Every number on this site traces back to one of the public datasets listed below. This page records the source, the licensing terms, and the date of the snapshot we are currently displaying. For the live mechanism behind those snapshots, see the data status page.
The EIA, an independent statistical agency within the U.S. Department of Energy, is the
source of our residential electricity rates by state. We read the
Open Data API v2
electricity/retail-sales series — residential sector, annual average price.
scripts/sync_eia_rates.php pulls
the series and rewrites config/state_data.php, which is then committed. The
series is annual, so the figure does not change between EIA's yearly releases; a more
frequent sync would return the same number.This page previously credited Electric Power Monthly Table 5.6.A, described the values as hand-keyed, and called the API integration "not yet built". All three were out of date: the sync went live in July 2026 and reads a different, annual series. The monthly table is a real EIA product, but it is not the one behind these numbers.
NREL, operated by the Alliance for Sustainable Energy under U.S. Department of Energy contract, publishes the National Solar Radiation Database. We derive state-level peak-sun-hour averages from NSRDB grid cells, weighted toward population centres.
The Residential Clean Energy Credit is codified at 26 U.S.C. § 25D and described in plain language on the IRS's Residential Clean Energy Credit page. Current IRS guidance says the credit is not available for new residential clean-energy expenditures after December 31, 2025.
DSIRE is operated by the N.C. Clean Energy Technology Center at North Carolina State University and is the most comprehensive single source of state, local, utility, and federal renewable-energy incentives in the United States. The state pages on this site render DSIRE programs filtered to residential solar; the authoritative current list is DSIRE itself.
scripts/dsire_import.php still works;
it simply has nothing to read. The states listed below were read by hand from the public
per-state pages at programs.dsireusa.org.
scripts/refresh_dsire.* wrappers ship with the code but are not scheduled and do not work, because they download from an archive URL that is no longer published. Scheduling them would fetch nothing and quietly report success.Our cost per watt comes from Lawrence Berkeley National Laboratory's Tracking the Sun / U.S. Distributed Solar and Storage data update, national residential PV-only systems. The default of $2.98/W DC is the cash-purchase median, and the $2.49–$4.14/W range is the 20th and 80th percentiles of that same cash-purchase sample. See emp.lbl.gov/tracking-the-sun.
This is chosen deliberately over the alternatives below, because it is a record of prices actually paid on hundreds of thousands of real transactions — not a modelled "should cost" and not a sample of quotes from one marketplace. The previous default of $2.90/W with a $3.30/W top end understated the upper tail badly: four in ten cash buyers pay more than $3.30/W, and the real 80th percentile is $4.14.
These are useful for sanity-checking the figure above and are worth knowing about, but none of them feeds a number on this site.
Since September 2026 the calculator values exported energy separately from energy used in the house, because crediting every kilowatt-hour at the retail rate is no longer true in much of the country. Two inputs drive that, and both are sourced.
Read from the tariff or the regulator, state by state, and recorded with the date checked in
config/export_models.php (the rates we model) and
config/export_regimes.php (how every state is classified). Currently modelled:
Currently Arizona and Utah.
Where a state's export credit cannot be reduced to a single figure — an hourly avoided-cost schedule, a rate set separately by each utility, or no statewide rule at all — we do not model it and we do not invent a factor. The payback shown for those states stays a ceiling and is labelled one. An admitted gap beats a fabricated number.
How much of its own generation a household uses as it is produced, rather than exporting it. We use a published range, not a single assumed figure: 28–53% from Forrester, Barbose & Miller, iScience 25(8):104714 (2022), Lawrence Berkeley National Laboratory (emp.lbl.gov), based on metered load from roughly 1,800 US homes across six utilities, with PV generation simulated onto that real consumption — the households themselves had no solar, so the overlap is modelled rather than observed. It is used in preference to a rule of thumb because its central case is a system sized to generate 100% of annual household consumption, which is exactly how this calculator sizes one. Step 6 of the methodology sets out the caveats that come with it, including that the underlying load data is from 2012–2013.
The 0.5%/yr panel degradation rate is the median value from the warranty datasheets of LG (NeON 2), Q-Cells (Q.PEAK DUO), REC (Alpha Pure), and SunPower (Maxeon) monocrystalline modules across the 2020–2024 product generations. Manufacturer warranties bound degradation by the worse of (a) the cited annual rate or (b) a 25-year end-of-life output floor (typically 84–92% of nameplate).
The 0.80 performance ratio is the long-standing NREL PVWatts default for well-designed residential rooftop systems.
Where we present content derived from DSIRE, the derivative content (our state-incentive summaries) inherits the CC BY-SA 4.0 license. All other original content on this site — methodology pages, guides, glossary, and the calculator's source code — is published under the same license. You may reuse it with attribution, including commercially, provided you license derivatives under the same terms.
EIA, NREL, IRS, DSIRE, SEIA, EnergySage, Wood Mackenzie, and equipment-manufacturer names referenced on this site are the property of their respective owners. References are nominative — used to identify the source of the data we cite — and do not imply affiliation, endorsement, or sponsorship.
If you find an out-of-date figure or a citation that fails to support the claim it references, please flag it via our contact form. Data corrections are treated as priority.
Four inputs. Sixty seconds. Every number traceable to a public source.
May we count this visit? If you allow it, Google Analytics counts calculator use among visitors who agree — so it measures that group, not our total traffic. It sets a cookie and Google receives your IP address. It is not loaded unless you agree, and the calculator works either way — we do not send it your bill, ZIP code or results. What we collect.