, along with the sample sizes of each inferred category. The data suggest that many R-PACE HVAC-only projects may be installations of new equipment rather than one-for-one replacement of existing types of HVAC equipment. Among projects we classify as A/C equipment replacement, average annual electric savings are 15%. Among projects we classify as heating equipment replacement (e.g., furnaces), average annual gas savings are 23%. For the projects that we classify as having installed new equipment in households that did not previously have that equipment, usage increased significantly—by 16% on average for inferred cooling equipment installs and by 40% on average for inferred heating equipment installs (e.g., furnaces). In general, the impact of inferred new cooling equipment projects on gas use and of inferred gas heating projects on electricity use are small (1–2%), as would be expected.
We view this effort to characterize the underlying baseline condition of HVAC projects as exploratory, yielding approximate distributions, because our algorithm likely mischaracterizes some projects.