Story ideas: Heat pumps

Central conclusion Time-of-use or heat-pump electricity tariffs make heat pumps cheaper to run

qualified qualify review qualified (qualify): 2 supporting (2 independent, strongest 0.7), 2 qualifying.

RMI 2018, the economics of electrifying buildings: In RMI's models a tariff alone does not cut costs: a standard heat pump on Houston's Free Nights rate costs more than on a flat rate, and only flexible control on steep tariffs (3:1 TOU, Free Nights) saved about $1,000-2,000 over 15 years, while mild default TOU gave too little.

Wilson 2024 (NREL and LBNL), heat pumps for all?: Where time-of-use rates exist there may be an opportunity to shift heat pump use off-peak and improve bill savings; not analysed.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff right 70%, direct lines 1898-1906Staggered-adoption study of 6,631 UK households: switching to Octopus's Cosy heat pump time-of-use tariff saved about £318/yr (about 18% of the electricity bill), £78 from the tariff structure alone and £240 from shifting load; savings are vs the same supplier's flat tariff, and the authors are linked to Octopus.
• Supports Mass Save 2026, air-source heat pump rebates right 35%, direct lines 30-33Mass Save: eligible customers can get a Seasonal Heat Pump Rate with lower electric rates November through April 'to help make heat pump operation more affordable'; time-of-use rates are only described as future options.

• Qualifies RMI 2018, the economics of electrifying buildings partly 50%, direct lines 748-755In RMI's models a tariff alone does not cut costs: a standard heat pump on Houston's Free Nights rate costs more than on a flat rate, and only flexible control on steep tariffs (3:1 TOU, Free Nights) saved about $1,000-2,000 over 15 years, while mild default TOU gave too little.
• Qualifies Wilson 2024 (NREL and LBNL), heat pumps for all? partly 20%, direct lines 754-757Where time-of-use rates exist there may be an opportunity to shift heat pump use off-peak and improve bill savings; not analysed.

Finding 1 (evidence) The evidence does not support "Heating with a heat pump costs less than heating with a natural gas furnace"

qualified qualify review qualified (qualify): 5 supporting (5 independent, strongest 0.55), 3 qualifying.

Wilson 2024 (NREL and LBNL), heat pumps for all?: For gas-heated homes with AC, the share with lower bills ranges from 38% (minimum-efficiency, median -$70/yr) to 99% (cold-climate high-efficiency, median $380/yr), and only 5-58% for gas homes without AC; it holds only for efficient equipment.

EIA, state residential electricity and gas prices: At 2024 average prices a seasonal COP of 2.5 beats a 95% furnace in only 11 states; 30 states need COP 3 or more, so it depends on location and efficiency.

Canadian Climate Institute 2023, heat pumps pay off: Canadian lifetime-cost modelling finds heat pumps beat gas heating plus AC in nearly all cases in Montreal, Halifax and Vancouver and usually in Toronto with gas backup, but not in many Edmonton cases; the comparison also counts cooling and equipment.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff wrong 55%, direct lines 1850-1855In the UK at 2024 prices on a standard flat tariff, heat pumps cost about £190 more per year to run than the gas boilers they replaced (boilers, not furnaces); only a heat pump time-of-use tariff reverses this.
• Supports Edwards 2024, inequities in heat pump bill savings wrong 50%, direct lines 56-61Edwards says heat pumps are cost-effective in many places but can cost more to operate than gas furnaces, especially in colder regions, so the blanket claim fails.
• Supports RMI 2018, the economics of electrifying buildings wrong 50%, direct lines 158-161In RMI's gas retrofits, heat pump households pay more for energy in Chicago and Providence and save too little to recover the extra capital in Houston and Oakland.
• Supports Cadmus 2016, ductless mini-split evaluation (MA and RI) wrong 50%, direct lines 962-966At 2015-16 Massachusetts/Rhode Island prices, the break-even temperature against an 80%-efficient gas system was above 70 F (above 58 F for cold-climate units), so a mini-split was essentially never cheaper to run than gas; the test units were 2012-14 models (a modelled HSPF-13 unit moves the break-even to 28-40 F).
• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results wrong 20%, direct lines 1666-1673Survey comments from participants who had gas repeatedly said bills were higher than with natural gas; this is anecdotal (no billing analysis, small affiliated sample).

• Qualifies Wilson 2024 (NREL and LBNL), heat pumps for all? partly 55%, direct lines 409-415For gas-heated homes with AC, the share with lower bills ranges from 38% (minimum-efficiency, median -$70/yr) to 99% (cold-climate high-efficiency, median $380/yr), and only 5-58% for gas homes without AC; it holds only for efficient equipment.
• Qualifies EIA, state residential electricity and gas prices partly 40%, direct lines 191-193At 2024 average prices a seasonal COP of 2.5 beats a 95% furnace in only 11 states; 30 states need COP 3 or more, so it depends on location and efficiency.
• Qualifies Canadian Climate Institute 2023, heat pumps pay off partly 40%, direct lines 1133-1150Canadian lifetime-cost modelling finds heat pumps beat gas heating plus AC in nearly all cases in Montreal, Halifax and Vancouver and usually in Toronto with gas backup, but not in many Edmonton cases; the comparison also counts cooling and equipment.

Finding 2 (evidence) Whether a heat pump is cheaper to run than gas depends mainly on local electricity and gas prices

qualified qualify review qualified (qualify): 4 supporting (4 independent, strongest 0.55), 3 qualifying.

Edwards 2024, inequities in heat pump bill savings: Edwards finds large regional differences in heat pump operating costs that depend on local climate as well as the price of electricity relative to gas, so prices are a main driver but not the only one.

Wilson 2024 (NREL and LBNL), heat pumps for all?: Bill savings are very sensitive to prices (2019 elec-to-gas ratio 4.1 cuts gas homes with positive savings from 32-92% to 22-83%), but the range itself shows equipment efficiency matters as much as prices, so 'mainly prices' overstates.

EIA, state residential electricity and gas prices: EIA-derived break-even seasonal COP equals the electricity-to-gas price ratio times furnace AFUE, so the outcome depends on local prices and on the heat pump's seasonal efficiency together.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Less 2021 (LBNL), cost of decarbonization retrofits right 55%, direct lines 164-171LBNL retrofit database (1,739 projects): about 70 electrification projects in states with high electricity costs raised household energy costs, and the authors conclude savings must be judged state by state given the variation in gas and electricity prices.
• Supports Canadian Climate Institute 2023, heat pumps pay off right 50%, direct lines 1060-1067Advocacy-adjacent Canadian lifetime-cost model (five cities): regional energy prices and climate are the biggest drivers of whether heat pumps beat gas, with prices and climate both mattering (e.g. low-gas-price, cold Edmonton vs Montreal and Halifax).
• Supports Cadmus 2016, ductless mini-split evaluation (MA and RI) right 45%, direct lines 955-966The report maps electricity and fuel prices against the outdoor temperature above which a mini-split is cheaper than fuel heat; the answer turns on the price pair and on the unit's efficiency at that temperature, and at 2015-16 New England prices gas won almost everywhere.
• Supports Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff right 45%, direct lines 1038-1044In the UK, total energy use fell 40% after switching from a gas boiler, yet running costs came out slightly higher because electricity costs more per kWh than gas, which shows prices decide the comparison (one country, one set of prices).

• Qualifies Edwards 2024, inequities in heat pump bill savings partly 55%, direct lines 855-857Edwards finds large regional differences in heat pump operating costs that depend on local climate as well as the price of electricity relative to gas, so prices are a main driver but not the only one.
• Qualifies Wilson 2024 (NREL and LBNL), heat pumps for all? partly 50%, direct lines 538-553Bill savings are very sensitive to prices (2019 elec-to-gas ratio 4.1 cuts gas homes with positive savings from 32-92% to 22-83%), but the range itself shows equipment efficiency matters as much as prices, so 'mainly prices' overstates.
• Qualifies EIA, state residential electricity and gas prices partly 40%, direct lines 119-125EIA-derived break-even seasonal COP equals the electricity-to-gas price ratio times furnace AFUE, so the outcome depends on local prices and on the heat pump's seasonal efficiency together.

Finding 3 (evidence) A heat pump beats gas on running cost only when its COP exceeds the electricity-to-gas price ratio

qualified qualify review qualified (qualify): 1 supporting (1 independent, strongest 0.4), 3 qualifying.

EIA, state residential electricity and gas prices: The break-even seasonal COP is the electricity-to-gas price ratio times the furnace AFUE (US 2024: 3.45 x 0.95 = 3.28), so the threshold is slightly below the raw price ratio.

Edwards 2024, inequities in heat pump bill savings: Edwards computes savings as fuel price times load divided by furnace efficiency, minus electricity price times load divided by heat pump COP, so the break-even COP is the price ratio multiplied by furnace efficiency, not the bare price ratio.

Cadmus 2016, ductless mini-split evaluation (MA and RI): The crossover equation sets the heat pump's cost per delivered Btu (electricity price over COP) equal to the gas cost (gas price over furnace efficiency), so the threshold is the price ratio times furnace efficiency, not the bare price ratio.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff right 40%, direct lines 1071-1075In the UK the electricity-to-gas price ratio was about 4.1 while the heat pump used roughly a third of the gas energy it displaced, and running costs came out £190/yr higher, consistent with the ratio rule.

• Qualifies EIA, state residential electricity and gas prices partly 60%, direct lines 123-131The break-even seasonal COP is the electricity-to-gas price ratio times the furnace AFUE (US 2024: 3.45 x 0.95 = 3.28), so the threshold is slightly below the raw price ratio.
• Qualifies Edwards 2024, inequities in heat pump bill savings partly 45%, direct lines 653-672Edwards computes savings as fuel price times load divided by furnace efficiency, minus electricity price times load divided by heat pump COP, so the break-even COP is the price ratio multiplied by furnace efficiency, not the bare price ratio.
• Qualifies Cadmus 2016, ductless mini-split evaluation (MA and RI) partly 40%, direct lines 3372-3384The crossover equation sets the heat pump's cost per delivered Btu (electricity price over COP) equal to the gas cost (gas price over furnace efficiency), so the threshold is the price ratio times furnace efficiency, not the bare price ratio.

Finding 4 (evidence) Most US households would see lower energy bills with a heat pump

supported approved supported (approved): 2 supporting (2 independent, strongest 0.55).

• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 55%, direct (reviewed) lines 362-366Bills rise in 39% of homes with minimum-efficiency units and 5% with cold-climate high-efficiency units, so a majority save in every scenario, though fewer than a third of gas homes do with minimum units.
• Supports Edwards 2024, inequities in heat pump bill savings right 50%, direct lines 290-294Edwards' Monte Carlo model at 2020 prices finds 53% of households in its dataset would clearly see lower energy bills, against 19% higher and 28% unclear, a thin majority under conservative single-stage assumptions.

Finding 5 (evidence) Switching from gas to a heat pump raises the household's energy bills

qualified qualify review qualified (qualify): no direct support, 5 qualifying, 1 context only.

Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff: UK households switching from gas boilers see running costs about £190/yr higher on a standard tariff (about £83 after dropping the gas standing charge), so bills do rise modestly; but the same paper finds a heat pump time-of-use tariff saves about £318/yr, enough to reverse the rise.

Wilson 2024 (NREL and LBNL), heat pumps for all?: With minimum-efficiency equipment most gas homes with AC see higher bills (only 38% positive), but with cold-climate high-efficiency units 99% see lower bills.

RMI 2018, the economics of electrifying buildings: RMI finds that switching from gas raises energy costs in cold Chicago and Providence but lowers them (though not enough to repay capital) in Houston and Oakland, so it holds only in colder climates. [partial: the verdict note limits its own scope ("though")]

Less 2021 (LBNL), cost of decarbonization retrofits: LBNL: among 273 electrification projects, some raised annual energy costs (by up to $1,000) where electricity was expensive, while many cut costs substantially, mostly where oil or propane was replaced; so higher bills happen in some cases, not as a rule.

PNNL 2025, DOE Cold Climate Heat Pump Challenge field results: Billing was not analysed, but several participants who had gas cited higher utility bills and one would not recommend a cold-climate heat pump to neighbours with gas; respondents were manufacturer employees or affiliates, and others reported lower bills.

Context only, Canadian Climate Institute 2023, heat pumps pay off (wrong): a passing remark in a financing section that points back to Canadian lifetime-cost modelling; it measures no bill change after switching from gas.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Qualifies Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff partly 55%, direct lines 1850-1863UK households switching from gas boilers see running costs about £190/yr higher on a standard tariff (about £83 after dropping the gas standing charge), so bills do rise modestly; but the same paper finds a heat pump time-of-use tariff saves about £318/yr, enough to reverse the rise.
• Qualifies Wilson 2024 (NREL and LBNL), heat pumps for all? partly 50%, direct lines 409-415With minimum-efficiency equipment most gas homes with AC see higher bills (only 38% positive), but with cold-climate high-efficiency units 99% see lower bills.
• Qualifies RMI 2018, the economics of electrifying buildings partly 50%, partial lines 158-161RMI finds that switching from gas raises energy costs in cold Chicago and Providence but lowers them (though not enough to repay capital) in Houston and Oakland, so it holds only in colder climates.Applicability: the verdict note limits its own scope ("though")
• Qualifies Less 2021 (LBNL), cost of decarbonization retrofits partly 40%, direct lines 2700-2710LBNL: among 273 electrification projects, some raised annual energy costs (by up to $1,000) where electricity was expensive, while many cut costs substantially, mostly where oil or propane was replaced; so higher bills happen in some cases, not as a rule.
• Qualifies PNNL 2025, DOE Cold Climate Heat Pump Challenge field results partly 25%, direct lines 1658-1691Billing was not analysed, but several participants who had gas cited higher utility bills and one would not recommend a cold-climate heat pump to neighbours with gas; respondents were manufacturer employees or affiliates, and others reported lower bills.

• Context Canadian Climate Institute 2023, heat pumps pay off wrong 30%, contextual (reviewed) lines 1591-1597The report says energy bills tend to be lower in homes with heat pumps, based on its Canadian modelling where heat pumps are the lowest lifetime-cost option in most cities; it is a model, and Edmonton is an exception.Applicability: a passing remark in a financing section that points back to Canadian lifetime-cost modelling; it measures no bill change after switching from gas

Finding 6 (evidence) The evidence does not support "Switching to a heat pump cuts heating bills by half or more"

qualified qualify review qualified (qualify): 3 supporting (3 independent, strongest 0.55), 2 qualifying.

Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff: In the UK, replacing a gas boiler with a heat pump raised measured running costs by about £190/yr on a standard tariff, far from halving bills, even though energy use fell 40%. [partial: the verdict note limits its own scope ("though")]

CEE 2017, cold-climate heat pumps in Minnesota: Cost reductions were 26-40% against propane furnaces and 53-56% against electric resistance, so 'half or more' held only for resistance-heated homes.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Edwards 2024, inequities in heat pump bill savings wrong 55%, direct lines 290-294Edwards' Monte Carlo bill model finds only 53% of households would clearly see lower energy bills from a heat pump, 19% would pay more and 28% show no clear change, which contradicts a general halving of bills.
• Supports RMI 2018, the economics of electrifying buildings wrong 45%, direct lines 158-161In RMI's gas retrofits, energy costs rise with a heat pump in Chicago and Providence and fall only modestly in Houston and Oakland.
• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? wrong 40%, direct lines 332-334Even national average site energy savings are 31-47% depending on efficiency, and bill savings are smaller because electricity costs more per unit; average bill savings are $100 (minimum-efficiency) to $740 a year.

• Qualifies Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff wrong 55%, partial lines 1850-1855In the UK, replacing a gas boiler with a heat pump raised measured running costs by about £190/yr on a standard tariff, far from halving bills, even though energy use fell 40%.Applicability: the verdict note limits its own scope ("though")
• Qualifies CEE 2017, cold-climate heat pumps in Minnesota partly 50%, direct lines 298-314Cost reductions were 26-40% against propane furnaces and 53-56% against electric resistance, so 'half or more' held only for resistance-heated homes.

Finding 7 (evidence) Time-of-use or heat-pump electricity tariffs make heat pumps cheaper to run

qualified qualify review qualified (qualify): 2 supporting (2 independent, strongest 0.7), 2 qualifying.

RMI 2018, the economics of electrifying buildings: In RMI's models a tariff alone does not cut costs: a standard heat pump on Houston's Free Nights rate costs more than on a flat rate, and only flexible control on steep tariffs (3:1 TOU, Free Nights) saved about $1,000-2,000 over 15 years, while mild default TOU gave too little.

Wilson 2024 (NREL and LBNL), heat pumps for all?: Where time-of-use rates exist there may be an opportunity to shift heat pump use off-peak and improve bill savings; not analysed.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff right 70%, direct lines 1898-1906Staggered-adoption study of 6,631 UK households: switching to Octopus's Cosy heat pump time-of-use tariff saved about £318/yr (about 18% of the electricity bill), £78 from the tariff structure alone and £240 from shifting load; savings are vs the same supplier's flat tariff, and the authors are linked to Octopus.
• Supports Mass Save 2026, air-source heat pump rebates right 35%, direct lines 30-33Mass Save: eligible customers can get a Seasonal Heat Pump Rate with lower electric rates November through April 'to help make heat pump operation more affordable'; time-of-use rates are only described as future options.

• Qualifies RMI 2018, the economics of electrifying buildings partly 50%, direct lines 748-755In RMI's models a tariff alone does not cut costs: a standard heat pump on Houston's Free Nights rate costs more than on a flat rate, and only flexible control on steep tariffs (3:1 TOU, Free Nights) saved about $1,000-2,000 over 15 years, while mild default TOU gave too little.
• Qualifies Wilson 2024 (NREL and LBNL), heat pumps for all? partly 20%, direct lines 754-757Where time-of-use rates exist there may be an opportunity to shift heat pump use off-peak and improve bill savings; not analysed.

Central conclusion Payback for a heat pump varies widely with climate, fuel prices and the equipment it replaces

supported approved supported (approved): 6 supporting (6 independent, strongest 0.7).

• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 70%, direct lines 368-373Bill savings vary by climate, prices and housing stock across locations, and vary even more between the 5th and 95th percentiles within one location.
• Supports Deetjen 2021, private economics of US residential heat pumps right 65%, direct lines 646-654Adoption economics vary by grid region, climate, housing and especially baseline heating fuel: gas rarely pays, resistance almost always does.
• Supports Edwards 2024, inequities in heat pump bill savings right 55%, direct lines 292-299Edwards finds bill benefits vary across the US, with the primary heating fuel a key driver and climate and energy prices also important.
• Supports CEE 2017, cold-climate heat pumps in Minnesota right 55%, direct lines 1376-1408Simple payback ranged from 0.6 years (furnace and AC both failed) to 28.8 years (no failure, versus a condensing furnace), depending on what equipment is replaced and the baseline furnace efficiency.
• Supports RMI 2018, the economics of electrifying buildings right 50%, direct lines 150-161RMI's lifetime cost results flip with fuel (propane and oil favour heat pumps), with whether an AC is replaced too, with new build versus retrofit, and with city climate and prices.
• Supports Canadian Climate Institute 2023, heat pumps pay off right 45%, direct lines 139-147Cost-competitiveness varies most with regional energy prices and climate: heat pumps win even under worst-case assumptions in Montreal, Halifax and Vancouver but not in cold, cheap-gas Edmonton; the replaced equipment is not varied.

Finding 1 (evidence) The evidence does not support "A heat pump pays for itself through lower energy bills within a few years"

qualified qualify review qualified (qualify): 3 supporting (3 independent, strongest 0.55), 3 qualifying, 1 context only.

RMI 2018, the economics of electrifying buildings: In RMI's gas retrofits, heat pumps either raise energy costs (Chicago, Providence) or save too little to recover the extra capital (Houston, Oakland), so for most gas homes they never pay back at 2018 prices. [partial: gas retrofits in four modelled cities either cost more to run or save too little to recover the capital; four cities and 2018 prices, so it narrows rather than settles the national promise]

CEE 2018, ducted and ductless cold-climate heat pumps in the field: Payback is 5-6 years only when the existing heating or cooling system has failed; as a pure retrofit it typically exceeds 10 years, so 'a few years' holds only at replacement time.

Canadian Climate Institute 2023, heat pumps pay off: The report says lower heat pump operating costs offset the higher capital cost within a few years (citing Kaufman 2019 and Goody 2014), but its own model shows this depends on region and fails in many Edmonton cases.

Context only, Edwards 2024, inequities in heat pump bill savings (wrong): a population base rate cannot rule out an individual outcome the sentence only says is possible.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? wrong 55%, direct (reviewed) lines 497-520Even over 16 years, unsubsidized NPV is positive for only 55% (minimum), 41% (medium) and 21% (high-efficiency) of homes, far from a few-year payback for all.
• Supports Deetjen 2021, private economics of US residential heat pumps wrong 50%, direct (reviewed) lines 570-574Only 21% of US single-family houses without heat pumps would benefit economically over a 15-year life from replacing their heater with one.
• Supports Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff wrong 50%, direct lines 1868-1871For UK households replacing gas boilers, operating costs are higher, not lower, on a standard tariff, leaving a net private cost of about £4,121 over 20 years, so there is no payback from bill savings.

• Qualifies RMI 2018, the economics of electrifying buildings wrong 50%, partial (reviewed) lines 158-161In RMI's gas retrofits, heat pumps either raise energy costs (Chicago, Providence) or save too little to recover the extra capital (Houston, Oakland), so for most gas homes they never pay back at 2018 prices.Applicability: gas retrofits in four modelled cities either cost more to run or save too little to recover the capital; four cities and 2018 prices, so it narrows rather than settles the national promise
• Qualifies CEE 2018, ducted and ductless cold-climate heat pumps in the field partly 50%, direct lines 513-516Payback is 5-6 years only when the existing heating or cooling system has failed; as a pure retrofit it typically exceeds 10 years, so 'a few years' holds only at replacement time.
• Qualifies Canadian Climate Institute 2023, heat pumps pay off partly 25%, direct lines 1488-1492The report says lower heat pump operating costs offset the higher capital cost within a few years (citing Kaufman 2019 and Goody 2014), but its own model shows this depends on region and fails in many Edmonton cases.

• Context Edwards 2024, inequities in heat pump bill savings wrong 45%, contextual lines 290-294Edwards finds 19% of households would face higher bills and 28% no clear change, so a heat pump cannot pay for itself through bills for nearly half of homes.Applicability: a population base rate cannot rule out an individual outcome the sentence only says is possible

Finding 2 (evidence) Payback for a heat pump varies widely with climate, fuel prices and the equipment it replaces

supported approved supported (approved): 6 supporting (6 independent, strongest 0.7).

• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 70%, direct lines 368-373Bill savings vary by climate, prices and housing stock across locations, and vary even more between the 5th and 95th percentiles within one location.
• Supports Deetjen 2021, private economics of US residential heat pumps right 65%, direct lines 646-654Adoption economics vary by grid region, climate, housing and especially baseline heating fuel: gas rarely pays, resistance almost always does.
• Supports Edwards 2024, inequities in heat pump bill savings right 55%, direct lines 292-299Edwards finds bill benefits vary across the US, with the primary heating fuel a key driver and climate and energy prices also important.
• Supports CEE 2017, cold-climate heat pumps in Minnesota right 55%, direct lines 1376-1408Simple payback ranged from 0.6 years (furnace and AC both failed) to 28.8 years (no failure, versus a condensing furnace), depending on what equipment is replaced and the baseline furnace efficiency.
• Supports RMI 2018, the economics of electrifying buildings right 50%, direct lines 150-161RMI's lifetime cost results flip with fuel (propane and oil favour heat pumps), with whether an AC is replaced too, with new build versus retrofit, and with city climate and prices.
• Supports Canadian Climate Institute 2023, heat pumps pay off right 45%, direct lines 139-147Cost-competitiveness varies most with regional energy prices and climate: heat pumps win even under worst-case assumptions in Montreal, Halifax and Vancouver but not in cold, cheap-gas Edmonton; the replaced equipment is not varied.

Finding 3 (evidence) The evidence does not support "Most US homes would save money over the equipment's life by switching to a heat pump"

qualified qualify review qualified (qualify): 1 supporting (1 independent, strongest 0.55), 4 qualifying.

Wilson 2024 (NREL and LBNL), heat pumps for all?: Picking the best package per home, 59% (65 million) have positive unsubsidized NPV over 16 years, but no single package exceeds 55%, and that is an idealized best choice.

Edwards 2024, inequities in heat pump bill savings: Edwards cites earlier studies finding only about 30% of households would benefit from a single-speed heat pump and newer cold-climate studies about 60%; its own bill model gives 53% with 28% uncertain, so the majority is thin and depends on assumptions.

RMI 2018, the economics of electrifying buildings: RMI concludes that most of the 56 million gas-heated US households would not find a heat pump cost-effective at 2018 prices, although propane, oil and new-build homes would. [partial: the verdict note limits its own scope ("although")]

Canadian Climate Institute 2023, heat pumps pay off: Modelling of five Canadian cities finds a heat pump is the lowest lifetime-cost heating and cooling option for most households; this is Canada with its rebates and carbon price, not the US.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Deetjen 2021, private economics of US residential heat pumps wrong 55%, direct lines 886-889With minimum-efficiency heat pumps, 7% discount rate and 15-year life, only 21% of US single-family houses would gain economically.

• Qualifies Wilson 2024 (NREL and LBNL), heat pumps for all? partly 50%, direct lines 645-647Picking the best package per home, 59% (65 million) have positive unsubsidized NPV over 16 years, but no single package exceeds 55%, and that is an idealized best choice.
• Qualifies Edwards 2024, inequities in heat pump bill savings partly 50%, direct lines 857-860Edwards cites earlier studies finding only about 30% of households would benefit from a single-speed heat pump and newer cold-climate studies about 60%; its own bill model gives 53% with 28% uncertain, so the majority is thin and depends on assumptions.
• Qualifies RMI 2018, the economics of electrifying buildings partly 40%, partial lines 1325-1327RMI concludes that most of the 56 million gas-heated US households would not find a heat pump cost-effective at 2018 prices, although propane, oil and new-build homes would.Applicability: the verdict note limits its own scope ("although")
• Qualifies Canadian Climate Institute 2023, heat pumps pay off partly 30%, direct lines 842-846Modelling of five Canadian cities finds a heat pump is the lowest lifetime-cost heating and cooling option for most households; this is Canada with its rebates and carbon price, not the US.

Finding 4 (evidence) High installation costs keep heat pumps from paying off for many households

qualified qualify review qualified (qualify): 2 supporting (2 independent, strongest 0.6), 5 qualifying.

Deetjen 2021, private economics of US residential heat pumps: Small reductions in capital cost could greatly increase economic adoption rates, which are limited to 21% of houses at current costs. [partial: the verdict note limits its own scope ("limited to")]

CEE 2017, cold-climate heat pumps in Minnesota: At $9,450 installed, a ccASHP with propane backup had a lifetime savings-to-investment ratio of 0.63 against a condensing furnace (does not pay off) but 1.24 against an 82% furnace.

Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff: In the UK the heat pump's net private cost after the grant is about £4,121 over 20 years on a standard tariff (it does not pay off), but switching to the heat pump time-of-use tariff is worth about £4,675, which more than covers it.

Canadian Climate Institute 2023, heat pumps pay off: In some Canadian regions heat pumps win on lifetime cost only with current rebates and carbon pricing, especially under high upfront-cost assumptions, while elsewhere they win even without support.

Less 2021 (LBNL), cost of decarbonization retrofits: LBNL: even the cheapest deep path (heat pump electrification with PV and weatherization, about $54k) needed about $20k of cost cuts to break even monthly; this applies to whole retrofit packages, not a heat pump alone.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 60%, direct lines 517-520Higher upfront costs cut positive unsubsidized NPV to 41% and 21% for the more efficient units despite bigger bill savings; cost reductions are needed for almost half of households.
• Supports RMI 2018, the economics of electrifying buildings right 45%, direct lines 158-161RMI finds gas customers face higher upfront retrofit costs, and in Houston and Oakland the energy savings are too small to make up the extra capital.

• Qualifies Deetjen 2021, private economics of US residential heat pumps right 45%, partial lines 919-923Small reductions in capital cost could greatly increase economic adoption rates, which are limited to 21% of houses at current costs.Applicability: the verdict note limits its own scope ("limited to")
• Qualifies CEE 2017, cold-climate heat pumps in Minnesota partly 45%, direct lines 1376-1383At $9,450 installed, a ccASHP with propane backup had a lifetime savings-to-investment ratio of 0.63 against a condensing furnace (does not pay off) but 1.24 against an 82% furnace.
• Qualifies Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff partly 35%, direct lines 2019-2024In the UK the heat pump's net private cost after the grant is about £4,121 over 20 years on a standard tariff (it does not pay off), but switching to the heat pump time-of-use tariff is worth about £4,675, which more than covers it.
• Qualifies Canadian Climate Institute 2023, heat pumps pay off partly 30%, direct lines 1202-1208In some Canadian regions heat pumps win on lifetime cost only with current rebates and carbon pricing, especially under high upfront-cost assumptions, while elsewhere they win even without support.
• Qualifies Less 2021 (LBNL), cost of decarbonization retrofits partly 25%, direct lines 134-139LBNL: even the cheapest deep path (heat pump electrification with PV and weatherization, about $54k) needed about $20k of cost cuts to break even monthly; this applies to whole retrofit packages, not a heat pump alone.

Finding 5 (evidence) Payback should count only a heat pump's extra cost over the system you would replace anyway

supported qualify review supported (qualify): 3 supporting (3 independent, strongest 0.4).

Suggestion: The strongest support is weak (0.4); say "one study suggests" or find stronger evidence.

• Supports Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff right 40%, direct lines 1821-1825Bernard et al. put the average Boiler Upgrade Scheme install at £12,713 total (private spend plus subsidy), £5,213 private, and count it against a £2,250 gas boiler the household would buy anyway, so the cost they weigh is the extra over the replaced system; a modelling assumption, not a test of the rule.
• Supports Deetjen 2021, private economics of US residential heat pumps right 40%, direct lines 500-506The study counts only the extra cost of a heat pump over replacing the existing heater with a similar one, assuming purchase at end of the old heater's life.
• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 40%, direct lines 709-716Wilson's consumer economics compare heat pump installation with a reference scenario of like-for-like replacement of equipment at end of lifetime, and find a majority of households can benefit against that reference; the shares are therefore the replacement-time case, not early replacement of working equipment.

Finding 6 (evidence) Replacing a working gas furnace with a heat pump rarely pays back on energy savings alone

supported approved supported (approved): 3 supporting (3 independent, strongest 0.5).

• Supports Deetjen 2021, private economics of US residential heat pumps right 50%, direct lines 647-652Even assuming replacement at end of life, switching from natural gas to a heat pump rarely produces a private benefit, almost never in cold climates; replacing a working furnace would only add cost.
• Supports RMI 2018, the economics of electrifying buildings right 50%, direct lines 423-431In RMI's four-city model at 2018 prices, homes with gas heat that only need a new furnace usually pay more over 15 years to electrify than to stay on gas, even when the furnace is being replaced anyway (an advocacy model of a few cities).
• Supports Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff right 40%, direct lines 1850-1871In the UK the switch from a gas boiler has a net private cost of about £4,121 over 20 years even after the £7,500 grant, because running costs are higher on a standard tariff, so energy savings never repay the purchase (UK boilers; time-of-use tariffs change this).

Finding 7 (evidence) A heat pump makes the most financial sense when the air conditioner also needs replacing

qualified qualify review qualified (qualify): 5 supporting (4 independent, strongest 0.5), 1 qualifying.

Canadian Climate Institute 2023, heat pumps pay off: Doubling as cooling is one reason heat pumps beat gas plus AC on lifetime cost, but the model always compares full replacements and does not test the case where only the furnace is due.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 50%, direct lines 522-526Homes without existing AC are doubly challenged: higher bills from new cooling and a higher incremental cost because there is no AC replacement to offset.
• Supports RMI 2018, the economics of electrifying buildings right 50%, direct lines 1268-1272RMI finds heat pump retrofits of gas homes generally cost more, with the notable exception of a customer replacing both furnace and AC, whom one heat pump serves (true in Houston, Chicago and Oakland but not Providence).
• Supports CEE 2018, ducted and ductless cold-climate heat pumps in the field right 50%, direct lines 513-516Retrofit paybacks typically exceed 10 years unless equipment needs replacing; at heating or cooling system failure the payback drops to 5-6 years.
• Supports CEE 2017, cold-climate heat pumps in Minnesota right 50%, direct lines 1386-1408Incremental cost falls to $4,270 at AC failure and $370 when furnace and AC both fail (payback 0.6-1.1 years), versus 14.5-28.8 years with no failure.
• Supports Less 2021 (LBNL), cost of decarbonization retrofits right 35%, direct lines 2288-2293LBNL: the heat pump's cost premium over gas equipment disappears once cooling must also be added (like-for-like heating plus cooling), which is the case when the air conditioner also needs replacing.

• Qualifies Canadian Climate Institute 2023, heat pumps pay off partly 30%, direct lines 842-846Doubling as cooling is one reason heat pumps beat gas plus AC on lifetime cost, but the model always compares full replacements and does not test the case where only the furnace is due.

Central conclusion A heat pump costs less to run than heating with oil, propane or electric resistance

qualified qualify review qualified (qualify): 3 supporting (3 independent, strongest 0.65), 4 qualifying.

CEE 2018, ducted and ductless cold-climate heat pumps in the field: Measured cost savings averaged 33% against propane furnaces and 53% against electric resistance in Minnesota homes; oil was not tested, so the claim is supported for two of its three fuels.

Cadmus 2016, ductless mini-split evaluation (MA and RI): Mini-splits were cheaper to run than electric resistance and propane at all observed temperatures, but against oil they were cheaper only above a price-dependent break-even (about 32 F in 2016, 12 F in 2015).

Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff: In the UK, households previously on electric heating used less electricity after a heat pump, which supports the resistance-heat part; oil and propane are not tested.

Canadian Climate Institute 2023, heat pumps pay off: Heat pumps use up to 65% less energy than electric baseboards, which supports the resistance-heat part; oil and propane are not modelled.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 65%, direct lines 393-401Homes heating with electricity, oil or propane that have AC see lower bills in 95-100% of cases (median $300-$650/yr); 73-86% without AC; still 87-99% (propane) and 66-100% (oil) at 2019 prices.
• Supports Edwards 2024, inequities in heat pump bill savings right 55%, direct lines 296-299In Edwards' county model, where homes heat with electricity (resistance) or non-gas fuels, heating is generally more expensive and heat pump bill savings are higher.
• Supports RMI 2018, the economics of electrifying buildings right 45%, direct lines 817-820In Providence, RMI's heat pumps have a lower 15-year net present cost than both oil and propane systems, and save at least $10,000 against propane.

• Qualifies CEE 2018, ducted and ductless cold-climate heat pumps in the field partly 55%, direct lines 376-382Measured cost savings averaged 33% against propane furnaces and 53% against electric resistance in Minnesota homes; oil was not tested, so the claim is supported for two of its three fuels.
• Qualifies Cadmus 2016, ductless mini-split evaluation (MA and RI) partly 55%, direct lines 3611-3619Mini-splits were cheaper to run than electric resistance and propane at all observed temperatures, but against oil they were cheaper only above a price-dependent break-even (about 32 F in 2016, 12 F in 2015).
• Qualifies Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff partly 30%, direct lines 1267-1273In the UK, households previously on electric heating used less electricity after a heat pump, which supports the resistance-heat part; oil and propane are not tested.
• Qualifies Canadian Climate Institute 2023, heat pumps pay off partly 30%, direct lines 463-467Heat pumps use up to 65% less energy than electric baseboards, which supports the resistance-heat part; oil and propane are not modelled.

Finding 1 (evidence) A heat pump costs less to run than heating with oil, propane or electric resistance

qualified qualify review qualified (qualify): 3 supporting (3 independent, strongest 0.65), 4 qualifying.

CEE 2018, ducted and ductless cold-climate heat pumps in the field: Measured cost savings averaged 33% against propane furnaces and 53% against electric resistance in Minnesota homes; oil was not tested, so the claim is supported for two of its three fuels.

Cadmus 2016, ductless mini-split evaluation (MA and RI): Mini-splits were cheaper to run than electric resistance and propane at all observed temperatures, but against oil they were cheaper only above a price-dependent break-even (about 32 F in 2016, 12 F in 2015).

Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff: In the UK, households previously on electric heating used less electricity after a heat pump, which supports the resistance-heat part; oil and propane are not tested.

Canadian Climate Institute 2023, heat pumps pay off: Heat pumps use up to 65% less energy than electric baseboards, which supports the resistance-heat part; oil and propane are not modelled.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 65%, direct lines 393-401Homes heating with electricity, oil or propane that have AC see lower bills in 95-100% of cases (median $300-$650/yr); 73-86% without AC; still 87-99% (propane) and 66-100% (oil) at 2019 prices.
• Supports Edwards 2024, inequities in heat pump bill savings right 55%, direct lines 296-299In Edwards' county model, where homes heat with electricity (resistance) or non-gas fuels, heating is generally more expensive and heat pump bill savings are higher.
• Supports RMI 2018, the economics of electrifying buildings right 45%, direct lines 817-820In Providence, RMI's heat pumps have a lower 15-year net present cost than both oil and propane systems, and save at least $10,000 against propane.

• Qualifies CEE 2018, ducted and ductless cold-climate heat pumps in the field partly 55%, direct lines 376-382Measured cost savings averaged 33% against propane furnaces and 53% against electric resistance in Minnesota homes; oil was not tested, so the claim is supported for two of its three fuels.
• Qualifies Cadmus 2016, ductless mini-split evaluation (MA and RI) partly 55%, direct lines 3611-3619Mini-splits were cheaper to run than electric resistance and propane at all observed temperatures, but against oil they were cheaper only above a price-dependent break-even (about 32 F in 2016, 12 F in 2015).
• Qualifies Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff partly 30%, direct lines 1267-1273In the UK, households previously on electric heating used less electricity after a heat pump, which supports the resistance-heat part; oil and propane are not tested.
• Qualifies Canadian Climate Institute 2023, heat pumps pay off partly 30%, direct lines 463-467Heat pumps use up to 65% less energy than electric baseboards, which supports the resistance-heat part; oil and propane are not modelled.

Finding 2 (evidence) Heat pumps save the most money in homes now heated with oil, propane or electric resistance

qualified qualify review qualified (qualify): 5 supporting (5 independent, strongest 0.6), 2 qualifying.

Deetjen 2021, private economics of US residential heat pumps: Counting capital cost, resistance and propane homes usually gain, but oil-heated homes usually take a private loss (65% are in cold climates), so oil is not among the best cases.

Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff: UK households that previously used electric heating saw electricity use fall after a heat pump, while gas-boiler households paid slightly more to run one, which fits the claim for resistance heat; oil and propane are not analysed.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 60%, direct lines 393-414Median bill savings of $300-$650/yr for electric, oil and propane homes with AC compare with -$70 to $380/yr for gas homes with AC.
• Supports Edwards 2024, inequities in heat pump bill savings right 55%, direct lines 296-299Edwards finds the primary heating fuel is a key driver: bill savings are highest where homes now heat with electricity or fuels other than gas.
• Supports RMI 2018, the economics of electrifying buildings right 45%, direct lines 862-865RMI concludes that electrification has the greatest immediate cost and carbon benefit for oil and propane customers.
• Supports Cadmus 2016, ductless mini-split evaluation (MA and RI) right 45%, direct lines 1184-1205Savings were highest against electric resistance, and the evaluators recommend targeting resistance-heated and propane-heated homes (always cheaper there) and, with caveats, oil homes; against gas there were essentially no running-cost savings.
• Supports Less 2021 (LBNL), cost of decarbonization retrofits right 40%, direct lines 2700-2706LBNL: the large energy cost reductions in electrification projects were most likely in homes with high pre-retrofit bills, such as those heating with propane or fuel oil.

• Qualifies Deetjen 2021, private economics of US residential heat pumps partly 45%, direct lines 653-665Counting capital cost, resistance and propane homes usually gain, but oil-heated homes usually take a private loss (65% are in cold climates), so oil is not among the best cases.
• Qualifies Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff partly 30%, direct lines 1267-1273UK households that previously used electric heating saw electricity use fall after a heat pump, while gas-boiler households paid slightly more to run one, which fits the claim for resistance heat; oil and propane are not analysed.

Central conclusion Heat pumps cut a home's carbon emissions compared with gas heating, even on today's grid

qualified qualify review qualified (qualify): 4 supporting (4 independent, strongest 0.65), 2 qualifying.

RMI 2018, the economics of electrifying buildings: On 2016 short-term marginal grid data, heat pumps cut emissions versus gas in Oakland, Houston and Providence but raised them in coal-heavy Chicago (about 50% more), so the claim holds on most grids but not all.

Less 2021 (LBNL), cost of decarbonization retrofits: LBNL: fuel switching boosted carbon cuts (68% vs 50%) in the electrification cluster, but those projects were mostly in states with low-carbon grids, and the authors expect lower carbon savings where electricity is more carbon-intensive.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 60%, direct lines 248-258Average GHG savings are positive in every state for every heat pump and future grid scenario, even the conservative high-renewable-cost one, with emissions levelized over a 16-year equipment lifetime (projected grids, not today's).
• Supports Deetjen 2021, private economics of US residential heat pumps right 50%, direct lines 635-642Heat pump adoption raises CO2 for only 2.1% of US houses (1.7 million), but the estimate assumes grid CO2 emissions fall 45% over the heat pump's 15-year life.
• Supports Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff right 65%, direct lines 193-198On the 2024 UK grid, replacing a gas boiler with a heat pump cut household energy CO2eq by about 1.16 t/yr, 36% of the pre-adoption footprint, from measured consumption changes and government emission factors.
• Supports Canadian Climate Institute 2023, heat pumps pay off right 40%, direct lines 459-462Even on relatively emissions-intensive electricity, heat pumps emit 20 to 30% less than gas furnaces (citing IEA 2022 and a consultancy), and near zero on clean grids.

• Qualifies RMI 2018, the economics of electrifying buildings partly 50%, direct lines 225-231On 2016 short-term marginal grid data, heat pumps cut emissions versus gas in Oakland, Houston and Providence but raised them in coal-heavy Chicago (about 50% more), so the claim holds on most grids but not all.
• Qualifies Less 2021 (LBNL), cost of decarbonization retrofits partly 35%, direct lines 1947-1952LBNL: fuel switching boosted carbon cuts (68% vs 50%) in the electrification cluster, but those projects were mostly in states with low-carbon grids, and the authors expect lower carbon savings where electricity is more carbon-intensive.

Finding 1 (evidence) Heat pumps cut a home's carbon emissions compared with gas heating, even on today's grid

qualified qualify review qualified (qualify): 4 supporting (4 independent, strongest 0.65), 2 qualifying.

RMI 2018, the economics of electrifying buildings: On 2016 short-term marginal grid data, heat pumps cut emissions versus gas in Oakland, Houston and Providence but raised them in coal-heavy Chicago (about 50% more), so the claim holds on most grids but not all.

Less 2021 (LBNL), cost of decarbonization retrofits: LBNL: fuel switching boosted carbon cuts (68% vs 50%) in the electrification cluster, but those projects were mostly in states with low-carbon grids, and the authors expect lower carbon savings where electricity is more carbon-intensive.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 60%, direct lines 248-258Average GHG savings are positive in every state for every heat pump and future grid scenario, even the conservative high-renewable-cost one, with emissions levelized over a 16-year equipment lifetime (projected grids, not today's).
• Supports Deetjen 2021, private economics of US residential heat pumps right 50%, direct lines 635-642Heat pump adoption raises CO2 for only 2.1% of US houses (1.7 million), but the estimate assumes grid CO2 emissions fall 45% over the heat pump's 15-year life.
• Supports Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff right 65%, direct lines 193-198On the 2024 UK grid, replacing a gas boiler with a heat pump cut household energy CO2eq by about 1.16 t/yr, 36% of the pre-adoption footprint, from measured consumption changes and government emission factors.
• Supports Canadian Climate Institute 2023, heat pumps pay off right 40%, direct lines 459-462Even on relatively emissions-intensive electricity, heat pumps emit 20 to 30% less than gas furnaces (citing IEA 2022 and a consultancy), and near zero on clean grids.

• Qualifies RMI 2018, the economics of electrifying buildings partly 50%, direct lines 225-231On 2016 short-term marginal grid data, heat pumps cut emissions versus gas in Oakland, Houston and Providence but raised them in coal-heavy Chicago (about 50% more), so the claim holds on most grids but not all.
• Qualifies Less 2021 (LBNL), cost of decarbonization retrofits partly 35%, direct lines 1947-1952LBNL: fuel switching boosted carbon cuts (68% vs 50%) in the electrification cluster, but those projects were mostly in states with low-carbon grids, and the authors expect lower carbon savings where electricity is more carbon-intensive.

Central conclusion Heat pump efficiency and capacity fall as the outdoor temperature drops

supported approved supported (approved): 9 supporting (8 independent, strongest 0.65).

• Supports CEE 2017, cold-climate heat pumps in Minnesota right 65%, direct lines 1143-1157Measured heat-pump-only cycles averaged COP about 1.3 at the 10F change point rising to 3.0-4.5 at 50-60F; outdoor temperature had the largest effect on COP across six Minnesota sites.
• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 60%, direct lines 1097-1100Capacity retention curves from lab and NEEP manufacturer data fall from 100% of rated output at 47F to 47% (single-speed) or 85% (cold-climate) at 5F.
• Supports Deetjen 2021, private economics of US residential heat pumps right 55%, direct lines 300-305Heat pump efficiency falls with ambient temperature, and at low temperatures the load can exceed capacity so the model falls back to resistance heat (COP 1).
• Supports Cadmus 2016, ductless mini-split evaluation (MA and RI) right 55%, direct lines 3120-3127The evaluation states unit efficiency drops as outdoor temperature drops and estimates from manufacturer data that capacity falls about 20% from 17 F to 5 F at fixed airflow; about half the units could not meet the space heating load below 17 F.
• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results right 55%, direct lines 1000-1003Heating COPs rose steadily with outdoor temperature (median 1.6 in the cold bins up to about 5.2), and the report notes conventional heat pumps lose capacity below 40 F and keep only a fraction in the teens.
• Supports RMI 2018, the economics of electrifying buildings right 50%, direct lines 966-970The cold-climate Mitsubishi unit RMI models has a COP at max capacity that falls from 3.5 at 47F to 2.1 at 17F and 1.4 at -13F; the report shows efficiency falling but gives no capacity figures.
• Supports CEE 2018, ducted and ductless cold-climate heat pumps in the field right 50%, direct lines 478-489A 1-ton ductless unit in northern Minnesota held its median capacity from 10F to -15F but delivered only about 2,600 Btu/h (about 20% of its rated 5F capacity) between -20F and -25F.
• Supports Edwards 2024, inequities in heat pump bill savings right 45%, direct lines 1638-1659Edwards fits heat pump COP from the NEEP database as a linear function of outdoor temperature (COP = 0.05T + 1.6 above 5F), with single-stage units dropping to about COP 1 below 5F; the source covers efficiency only, not capacity.
• Supports Canadian Climate Institute 2023, heat pumps pay off right 35%, direct lines 376-378Older heat pumps lose efficiency in colder temperatures, less so for newer models; the passage says nothing about capacity.

Finding 1 (evidence) Heat pump efficiency and capacity fall as the outdoor temperature drops

supported approved supported (approved): 9 supporting (8 independent, strongest 0.65).

• Supports CEE 2017, cold-climate heat pumps in Minnesota right 65%, direct lines 1143-1157Measured heat-pump-only cycles averaged COP about 1.3 at the 10F change point rising to 3.0-4.5 at 50-60F; outdoor temperature had the largest effect on COP across six Minnesota sites.
• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 60%, direct lines 1097-1100Capacity retention curves from lab and NEEP manufacturer data fall from 100% of rated output at 47F to 47% (single-speed) or 85% (cold-climate) at 5F.
• Supports Deetjen 2021, private economics of US residential heat pumps right 55%, direct lines 300-305Heat pump efficiency falls with ambient temperature, and at low temperatures the load can exceed capacity so the model falls back to resistance heat (COP 1).
• Supports Cadmus 2016, ductless mini-split evaluation (MA and RI) right 55%, direct lines 3120-3127The evaluation states unit efficiency drops as outdoor temperature drops and estimates from manufacturer data that capacity falls about 20% from 17 F to 5 F at fixed airflow; about half the units could not meet the space heating load below 17 F.
• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results right 55%, direct lines 1000-1003Heating COPs rose steadily with outdoor temperature (median 1.6 in the cold bins up to about 5.2), and the report notes conventional heat pumps lose capacity below 40 F and keep only a fraction in the teens.
• Supports RMI 2018, the economics of electrifying buildings right 50%, direct lines 966-970The cold-climate Mitsubishi unit RMI models has a COP at max capacity that falls from 3.5 at 47F to 2.1 at 17F and 1.4 at -13F; the report shows efficiency falling but gives no capacity figures.
• Supports CEE 2018, ducted and ductless cold-climate heat pumps in the field right 50%, direct lines 478-489A 1-ton ductless unit in northern Minnesota held its median capacity from 10F to -15F but delivered only about 2,600 Btu/h (about 20% of its rated 5F capacity) between -20F and -25F.
• Supports Edwards 2024, inequities in heat pump bill savings right 45%, direct lines 1638-1659Edwards fits heat pump COP from the NEEP database as a linear function of outdoor temperature (COP = 0.05T + 1.6 above 5F), with single-stage units dropping to about COP 1 below 5F; the source covers efficiency only, not capacity.
• Supports Canadian Climate Institute 2023, heat pumps pay off right 35%, direct lines 376-378Older heat pumps lose efficiency in colder temperatures, less so for newer models; the passage says nothing about capacity.

Finding 2 (evidence) The evidence does not support "Heat pumps do not work in very cold climates"

qualified qualify review qualified (qualify): 4 supporting (4 independent, strongest 0.6), 1 qualifying, 1 context only.

Wilson 2024 (NREL and LBNL), heat pumps for all?: Over 25,000 products on the NEEP cold-climate list keep a COP of 2 or more at full capacity at 5F, and modelled per-household GHG savings are generally higher in cold climates. [partial: a count of products rated at 5F on a certification list, not measured operation in homes]

Context only, RMI 2018, the economics of electrifying buildings (wrong): background sentence citing a footnote about rated models; RMI did not test cold-weather operation.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports CEE 2018, ducted and ductless cold-climate heat pumps in the field wrong 60%, direct lines 495-518In monitored Minnesota homes cold-climate ASHPs cut heating costs 30-56%; cold weather raised backup and defrost use and lowered heat pump performance but overall system efficiencies remained high.
• Supports Cadmus 2016, ductless mini-split evaluation (MA and RI) wrong 55%, direct lines 2926-2935Across two New England winters, metered mini-splits kept heating below 0 F with average COP about 1.5 at -10 to 0 F, which the report calls far superior to resistance heat.
• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results wrong 55%, direct lines 180-189Prototype cold-climate heat pumps in occupied northern US and Canadian homes kept a median COP of 1.9 at 0-5 F, with most units above the ENERGY STAR 1.75 threshold in the field.
• Supports Canadian Climate Institute 2023, heat pumps pay off wrong 45%, direct lines 528-541Standard heat pumps perform well to -8 to -10°C and cold-climate models to -25°C; Norway, with temperatures similar to Canada's, leads the world in heat pump penetration, with confirmed real-world performance there.

• Qualifies Wilson 2024 (NREL and LBNL), heat pumps for all? wrong 50%, partial (reviewed) lines 157-163Over 25,000 products on the NEEP cold-climate list keep a COP of 2 or more at full capacity at 5F, and modelled per-household GHG savings are generally higher in cold climates.Applicability: a count of products rated at 5F on a certification list, not measured operation in homes

• Context RMI 2018, the economics of electrifying buildings wrong 50%, contextual (reviewed) lines 356-358RMI notes that hundreds of models operate efficiently at 5F and some heat at -13F, and it models cold-climate heat pumps heating homes in Chicago and Providence.Applicability: background sentence citing a footnote about rated models; RMI did not test cold-weather operation

Finding 3 (evidence) Modern cold-climate heat pumps keep heating well below 0°F

qualified qualify review qualified (qualify): 3 supporting (3 independent, strongest 0.5), 3 qualifying.

CEE 2018, ducted and ductless cold-climate heat pumps in the field: A ductless cold-climate unit delivered a consistent median capacity from 10F to -15F and kept operating to -25F at reduced capacity, though the authors say it needs backup below -15F; one monitored unit. [partial: the verdict note limits its own scope ("though")]

Winkler 2023 (NREL), cold-climate heat pump field validation: Among these ducted variable-speed units, Site 10's compressor did not run below 0F despite a disabled lockout, and the dual-fuel unit did not run below about 4F.

Canadian Climate Institute 2023, heat pumps pay off: Cold-climate heat pumps can perform well down to -25°C (about -13°F), citing NRCan, though backup may still be needed in the coldest parts of Canada. [partial: the verdict note limits its own scope ("though")]

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports RMI 2018, the economics of electrifying buildings right 50%, direct lines 966-970RMI's modelled cold-climate unit provides substantial heat down to -13F (COP 1.4 there), based on manufacturer performance data.
• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results right 45%, direct lines 1238-1247The prototype cold-climate units ran mainly in heat-pump-only mode even in the lowest outdoor bins, with resistance-only operation minimal above 0 F; but only 8 of 22 sites logged 10+ hours below 0 F, and auxiliary heat was on about 25% of the time at -10 to 0 F.
• Supports Cadmus 2016, ductless mini-split evaluation (MA and RI) right 40%, direct lines 2997-3002Even non-cold-climate mini-splits from 2012-14 kept operating below 0 F (at lower efficiency than cold-climate models), with COPs averaging about 1.5 at -10 to 0 F; capacity at those temperatures was not measured directly.

• Qualifies CEE 2018, ducted and ductless cold-climate heat pumps in the field right 50%, partial lines 478-489A ductless cold-climate unit delivered a consistent median capacity from 10F to -15F and kept operating to -25F at reduced capacity, though the authors say it needs backup below -15F; one monitored unit.Applicability: the verdict note limits its own scope ("though")
• Qualifies Winkler 2023 (NREL), cold-climate heat pump field validation partly 40%, direct lines 1946-1951Among these ducted variable-speed units, Site 10's compressor did not run below 0F despite a disabled lockout, and the dual-fuel unit did not run below about 4F.
• Qualifies Canadian Climate Institute 2023, heat pumps pay off right 35%, partial lines 528-536Cold-climate heat pumps can perform well down to -25°C (about -13°F), citing NRCan, though backup may still be needed in the coldest parts of Canada.Applicability: the verdict note limits its own scope ("though")

Finding 4 (evidence) A heat pump still uses less energy than resistance heat even in very cold weather

qualified qualify review qualified (qualify): 5 supporting (4 independent, strongest 0.6), 1 qualifying.

Edwards 2024, inequities in heat pump bill savings: Edwards assumes single-stage heat pumps fall to COP 1 (no better than resistance) below 5F, while cold-climate two-stage units keep a higher COP, so the claim depends on equipment type.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Cadmus 2016, ductless mini-split evaluation (MA and RI) right 60%, direct lines 2926-2935In the 2016 winter field data, mini-splits averaged COP 1.5 at -10 to 0 F, giving about 50% more heat per kWh than resistance heat; some non-cold-climate units got close to COP 1.0 in very cold hours, but those hours were few.
• Supports CEE 2018, ducted and ductless cold-climate heat pumps in the field right 55%, direct lines 425-435Because the heat pump is prioritized and resistance only supplements it, the all-electric system's COP stayed above 1.0 even at the coldest observed temperature (about -7F).
• Supports CEE 2017, cold-climate heat pumps in Minnesota right 55%, direct lines 1199-1208Ductless units averaged daily COP 1.7 and 1.3 between 0F and -13F, and one ran with COP above 1.0 at -19F, beating the electric resistance alternative (COP 1).
• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results right 55%, direct lines 180-189Heat-pump-only COP had a median of 1.9 at 0-5 F, so the heat pump used roughly half the electricity of resistance heat; once auxiliary heat is included, system COP is lower in the coldest bins.
• Supports RMI 2018, the economics of electrifying buildings right 50%, direct lines 966-970RMI's cold-climate unit keeps a COP of 1.4 at -13F, so it still uses less electricity than resistance heat (COP 1).

• Qualifies Edwards 2024, inequities in heat pump bill savings partly 40%, direct lines 1649-1659Edwards assumes single-stage heat pumps fall to COP 1 (no better than resistance) below 5F, while cold-climate two-stage units keep a higher COP, so the claim depends on equipment type.

Finding 5 (evidence) Cold-climate heat pumps often rely on backup resistance heat in the coldest weather, which raises bills

qualified qualify review qualified (qualify): 1 supporting (1 independent, strongest 0.45), 4 qualifying.

Winkler 2023 (NREL), cold-climate heat pump field validation: Auxiliary heat runtime generally rose in colder bins and heavy use cut system COP over 30% at two sites, but the heavy users were undersized or faulty while six sites used little auxiliary heat.

CEE 2018, ducted and ductless cold-climate heat pumps in the field: In all-electric systems the booster resistance heat ran at COP 1 below about 30F, but because the heat pump was prioritized the system COP stayed above 1.0 even at -7F, so backup use dented rather than erased efficiency. [partial: the verdict note limits its own scope ("rather than")]

Wilson 2024 (NREL and LBNL), heat pumps for all?: In Minnesota a high-efficiency heat pump sized for the heating load saves a median $182/yr, but sized for cooling it raises bills $96/yr through more resistance backup; reliance and the bill hit depend on sizing, not cold-climate units as such. [partial: the verdict note limits its own scope ("not cold-climate units as such")]

Canadian Climate Institute 2023, heat pumps pay off: In Edmonton's cold climate the model finds heat pumps less efficient and more reliant on the backup system, which together with cheap gas makes all-electric heat pumps less cost-competitive; how often backup runs is not reported.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results right 45%, direct lines 1169-1193Auxiliary resistance heat's median share of heating energy rose from 0% in mild bins to about 11% at 5 F and higher below, and was on 25% of the time at -10 to 0 F; the report says auxiliary heat is less efficient and usually costs more (bills not measured).

• Qualifies Winkler 2023 (NREL), cold-climate heat pump field validation partly 50%, direct lines 2150-2157Auxiliary heat runtime generally rose in colder bins and heavy use cut system COP over 30% at two sites, but the heavy users were undersized or faulty while six sites used little auxiliary heat.
• Qualifies CEE 2018, ducted and ductless cold-climate heat pumps in the field partly 45%, partial lines 425-435In all-electric systems the booster resistance heat ran at COP 1 below about 30F, but because the heat pump was prioritized the system COP stayed above 1.0 even at -7F, so backup use dented rather than erased efficiency.Applicability: the verdict note limits its own scope ("rather than")
• Qualifies Wilson 2024 (NREL and LBNL), heat pumps for all? partly 40%, partial lines 942-946In Minnesota a high-efficiency heat pump sized for the heating load saves a median $182/yr, but sized for cooling it raises bills $96/yr through more resistance backup; reliance and the bill hit depend on sizing, not cold-climate units as such.Applicability: the verdict note limits its own scope ("not cold-climate units as such")
• Qualifies Canadian Climate Institute 2023, heat pumps pay off partly 35%, direct lines 1153-1158In Edmonton's cold climate the model finds heat pumps less efficient and more reliant on the backup system, which together with cheap gas makes all-electric heat pumps less cost-competitive; how often backup runs is not reported.

Finding 6 (evidence) Defrost cycles reduce a heat pump's output and efficiency in cold, humid weather

supported approved supported (approved): 5 supporting (4 independent, strongest 0.6).

• Supports CEE 2017, cold-climate heat pumps in Minnesota right 60%, direct lines 1484-1519Defrost cut ccASHP COP at site 2 (e.g. 1.86 to 1.65 at 10-20F), drove backup energy use between the 10F changeover and 40F, and eliminating it would have raised annual energy savings by 8.3%.
• Supports Winkler 2023 (NREL), cold-climate heat pump field validation right 55%, direct lines 2363-2372Defrost consumed over 20% of compressor heating energy at Sites 5, 7, 8 and 11; defrost cycles peaked at 20-40F wet-bulb, since cold air holds little moisture.
• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results right 55%, direct lines 1249-1310Frost on the outdoor coil in cold humid air cuts heat exchange, and the reverse-cycle defrost pulls heat from the house and drops supply air temperature; defrost took a median 5-8% of energy at 5-35 F and up to 16% of the time at one site.
• Supports CEE 2018, ducted and ductless cold-climate heat pumps in the field right 50%, direct lines 346-352Frost on the outdoor coil restricts airflow, and defrost cycles reverse the refrigeration cycle, using energy, reducing heat pump availability and increasing auxiliary heating; the humidity dependence is not tested here.
• Supports Cadmus 2016, ductless mini-split evaluation (MA and RI) right 40%, direct lines 3320-3329Site conditions caused long defrost periods that lowered efficiency, and at one windy oceanfront site prolonged defrost cycles pulled heat out of the house; this is anecdotal plus the evaluators' general observation.

Finding 7 (evidence) A dual-fuel system (heat pump with a gas furnace backup) is cheaper to run than either alone in cold climates

qualified qualify review qualified (qualify): no direct support, 4 qualifying.

Canadian Climate Institute 2023, heat pumps pay off: Lifetime-cost model: in Toronto a heat pump with gas backup is cheapest for nearly all households, and in Edmonton for 1940 and 1980 detached and townhouses, but gas with AC wins other Edmonton cases; this is lifetime cost including equipment, not running cost alone.

Cadmus 2016, ductless mini-split evaluation (MA and RI): Switching between heat pump and fuel at the crossover temperature is cheapest in principle, but at 2016 prices the gas crossover was above 70 F for the measured units, so pairing with gas saved nothing; only a modelled HSPF-13 unit (28-40 F crossover) would beat gas alone.

Wilson 2024 (NREL and LBNL), heat pumps for all?: Hybrid heat pumps with fossil backup are said to mitigate bill increases without necessarily raising upfront cost, but they were not analysed.

Deetjen 2021, private economics of US residential heat pumps: The authors note private economics might improve if supplemental heat came from gas rather than resistance, but they do not model it. [partial: the verdict note limits its own scope ("rather than")]

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Qualifies Canadian Climate Institute 2023, heat pumps pay off partly 40%, direct lines 1142-1150Lifetime-cost model: in Toronto a heat pump with gas backup is cheapest for nearly all households, and in Edmonton for 1940 and 1980 detached and townhouses, but gas with AC wins other Edmonton cases; this is lifetime cost including equipment, not running cost alone.
• Qualifies Cadmus 2016, ductless mini-split evaluation (MA and RI) partly 30%, direct lines 3436-3459Switching between heat pump and fuel at the crossover temperature is cheapest in principle, but at 2016 prices the gas crossover was above 70 F for the measured units, so pairing with gas saved nothing; only a modelled HSPF-13 unit (28-40 F crossover) would beat gas alone.
• Qualifies Wilson 2024 (NREL and LBNL), heat pumps for all? partly 25%, direct lines 730-732Hybrid heat pumps with fossil backup are said to mitigate bill increases without necessarily raising upfront cost, but they were not analysed.
• Qualifies Deetjen 2021, private economics of US residential heat pumps partly 20%, partial lines 545-547The authors note private economics might improve if supplemental heat came from gas rather than resistance, but they do not model it.Applicability: the verdict note limits its own scope ("rather than")

Finding 8 (evidence) Heat pumps are being used successfully in cold countries like Norway and Canada

supported qualify review supported (qualify): 3 supporting (3 independent, strongest 0.45).

Suggestion: The strongest support is weak (0.45); say "one study suggests" or find stronger evidence.

• Supports Canadian Climate Institute 2023, heat pumps pay off right 45%, direct lines 528-541Norway leads the world in heat pump penetration, with Finland and Sweden close behind and studies confirming real-world performance; elsewhere the report notes heat pumps heat 21 to 32% of homes in Canada's Maritime provinces.
• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 35%, direct lines 126-128Air-to-air heat pumps are the dominant heat pump type in Nordic and Baltic countries (Canada is not mentioned).
• Supports Edwards 2024, inequities in heat pump bill savings right 35%, direct lines 70-71Edwards reports high heat pump penetration in Norway, Finland and Sweden; Canada is not covered and success is inferred from adoption.

Central conclusion From 2023 through 2025 a federal tax credit covered 30% of a qualifying heat pump, up to $2,000

supported approved supported (approved): 2 supporting (2 independent, strongest 0.85).

• Supports IRS, the 25C home improvement credit right 85%, direct lines 10-25IRS 25C page: from 2023 the credit equals 30% of qualified expenses, with a $2,000-per-year cap for qualified heat pumps (installation labor included).
• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 30%, direct lines 620-625The paper refers to claiming a $2,000 IRA tax credit for a heat pump in 2023, consistent with the cap; it does not state the 30% rate.

Finding 1 (evidence) From 2023 through 2025 a federal tax credit covered 30% of a qualifying heat pump, up to $2,000

supported approved supported (approved): 2 supporting (2 independent, strongest 0.85).

• Supports IRS, the 25C home improvement credit right 85%, direct lines 10-25IRS 25C page: from 2023 the credit equals 30% of qualified expenses, with a $2,000-per-year cap for qualified heat pumps (installation labor included).
• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 30%, direct lines 620-625The paper refers to claiming a $2,000 IRA tax credit for a heat pump in 2023, consistent with the cap; it does not state the 30% rate.

Finding 2 (evidence) The evidence does not support "A federal tax credit for heat pumps is available for installations in 2026"

supported approved supported (approved): 2 supporting (2 independent, strongest 0.85).

• Supports IRS 2025, end of the 25C and 25D credits wrong 85%, direct lines 25-32IRS fact sheet on the 2025 budget law (OBBB): the 25C home improvement credit is not allowed for property placed in service after December 31, 2025, and the 25D clean energy credit for expenditures after that date, so no federal heat pump credit applies to 2026 installs.
• Supports IRS, the 25C home improvement credit wrong 80%, direct lines 4-6IRS 25C page: the credit applies to improvements made through December 31, 2025; nothing is available for 2026 installs.

Finding 3 (evidence) The evidence does not support "Federal Inflation Reduction Act rebates for heat pumps are available to most homeowners"

supported approved supported (approved): 3 supporting (3 independent, strongest 0.6).

• Supports DOE 2026, Program Notice 26-2 on electric home rebates wrong 60%, direct lines 145-155DOE Program Notice 26-2: the federal electrification rebate no longer funds replacing non-electric (gas, oil, propane) heating, which rules out most homes that would switch to a heat pump; it is also an income-limited program run by the states.
• Supports National Housing Trust 2026, state rebate tracker wrong 55%, direct lines 17-31The DOE home energy rebates target low- to moderate-income households, and as of Feb 2026 most states had no launch date, so they are not available to most homeowners.
• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? wrong 40%, direct lines 613-620The $4.5B IRA electrification rebates could cover roughly 500,000 low-income or 1 million moderate-income households, and efficiency rebates 0.5-2 million more, a small fraction of over 110 million households.

Finding 4 (evidence) Federal heat pump rebates for low- and moderate-income households depend on slow state-by-state rollout

supported approved supported (approved): 1 supporting (1 independent, strongest 0.7).

• Supports National Housing Trust 2026, state rebate tracker right 70%, direct lines 98-126As of Feb 2026, the National Housing Trust tracker shows most states' HEAR and HER programs only conditionally approved with no expected launch date (e.g. Alabama, Alaska, Arkansas), and others partly launched or in pilots, three-plus years after the 2022 law.

Finding 5 (evidence) Since mid-2026 the federal electric appliance rebate no longer pays to replace gas or oil heating with a heat pump

supported approved supported (approved): 2 supporting (2 independent, strongest 0.85).

• Supports DOE 2026, Program Notice 26-2 on electric home rebates right 85%, direct (reviewed) lines 145-155DOE Program Notice 26-2 (effective May 29, 2026) removes fuel-switching from the HEEHR (formerly HEAR) rebate, allowing HVAC rebates only for replacing existing electric equipment; launched programs must comply within three months, though existing reservations can still be honored.
• Supports North Carolina 2026, Energy Saver program changes right 75%, direct lines 4-16North Carolina's state energy office: under DOE's May 29, 2026 guidance, from Sept. 1, 2026 HEAR rebates are no longer allowed for replacing gas or other fossil-fuel appliances; only electric-to-electric upgrades qualify.

Finding 6 (evidence) State and utility rebates can cover a large share of a heat pump's cost

qualified qualify review qualified (qualify): 1 supporting (1 independent, strongest 0.4), 1 qualifying.

Less 2021 (LBNL), cost of decarbonization retrofits: LBNL: 71% of projects got rebates, with a median of $1,327 (21% of gross cost, mean 30%); electrification programs such as MA DOER Home MVP gave higher shares (37%), so rebates covered a meaningful but usually minority share.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Mass Save 2026, air-source heat pump rebates right 40%, direct lines 98-109Mass Save example: the whole-home rebate of $8,500 covers about 39% of the $22,000 average install, and the income-based $16,000 covers about 73%; this is one generous utility program, not a national picture.

• Qualifies Less 2021 (LBNL), cost of decarbonization retrofits partly 35%, direct lines 2532-2536LBNL: 71% of projects got rebates, with a median of $1,327 (21% of gross cost, mean 30%); electrification programs such as MA DOER Home MVP gave higher shares (37%), so rebates covered a meaningful but usually minority share.

Finding 7 (evidence) Incentives can make a heat pump cheaper upfront than a furnace and air conditioner

qualified qualify review qualified (qualify): 2 supporting (2 independent, strongest 0.4), 1 qualifying.

Canadian Climate Institute 2023, heat pumps pay off: In Canada heat pumps still cost more upfront despite rebates meant to lower capital costs, though some programs cover the full cost for low-income households. [partial: the verdict note limits its own scope ("though")]

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 40%, direct lines 633-641Applying $13,500 in rebates and credits shifts the incremental upgrade cost distribution left by that amount, and positive NPV rises to 84-90% of homes.
• Supports Edwards 2024, inequities in heat pump bill savings right 40%, direct lines 1829-1836Edwards' national medians already put a heat pump ($6,185) about $270 below a gas furnace plus AC, with over 80% overlap in cost distributions, so even small incentives would make it cheaper upfront.

• Qualifies Canadian Climate Institute 2023, heat pumps pay off partly 30%, partial lines 1482-1488In Canada heat pumps still cost more upfront despite rebates meant to lower capital costs, though some programs cover the full cost for low-income households.Applicability: the verdict note limits its own scope ("though")

Central conclusion The installed cost of a heat pump varies widely with the home, location and system type

supported approved supported (approved): 5 supporting (5 independent, strongest 0.6).

• Supports Less 2021 (LBNL), cost of decarbonization retrofits right 60%, direct lines 3046-3056LBNL (642 heat pump installs, median $8,027): costs vary substantially by heat pump type (ground source nearly 3x others), and the main cost drivers were capacity, type, floor area, efficiency, climate zone, program and stories.
• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 55%, direct lines 1286-1293Installed costs remain extremely variable even after corrections, driven by deferred maintenance, pricing strategies, market experience, integration with existing systems and local codes.
• Supports Edwards 2024, inequities in heat pump bill savings right 50%, direct lines 847-852Edwards reports that heat pump costs vary substantially across households and regions (typically $5,000-$10,000 for air-source) with home size, efficiency, climate, labour and installer experience.
• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results right 45%, direct lines 1969-1976Invoice totals varied widely by region and install complexity; upgrade costs alone ranged from $3,258 to $18,400 excluding the equipment, with major duct work tied to big price jumps.
• Supports Canadian Climate Institute 2023, heat pumps pay off right 40%, direct lines 1160-1168Heat pump equipment and installation costs vary widely in Canada, driven by limited suppliers in some regions and largely by brand, while gas systems vary less.

Finding 1 (evidence) A heat pump costs more to buy and install than a gas furnace

qualified qualify review qualified (qualify): 5 supporting (5 independent, strongest 0.55), 3 qualifying.

Less 2021 (LBNL), cost of decarbonization retrofits: LBNL: among 2,298 HVAC measures, fuel-fired heating systems averaged $3,000 less than heat pump installations (heat pump median $8,027 vs furnace median $5,025 in 2019 dollars), though heat pumps also provide cooling. [partial: the verdict note limits its own scope ("though")]

Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff: UK figures: an air-source heat pump installation averaged £12,713 before the grant (£5,213 after it) against an assumed £2,250 gas boiler; this compares with a UK boiler rather than a furnace. [partial: the verdict note limits its own scope ("rather than")]

CEE 2017, cold-climate heat pumps in Minnesota: Database costs put an air source heat pump at $2,994-$5,550 versus a propane furnace at $2,060-$3,900; a propane rather than gas furnace, and cost estimates rather than quotes. [partial: the verdict note limits its own scope ("rather than")]

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Deetjen 2021, private economics of US residential heat pumps right 55%, direct lines 804-807Replacing the existing heater with a minimum-efficiency heat pump costs an average of $6,600 more than replacing it like-for-like.
• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 55%, direct lines 451-457In cold-climate homes without central AC, the mean incremental cost over a furnace alone is $3,000, $15,000 and $18,000 for the three heat pump tiers.
• Supports RMI 2018, the economics of electrifying buildings right 50%, direct lines 158-161RMI finds that customers with existing gas service face higher upfront costs to retrofit to electric space and water heating than to install new gas devices.
• Supports Canadian Climate Institute 2023, heat pumps pay off right 50%, direct lines 1176-1197Canadian mid-case capital costs for a detached home: gas furnace about $3,750 to $6,030 against $7,570 to $13,260 for a standard ducted air-source heat pump and $14,840 to $19,880 for a cold-climate one (the heat pump also replaces the AC).
• Supports Edwards 2024, inequities in heat pump bill savings right 45%, direct lines 1838-1854From Edwards' national cost medians, a heat pump costs $6,185 while a gas furnace plus AC costs $6,455 and the AC alone $4,684, which puts the furnace alone at roughly $1,800, far below the heat pump.

• Qualifies Less 2021 (LBNL), cost of decarbonization retrofits right 55%, partial lines 3027-3031LBNL: among 2,298 HVAC measures, fuel-fired heating systems averaged $3,000 less than heat pump installations (heat pump median $8,027 vs furnace median $5,025 in 2019 dollars), though heat pumps also provide cooling.Applicability: the verdict note limits its own scope ("though")
• Qualifies Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff right 50%, partial lines 1821-1825UK figures: an air-source heat pump installation averaged £12,713 before the grant (£5,213 after it) against an assumed £2,250 gas boiler; this compares with a UK boiler rather than a furnace.Applicability: the verdict note limits its own scope ("rather than")
• Qualifies CEE 2017, cold-climate heat pumps in Minnesota right 30%, partial lines 1360-1375Database costs put an air source heat pump at $2,994-$5,550 versus a propane furnace at $2,060-$3,900; a propane rather than gas furnace, and cost estimates rather than quotes.Applicability: the verdict note limits its own scope ("rather than")

Finding 2 (evidence) A heat pump costs about the same as or less than a new furnace plus central air conditioner

disputed qualify review disputed (qualify): 4 supporting (4 independent, strongest 0.5), 1 contradicting, 1 qualifying.

Contradicted by Canadian Climate Institute 2023, heat pumps pay off: Compared with the gas-plus-AC baseline, the Canadian model finds heat pumps have higher upfront costs and win only on lifetime costs through efficiency and doubling as cooling.

Suggestion: Present the disagreement: name the studies on each side instead of stating one side as fact.

• Supports Edwards 2024, inequities in heat pump bill savings right 50%, direct lines 1829-1836Edwards' median installed heat pump ($6,185) costs about $270 less than a gas furnace plus central AC and about $90 more than resistance plus AC, with over 80% of the cost distributions overlapping.
• Supports RMI 2018, the economics of electrifying buildings right 45%, direct lines 455-460RMI says customers replacing both an AC and a furnace can save installation cost by choosing one heat pump for both jobs.
• Supports Less 2021 (LBNL), cost of decarbonization retrofits right 45%, direct lines 2288-2293LBNL archetypes: electric equipment (ductless heat pump plus heat pump water heater, $14,754) cost more than gas ($8,786), but the gap almost exactly equals the $5,930 cost of adding central air, implying no premium when comparing heating plus cooling like for like.
• Supports CEE 2017, cold-climate heat pumps in Minnesota right 40%, direct lines 1386-1408When both furnace and AC need replacing, a ccASHP with propane furnace backup costs only $370 more than a new furnace plus AC (payback under about one year).

• Contradicts Canadian Climate Institute 2023, heat pumps pay off wrong 35%, direct lines 842-846Compared with the gas-plus-AC baseline, the Canadian model finds heat pumps have higher upfront costs and win only on lifetime costs through efficiency and doubling as cooling.

• Qualifies Wilson 2024 (NREL and LBNL), heat pumps for all? partly 50%, direct lines 451-457In cold-climate ducted homes with central AC the minimum-efficiency heat pump averages $2,000 less than furnace plus AC, but medium and high-efficiency units cost $8,000 and $13,000 more.

Finding 3 (evidence) A heat pump costs about the same as a central air conditioner of the same capacity

disputed qualify review disputed (qualify): 1 supporting (1 independent, strongest 0.45), 1 contradicting, 1 qualifying.

Contradicted by Edwards 2024, inequities in heat pump bill savings: Edwards' national installed-cost medians at the same 36,000 BTU/h size are $6,185 for a heat pump and $4,684 for a central AC, about 30% apart.

Suggestion: Present the disagreement: name the studies on each side instead of stating one side as fact.

• Supports CEE 2017, cold-climate heat pumps in Minnesota right 45%, direct lines 1360-1391NREL cost database: heat pump $2,994 base / $5,550 efficient versus central AC $3,164 / $5,180; the report says heat pumps cost very similar to split AC of the same SEER.

• Contradicts Edwards 2024, inequities in heat pump bill savings wrong 45%, direct lines 1838-1854Edwards' national installed-cost medians at the same 36,000 BTU/h size are $6,185 for a heat pump and $4,684 for a central AC, about 30% apart.

• Qualifies Wilson 2024 (NREL and LBNL), heat pumps for all? partly 45%, direct lines 185-191Wholesale prices are only $200-$500 more than equivalent ACs, but installed heat pumps sized for heating often cost more because of larger capacity and electrical work.

Finding 4 (evidence) Choosing a heat pump instead of a new central air conditioner adds efficient heating for little extra cost

qualified qualify review qualified (qualify): 1 supporting (1 independent, strongest 0.4), 2 qualifying.

Wilson 2024 (NREL and LBNL), heat pumps for all?: Wholesale heat pumps cost only $200-$500 more than equivalent ACs, but heat pumps sized to fully electrify heating often cost more to install than AC plus furnace because of larger capacity, wiring and panel upgrades.

Edwards 2024, inequities in heat pump bill savings: Edwards' national installed-cost medians put a heat pump at $6,185 against $4,684 for a central AC of the same 36,000 BTU/h size, so it adds heating for about $1,500 (roughly 30%) more, which is modest but not trivial.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports CEE 2017, cold-climate heat pumps in Minnesota right 40%, direct lines 1360-1391Database costs put an efficient heat pump only about $370 above an equally efficient central AC ($5,550 vs $5,180), adding heating for little extra cost.

• Qualifies Wilson 2024 (NREL and LBNL), heat pumps for all? partly 50%, direct lines 185-191Wholesale heat pumps cost only $200-$500 more than equivalent ACs, but heat pumps sized to fully electrify heating often cost more to install than AC plus furnace because of larger capacity, wiring and panel upgrades.
• Qualifies Edwards 2024, inequities in heat pump bill savings partly 45%, direct lines 1838-1854Edwards' national installed-cost medians put a heat pump at $6,185 against $4,684 for a central AC of the same 36,000 BTU/h size, so it adds heating for about $1,500 (roughly 30%) more, which is modest but not trivial.

Finding 5 (evidence) The installed cost of a heat pump varies widely with the home, location and system type

supported approved supported (approved): 5 supporting (5 independent, strongest 0.6).

• Supports Less 2021 (LBNL), cost of decarbonization retrofits right 60%, direct lines 3046-3056LBNL (642 heat pump installs, median $8,027): costs vary substantially by heat pump type (ground source nearly 3x others), and the main cost drivers were capacity, type, floor area, efficiency, climate zone, program and stories.
• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 55%, direct lines 1286-1293Installed costs remain extremely variable even after corrections, driven by deferred maintenance, pricing strategies, market experience, integration with existing systems and local codes.
• Supports Edwards 2024, inequities in heat pump bill savings right 50%, direct lines 847-852Edwards reports that heat pump costs vary substantially across households and regions (typically $5,000-$10,000 for air-source) with home size, efficiency, climate, labour and installer experience.
• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results right 45%, direct lines 1969-1976Invoice totals varied widely by region and install complexity; upgrade costs alone ranged from $3,258 to $18,400 excluding the equipment, with major duct work tied to big price jumps.
• Supports Canadian Climate Institute 2023, heat pumps pay off right 40%, direct lines 1160-1168Heat pump equipment and installation costs vary widely in Canada, driven by limited suppliers in some regions and largely by brand, while gas systems vary less.

Finding 6 (evidence) Installing a heat pump can require costly electrical panel or service upgrades

supported approved supported (approved): 6 supporting (6 independent, strongest 0.55).

• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results right 55%, direct lines 1953-1976Across 21 installs, 38% needed a new panel, 24% a service line upgrade and 90% new breakers for auxiliary heat; electrical upgrades alone cost $3,750-$8,970 in labour and materials.
• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? right 50%, direct lines 849-859Panel, service wire, transformer and pole upgrades are estimated at $2,000-$30,000; only 50-60% of homes already have panels above 150 amps.
• Supports Less 2021 (LBNL), cost of decarbonization retrofits right 50%, direct lines 3676-3681LBNL: electrification projects are often burdened by panel upgrades where amperage is insufficient; a typical 200 A panel upgrade costs $1,954, plus dedicated circuits, and many homes with 50- or 100-amp service cannot support the new loads.
• Supports RMI 2018, the economics of electrifying buildings right 45%, direct lines 135-139RMI notes that many homes need additional electrical work to accommodate heat pumps.
• Supports Canadian Climate Institute 2023, heat pumps pay off right 45%, direct lines 770-773The model adds an electrical panel upgrade cost (set at $3,400 in Table 2) for 1940 and 1980 homes where one is likely needed, including all heat-pump-with-electric-backup scenarios.
• Supports Edwards 2024, inequities in heat pump bill savings right 40%, direct lines 885-888Edwards lists electrical panel and other home upgrades among the upfront costs that can block heat pump adoption.

Central conclusion The evidence does not support "Heat pumps need a home to be well insulated to work economically"

qualified qualify review qualified (qualify): 2 supporting (2 independent, strongest 0.45), 2 qualifying.

RMI 2018, the economics of electrifying buildings: RMI says electric space heating is less suitable in inefficient buildings, especially in cold climates, yet its leaky retrofit homes in mild Houston and Oakland still had lower heat pump energy costs.

Less 2021 (LBNL), cost of decarbonization retrofits: LBNL: the cheapest high-savings approach paired heat pumps with only typical weatherization (attic, cavity fill, air sealing), while aggressive envelope upgrades cost double for modest marginal energy cost savings; basic weatherization helps, deep insulation was not needed.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? wrong 45%, direct lines 522-529Envelope upgrades only slightly raise the share with positive NPV in cold climates and generally lower it in warmer ones; high-efficiency units avoid bill increases in 95% of homes without them.
• Supports CEE 2017, cold-climate heat pumps in Minnesota wrong 30%, direct lines 1271-1299Modeled savings depended mostly on house type and were largest in leaky/large homes (about $1,033/yr vs an 80% furnace in Duluth versus about $111 for a passive house); the table is garbled in extraction.

• Qualifies RMI 2018, the economics of electrifying buildings partly 45%, direct lines 972-977RMI says electric space heating is less suitable in inefficient buildings, especially in cold climates, yet its leaky retrofit homes in mild Houston and Oakland still had lower heat pump energy costs.
• Qualifies Less 2021 (LBNL), cost of decarbonization retrofits partly 25%, direct lines 1931-1945LBNL: the cheapest high-savings approach paired heat pumps with only typical weatherization (attic, cavity fill, air sealing), while aggressive envelope upgrades cost double for modest marginal energy cost savings; basic weatherization helps, deep insulation was not needed.

• Other Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff unknown, direct lines 1235-1244The UK study found no interaction between EPC rating and the heat pump's electricity increase, but because the grant required loft and cavity insulation, few homes in the sample were poorly insulated, so it cannot test the claim.

Finding 1 (evidence) The evidence does not support "Heat pumps need a home to be well insulated to work economically"

qualified qualify review qualified (qualify): 2 supporting (2 independent, strongest 0.45), 2 qualifying.

RMI 2018, the economics of electrifying buildings: RMI says electric space heating is less suitable in inefficient buildings, especially in cold climates, yet its leaky retrofit homes in mild Houston and Oakland still had lower heat pump energy costs.

Less 2021 (LBNL), cost of decarbonization retrofits: LBNL: the cheapest high-savings approach paired heat pumps with only typical weatherization (attic, cavity fill, air sealing), while aggressive envelope upgrades cost double for modest marginal energy cost savings; basic weatherization helps, deep insulation was not needed.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Wilson 2024 (NREL and LBNL), heat pumps for all? wrong 45%, direct lines 522-529Envelope upgrades only slightly raise the share with positive NPV in cold climates and generally lower it in warmer ones; high-efficiency units avoid bill increases in 95% of homes without them.
• Supports CEE 2017, cold-climate heat pumps in Minnesota wrong 30%, direct lines 1271-1299Modeled savings depended mostly on house type and were largest in leaky/large homes (about $1,033/yr vs an 80% furnace in Duluth versus about $111 for a passive house); the table is garbled in extraction.

• Qualifies RMI 2018, the economics of electrifying buildings partly 45%, direct lines 972-977RMI says electric space heating is less suitable in inefficient buildings, especially in cold climates, yet its leaky retrofit homes in mild Houston and Oakland still had lower heat pump energy costs.
• Qualifies Less 2021 (LBNL), cost of decarbonization retrofits partly 25%, direct lines 1931-1945LBNL: the cheapest high-savings approach paired heat pumps with only typical weatherization (attic, cavity fill, air sealing), while aggressive envelope upgrades cost double for modest marginal energy cost savings; basic weatherization helps, deep insulation was not needed.

• Other Bernard 2024 (NBER), UK heat pumps and a time-of-use tariff unknown, direct lines 1235-1244The UK study found no interaction between EPC rating and the heat pump's electricity increase, but because the grant required loft and cavity insulation, few homes in the sample were poorly insulated, so it cannot test the claim.

Finding 2 (evidence) Air sealing and insulation should come before or alongside a heat pump

qualified qualify review qualified (qualify): 5 supporting (5 independent, strongest 0.45), 1 qualifying.

Wilson 2024 (NREL and LBNL), heat pumps for all?: ResStock modelling: envelope upgrades cut bill-increase cases (39% to 19% with minimum-efficiency units) and shrink heat pump capacity, but add about $9,000 so the net upfront cost rises, raising positive-NPV share slightly in cold climates and lowering it in warmer ones.

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports RMI 2018, the economics of electrifying buildings right 45%, direct lines 972-977RMI's modelled new homes use 55-67% less heating energy than the leaky retrofits, and it recommends bundling insulation and sealing with heat pumps in inefficient homes, especially in cold climates (a recommendation, not a tested sequence).
• Supports Guidehouse 2025, review of the MA and CT heat pump study right 30%, direct lines 345-356The MA/CT metering study lists weatherizing the home before heat pump installation as a program best practice for performance consistent with its savings.
• Supports DOE 2026, Program Notice 26-2 on electric home rebates right 30%, direct lines 157-160DOE now requires homes to use rebates for insulation and air sealing before heating and cooling upgrades unless already sealed to a state level, 'to ensure upgrades do not increase consumer energy costs'; this is a program rule, not measured evidence.
• Supports CEE 2017, cold-climate heat pumps in Minnesota right 25%, direct lines 1600-1612The policy analysis recommends that a ccASHP program include insulation and air-sealing incentives alongside the heat pump; a recommendation, not a measured result.
• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results right 20%, direct lines 2228-2237The authors call weatherizing before replacing a furnace with a heat pump best practice for cost savings and comfort, but the study homes were not weatherized and the effect was not tested.

• Qualifies Wilson 2024 (NREL and LBNL), heat pumps for all? partly 50%, direct lines 465-479ResStock modelling: envelope upgrades cut bill-increase cases (39% to 19% with minimum-efficiency units) and shrink heat pump capacity, but add about $9,000 so the net upfront cost rises, raising positive-NPV share slightly in cold climates and lowering it in warmer ones.

Central conclusion Studies disagree on whether real-world heat pump efficiency is lower than manufacturers' rated figures

supported approved supported (approved): no direct support, 1 contradicting, 2 qualifying.

Contradicted by CEE 2018, ducted and ductless cold-climate heat pumps in the field: Installed furnace efficiencies ran below rating, but the heat pumps' rated COPs at 17F were within the measurement error of the field data at all sites; small sample.

• Contradicts CEE 2018, ducted and ductless cold-climate heat pumps in the field wrong 35%, direct lines 413-419Installed furnace efficiencies ran below rating, but the heat pumps' rated COPs at 17F were within the measurement error of the field data at all sites; small sample.

• Qualifies Cadmus 2016, ductless mini-split evaluation (MA and RI) right 65%, partial lines 1073-1079Directly metered seasonal efficiencies of 152 mini-splits fell below their rated values on average, though some units met or exceeded their ratings; the evaluators blame usage patterns and test conditions, not misrated equipment.Applicability: the verdict note limits its own scope ("though")
• Qualifies Winkler 2023 (NREL), cold-climate heat pump field validation partly 55%, direct lines 241-252Steady-state COPs at 7 of 12 sites matched manufacturer values while the other 5 were lower, and measured capacity aligned with maximum rated capacity at only 5 of 12 sites.

Finding 1 (evidence) Real-world heat pump efficiency is lower than manufacturers' rated figures

unsupported blocked unsupported (blocked): no direct support, 1 contradicting, 2 qualifying.

Contradicted by CEE 2018, ducted and ductless cold-climate heat pumps in the field: Installed furnace efficiencies ran below rating, but the heat pumps' rated COPs at 17F were within the measurement error of the field data at all sites; small sample.

Suggestion: The evidence points the other way; reverse the claim or report it as what the sources say.

• Contradicts CEE 2018, ducted and ductless cold-climate heat pumps in the field wrong 35%, direct lines 413-419Installed furnace efficiencies ran below rating, but the heat pumps' rated COPs at 17F were within the measurement error of the field data at all sites; small sample.

• Qualifies Cadmus 2016, ductless mini-split evaluation (MA and RI) right 65%, partial lines 1073-1079Directly metered seasonal efficiencies of 152 mini-splits fell below their rated values on average, though some units met or exceeded their ratings; the evaluators blame usage patterns and test conditions, not misrated equipment.Applicability: the verdict note limits its own scope ("though")
• Qualifies Winkler 2023 (NREL), cold-climate heat pump field validation partly 55%, direct lines 241-252Steady-state COPs at 7 of 12 sites matched manufacturer values while the other 5 were lower, and measured capacity aligned with maximum rated capacity at only 5 of 12 sites.

Finding 2 (evidence) The evidence does not support "Well-installed air-source heat pumps reach seasonal efficiencies around four in real homes"

qualified qualify review qualified (qualify): 1 supporting (1 independent, strongest 0.5), 1 qualifying.

CEE 2017, cold-climate heat pumps in Minnesota: Heat-pump-only cycle COPs were about 1.3 at 10F and 3.0-4.5 only in shoulder seasons; weather-normalized annual heat-pump-only COPs were 2.51-2.78 at three Minnesota sites, well short of four. [partial: the verdict note limits its own scope ("only in")]

Suggestion: Add the qualification: the evidence supports a narrower or weaker statement.

• Supports Winkler 2023 (NREL), cold-climate heat pump field validation wrong 50%, direct lines 1823-1825Across 12 monitored central heat pumps in cold northwest US climates, seasonal compressor-based heating COP averaged 2.5 (1.7-3.5) and system COP 1.5-2.9; none approached four.

• Qualifies CEE 2017, cold-climate heat pumps in Minnesota wrong 40%, partial lines 1143-1157Heat-pump-only cycle COPs were about 1.3 at 10F and 3.0-4.5 only in shoulder seasons; weather-normalized annual heat-pump-only COPs were 2.51-2.78 at three Minnesota sites, well short of four.Applicability: the verdict note limits its own scope ("only in")

Central conclusion Studies disagree on whether contractors routinely oversize heating and cooling equipment

supported approved supported (approved): 1 supporting (1 independent, strongest 0.4), 1 contradicting, 3 qualifying.

Contradicted by Winkler 2023 (NREL), cold-climate heat pump field validation: Heating-load audits found the heat pumps at Sites 5 and 8 very undersized (under 40% of design load) and Sites 10, 11 and 13 also undersized; oversizing was the exception in this 12-site northwest sample.

• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results right 40%, direct lines 1943-1951The furnaces being replaced (mostly 66-100 kBtu/h gas) tended to be well above the homes' Manual J heating loads, so the cold-climate heat pumps installed were much smaller; this covers 21 sites and furnaces only.

• Contradicts Winkler 2023 (NREL), cold-climate heat pump field validation wrong 30%, direct lines 1131-1139Heating-load audits found the heat pumps at Sites 5 and 8 very undersized (under 40% of design load) and Sites 10, 11 and 13 also undersized; oversizing was the exception in this 12-site northwest sample.

• Qualifies Cadmus 2016, ductless mini-split evaluation (MA and RI) partly 40%, direct lines 3103-3127Program mini-splits averaged about 2.1-2.6 times the Manual J cooling load of the spaces they served, but for heating nearly half were undersized, so the oversizing was one-directional and probably came from sizing to heating.
• Qualifies Less 2021 (LBNL), cost of decarbonization retrofits partly 30%, direct lines 2842-2853LBNL: fuel-fired heating equipment was sized far larger (typically 356 ft2 per ton) than heat pumps or air conditioners (543 and 555 ft2 per ton), suggesting generous furnace sizing, while many heat pumps were sized aggressively small; no load calculations were compared.
• Qualifies Wilson 2024 (NREL and LBNL), heat pumps for all? partly 25%, direct lines 763-765The authors cite overly conservative equipment sizing as a cause of high upfront costs to be addressed with sizing guidance, without measuring how common it is.

Finding 1 (evidence) Contractors routinely oversize heating and cooling equipment

disputed qualify review disputed (qualify): 1 supporting (1 independent, strongest 0.4), 1 contradicting, 3 qualifying.

Contradicted by Winkler 2023 (NREL), cold-climate heat pump field validation: Heating-load audits found the heat pumps at Sites 5 and 8 very undersized (under 40% of design load) and Sites 10, 11 and 13 also undersized; oversizing was the exception in this 12-site northwest sample.

Suggestion: Present the disagreement: name the studies on each side instead of stating one side as fact.

• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results right 40%, direct lines 1943-1951The furnaces being replaced (mostly 66-100 kBtu/h gas) tended to be well above the homes' Manual J heating loads, so the cold-climate heat pumps installed were much smaller; this covers 21 sites and furnaces only.

• Contradicts Winkler 2023 (NREL), cold-climate heat pump field validation wrong 30%, direct lines 1131-1139Heating-load audits found the heat pumps at Sites 5 and 8 very undersized (under 40% of design load) and Sites 10, 11 and 13 also undersized; oversizing was the exception in this 12-site northwest sample.

• Qualifies Cadmus 2016, ductless mini-split evaluation (MA and RI) partly 40%, direct lines 3103-3127Program mini-splits averaged about 2.1-2.6 times the Manual J cooling load of the spaces they served, but for heating nearly half were undersized, so the oversizing was one-directional and probably came from sizing to heating.
• Qualifies Less 2021 (LBNL), cost of decarbonization retrofits partly 30%, direct lines 2842-2853LBNL: fuel-fired heating equipment was sized far larger (typically 356 ft2 per ton) than heat pumps or air conditioners (543 and 555 ft2 per ton), suggesting generous furnace sizing, while many heat pumps were sized aggressively small; no load calculations were compared.
• Qualifies Wilson 2024 (NREL and LBNL), heat pumps for all? partly 25%, direct lines 763-765The authors cite overly conservative equipment sizing as a cause of high upfront costs to be addressed with sizing guidance, without measuring how common it is.

Central conclusion The evidence does not support "Heat pumps need more maintenance or repairs than gas furnaces"

supported qualify review supported (qualify): 1 supporting (1 independent, strongest 0.2).

Suggestion: The strongest support is weak (0.2); say "one study suggests" or find stronger evidence.

• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results wrong 20%, direct lines 1837-1849Seven homeowners needed contractor visits, mostly minor control, sensor or condensate issues fixed in one visit, and the team judged this not atypical compared with their previous systems; the units were prototypes and the sample was small.

Finding 1 (evidence) The evidence does not support "Heat pumps need more maintenance or repairs than gas furnaces"

supported qualify review supported (qualify): 1 supporting (1 independent, strongest 0.2).

Suggestion: The strongest support is weak (0.2); say "one study suggests" or find stronger evidence.

• Supports PNNL 2025, DOE Cold Climate Heat Pump Challenge field results wrong 20%, direct lines 1837-1849Seven homeowners needed contractor visits, mostly minor control, sensor or condensate issues fixed in one visit, and the team judged this not atypical compared with their previous systems; the units were prototypes and the sample was small.