Home Energy Rebates Need Installed-Savings Accountability, Not Reservation Counts Alone
Home Energy Rebates are easy to overreport because awards, launches, reservations, installations, modeled savings, measured savings, payments, consumer protection, and equity outcomes sound like one delivery pipeline but prove different claims. This paper synthesizes Public Law 117-169, DOE program requirements, consumer-protection guidance, measured-path guidance, DOE/NREL software verification procedures, California, Wisconsin, and Oregon program pages, ACEEE evaluation and energy-burden evidence, and NBER weatherization evaluation evidence. The contribution is an Installed-Savings Accountability Chain: public reporting should stop at the strongest stage supported by evidence. A funding award is not household access; a launch is not an installation; a reservation is not completed work; modeled savings are not verified savings; a payment is not necessarily a resolved consumer or equity outcome. The chain helps agencies report progress without turning future household benefits into current counts.
Introduction
The Inflation Reduction Act created large household energy rebate programs, but large appropriations create a reporting trap. A public dashboard can say money was authorized, awarded, launched, reserved, installed, paid, modeled, measured, or equitably delivered. Those are not interchangeable verbs. A 14000 dollar electrification cap and an 8000 dollar HOMES pathway are useful to consumers, but the existence of those caps does not prove that a heat pump was installed, inspected, paid for, or produced lower bills [[cite:publicLaw117169,doeProgramPage]].
This paper asks how Home Energy Rebate programs should report progress so award totals and reservation counts are not mistaken for installed savings. The answer is a weakest-proven-stage rule: report the strongest claim for which evidence exists, and stop there. Award evidence supports an award claim. Reservation evidence supports a reservation claim. Installation evidence supports an installation claim. Modeled savings support a prediction claim. Measured post-installation utility data support a verified-savings claim [[cite:doeProgramRequirements,doeMeasuredPathway]].
The issue is timely because public pages already show different states at different stages. California's page shows a fully reserved single-family electrification lane and a HOMES Pay for Performance design that will depend on meter-measured savings. Wisconsin describes operational programs and payment processing. Oregon describes an award but no currently available rebates because launch approval and operational prerequisites remain incomplete [[cite:californiaCec,wisconsinFocus,oregonOdoe]]. A single national progress phrase cannot safely cover all three cases.
Methods
I conducted a policy-accountability synthesis on 2026-06-27. The research set included six AlexandrAI graph searches to avoid duplicating prior archive work and twelve external searches across statute, DOE guidance, state program pages, evaluation literature, equity research, and measured-savings evidence. Forty-plus sources were screened, and twelve were fully read for citation-level use. The source set intentionally combines official rules, live state examples, and evaluation caution rather than relying on program announcements alone.
The primary sources were Public Law 117-169, DOE's current program page, DOE program requirements, DOE consumer-protection required elements, DOE measured-path guidance, DOE/NREL measured-path software verification procedures, and California, Wisconsin, and Oregon program pages [[cite:publicLaw117169,doeProgramPage,doeProgramRequirements,doeConsumerProtection,doeMeasuredPathway,doeSoftwareVerification,californiaCec,wisconsinFocus,oregonOdoe]]. The evaluation and equity sources were an ACEEE/DOE IRA evaluation paper, an ACEEE household energy burden brief, and the NBER summary of Fowlie, Greenstone, and Wolfram's weatherization evaluation [[cite:aceeeIraLessons,aceeeEnergyBurden,fowlieNber]].
The analysis used a stage-discipline method. For each source, I asked what public claim the evidence actually proves and what stronger claim it does not prove. For example, California's fully reserved signal proves demand and reservation pressure, not completed installations. Oregon's award proves funding status, not consumer availability. A modeled savings estimate proves a predicted engineering or approved-model result, not actual utility-bill savings. Table 1 summarizes the source families and the reporting risks they reduce.
Program Design
The statutory design already contains multiple accountability stages. Section 50121 appropriates 4.3 billion dollars for HOMES rebate programs and requires approved procedures for modeled savings, open-source advanced measurement and verification software for measured performance rebates, quality monitoring certificates, and anti-duplication for the same upgrade [[cite:publicLaw117169]]. Section 50122 appropriates 4.275 billion dollars for state high-efficiency electric home rebate programs and 225 million dollars for Tribes, requires income verification and point-of-sale delivery, caps multiple rebates at 14000 dollars, and prohibits stacking federal rebates for the same qualified project [[cite:publicLaw117169]].
DOE's program page describes the two program families for consumers: HOMES can provide rebates for whole-home energy upgrades, including up to 8000 dollars based on modeled savings levels with a minimum 20 percent savings threshold; the electrification program can provide up to 14000 dollars for efficient upgrades at point of sale through retail outlets or contractors [[cite:doeProgramPage]]. That page also states that rebates are available in select states and directs consumers to state or territory energy offices for current status and eligibility. The status referral is not administrative trivia; it is evidence that national eligibility language does not equal local availability.
DOE's program requirements sharpen the split. The requirements say the two IRA sections together authorize 8.8 billion dollars for U.S. households. They distinguish modeled Home Efficiency Rebates, which use calibrated home energy models and a 20 percent predicted savings threshold, from measured rebates, which use DOE-approved open-source M&V to measure post-installation savings and require at least 15 percent savings [[cite:doeProgramRequirements]]. The same document requires states to collect required data during the award period and includes duplicate-rebate protections for the same measure [[cite:doeProgramRequirements]].
State Examples
State pages show why progress should be staged. California reports that, as of February 24, 2026, HEEHRA single-family retrofit rebates were fully reserved statewide; unapproved reservation requests were waitlisted, and no new income verification requests were accepted at that time. California also says projects need an approved reservation before purchase or installation and are not retroactive [[cite:californiaCec]]. This is a strong reservation-stage signal, but it is not an installation-stage or measured-savings signal.
California also illustrates the difference between award, program design, and availability. The state reports a 590 million dollar DOE Home Energy Rebates award, including 290 million dollars for HEEHRA and 291 million dollars for HOMES. But the same page says HOMES rebates are not yet available and describes a Pay for Performance design in which payments to aggregators are based on energy savings measured at the meter for properties completing HOMES projects [[cite:californiaCec]]. A dashboard that reports "590 million dollars awarded" and "HOMES measured savings" without the intervening availability and completion states would overstate current household outcomes.
Wisconsin and Oregon provide the contrast cases. Wisconsin says it was allocated 149 million dollars, that Focus on Energy was chosen to deliver the programs, and that both HOMES and HEAR are now available. Its FAQ states that the programs are operational and that incentive or rebate payments are processed as received [[cite:wisconsinFocus]]. Oregon says it was awarded over 113 million dollars but that no rebates are currently available; launch depends on DOE approval, participant and contractor agreements, and a program portal, and Oregon reports that DOE suspended launch approvals while reviewing priorities and requirements [[cite:oregonOdoe]].
Savings Evidence
Savings claims require the most careful labels. DOE's requirements permit modeled and measured HOMES pathways. A modeled path can be a legitimate basis for a rebate when it uses approved calibrated models and meets the predicted-savings threshold, but it remains a prediction until actual energy-use evidence arrives [[cite:doeProgramRequirements]]. The NBER summary of Fowlie, Greenstone, and Wolfram's weatherization field experiment is a cautionary counterexample: in that study, model-projected savings were roughly 2.5 times actual savings and upfront investment costs were about twice actual energy savings [[cite:fowlieNber]]. The point is not that IRA rebates will repeat that result; the point is that modeled savings should be labeled as modeled.
The measured pathway is stronger but operationally more demanding. DOE says measured-path rebates are calculated on actual verified savings using pre- and post-installation utility consumption data and DOE-approved open-source M&V [[cite:doeMeasuredPathway]]. DOE also says measured-path programs need substantial data windows: if measuring less than twelve months after installation, they should include at least one peak energy season and both peaks in dual-peaking climates. The same guidance notes that states cannot use DOE funds for measured-path rebates until actual savings are verified, which can create carry risk and reimbursement delays, often 12 to 15 months after project completion [[cite:doeMeasuredPathway]].
DOE/NREL software verification procedures add a second accountability layer. The procedures address open-source M&V software that documents weather-normalized energy use before and after installation, with checks for alignment, transparency, consistency, accuracy, weather normalization, regression model selection, real-world validation, and missing-data handling [[cite:doeSoftwareVerification]]. This supports measured-savings credibility, but software verification is not a substitute for reporting the project count, data window, baseline, weather normalization method, and savings distribution.
Consumer Protection and Quality
Consumer protection is a post-award and post-installation evidence problem. DOE requires consumer protection plans for rebate programs and includes consumer feedback systems, complaint systems, and satisfaction surveys after project completion [[cite:doeConsumerProtection]]. It also calls for data review with validation controls, post-installation certificates, post-installation photos of major upgrades or equipment, and records on quality-control inspections and sampling rates [[cite:doeConsumerProtection]].
Those requirements imply that an installation count is not the final accountability unit. A stronger public report would say how many completed projects have certificates, how many have required photos, how many were inspected remotely or onsite, how many failed quality checks, how many complaints were opened, how many complaints were resolved, and how many corrections were completed. DOE's sample language includes independent onsite post-installation inspections for the first projects of new contractors and continuing inspection samples when no issues are found [[cite:doeConsumerProtection]].
The contractor channel matters for delivery. California tells homeowners to work with certified and HEEHRA-trained contractors and requires approved reservations before work begins. Wisconsin makes registered contractors one of the core participation steps. Oregon names participant and contractor agreements as a launch dependency [[cite:californiaCec,wisconsinFocus,oregonOdoe]]. Public progress reports should therefore count active delivery capacity separately from household reservations and installations.
Equity and Energy Burden
The equity rationale is strong, but eligibility language is not an equity outcome. ACEEE reports that one-quarter of U.S. households have high energy burdens, 13 percent have severe burdens, and 67 percent of low-income households have high energy burdens; among low-income households with high burdens, 60 percent have severe burdens [[cite:aceeeEnergyBurden]]. The brief also says weatherization can reduce low-income household energy burdens by about 25 percent [[cite:aceeeEnergyBurden]]. Those facts justify targeted rebates, but they do not prove that a particular rebate reached high-burden households or reduced bills.
DOE requirements reserve portions of funding for low-income and low-income multifamily households and connect eligible rebate levels to income and project cost [[cite:doeProgramRequirements]]. The statute likewise makes income verification central to the electrification program and provides larger HOMES rebate amounts for low- or moderate-income households [[cite:publicLaw117169]]. The reporting implication is straightforward: count verified low-income households or units, completed installations for those households, net customer cost, and bill or burden evidence where available.
Equity reports should also handle multifamily delivery carefully. California's page distinguishes single-family and multifamily HEEHRA processes and describes per-unit multifamily rebates; Wisconsin's page says multifamily projects involve income qualification, energy assessment, reservation, installation, and payment request steps [[cite:californiaCec,wisconsinFocus]]. A building-level award does not automatically prove tenant-level benefit. Reports should say whether the benefit was applied to tenant units, common areas, owner costs, or measured household bills.
Installed-Savings Accountability Chain
Table 3 gives the paper's main contribution: an Installed-Savings Accountability Chain for Home Energy Rebates. It is intentionally sequential because the public claim should stop at the last documented stage. A state with an award and no launch should not claim installations. A launched program with waitlisted reservations should not claim completed projects. A completed project with a modeled estimate should not claim measured savings. A measured savings record should not automatically claim complaint resolution or equity impact.
The chain also avoids punishing programs for being honest about early stages. Oregon's award-without-launch status is not a failure of household savings; it is a launch-stage status. California's reservation pressure is not proof of verified savings; it is evidence of demand and budget pressure. Wisconsin's operational status is meaningful, but outcome reporting still needs installed project, payment, and savings data [[cite:oregonOdoe,californiaCec,wisconsinFocus]].
Reporting Rules
Public reporting should use plain labels that match the evidence. Award amount, launch status, income verification, reservation, installation, modeled savings, measured savings, and consumer complaint activity can all be useful. The problem begins when one metric is allowed to stand in for another. Table 4 translates common metrics into safer labels and unsafe inferences.
The strongest dashboards would publish counts by stage and by pathway. For modeled HOMES, they would report modeled savings separately from later measured savings, if measured follow-up is available. For measured HOMES, they would report the number of projects awaiting post-installation data, the number with complete data windows, the number with verified savings, the distribution of savings, and the number reimbursed from federal funds. For electrification rebates, they would report income verification, point-of-sale or contractor discount, installation, quality-control evidence, and complaint resolution.
Limits
First, the state pages are date-bound. California, Wisconsin, and Oregon statuses were read on 2026-06-27 and can change as reservations reopen, launch approvals arrive, or program rules shift [[cite:californiaCec,wisconsinFocus,oregonOdoe]]. The examples are therefore used to illustrate stage mismatch, not to maintain a national tracker.
Second, the Fowlie, Greenstone, and Wolfram evidence is not an evaluation of IRA Home Energy Rebates. It is weatherization evidence used to discipline claims about modeled savings. IRA program design, consumer protections, software procedures, and data requirements may produce different results, but the older field experiment is still relevant as a warning against treating projected savings as realized savings [[cite:fowlieNber,doeSoftwareVerification,aceeeIraLessons]].
Third, verified savings are not the only policy objective. Comfort, health, resilience, indoor air quality, affordability, contractor workforce development, and emissions can matter. This paper focuses on installed-savings accountability because rebate programs are often publicly justified through household savings, but a complete evaluation portfolio should include broader outcomes when data are available [[cite:doeProgramPage,aceeeIraLessons,aceeeEnergyBurden]].
Fourth, the chain is a reporting discipline, not a legal opinion. Statute and DOE guidance define program duties; this paper organizes the evidence so public claims do not overrun it [[cite:publicLaw117169,doeProgramRequirements,doeConsumerProtection]].
Conclusion
Home Energy Rebates need installed-savings accountability, not reservation counts alone. The core rule is simple: a public claim should name the stage it proves. Awarded dollars, launch approvals, income verifications, reservations, installations, modeled savings, measured savings, payments, consumer-protection outcomes, and equity outcomes are all useful, but each answers a different question.
The evidence base supports a weakest-proven-stage reporting model. The statute and DOE requirements define authority, pathways, income rules, data, and duplicate checks. DOE consumer-protection guidance defines post-installation quality evidence. DOE measured-path and software guidance define how verified savings can be calculated and why it takes time. State pages show that availability, reservation pressure, operational processing, and launch delay can coexist across the country. Evaluation and energy-burden evidence explain why prediction, verification, and equity measurement should stay distinct.
The practical output is the Installed-Savings Accountability Chain. It lets public agencies and program administrators report real progress without collapsing future benefits into today's counts. A reservation is a reservation. An installation is an installation. A model is a model. Verified savings are verified savings. The public can understand all four, but only if reports stop treating them as the same thing.