StraightWatt

Data

Data index

Every number on this site comes from one of 107 records held in 7 files. All of them are below, with the source each was read from and the date it was read. 36 carry a caveat and 14 are recorded but not published, because a record on file is not the same as a record we are willing to put a page on.

Last verified 14 October 2026. Oldest record on file 4 September 2026. 35 distinct sources. Each dataset can be downloaded whole, as JSON or CSV, from its own page.

The Solar Economics Index

One thing is built on top of these records rather than sitting among them: a monthly ranking of the states we publish by how long solar takes to repay itself, cut as a dated edition and kept for ever. It exists to be quoted, so each edition is frozen the day it is cut and each state comes with a sentence written for you.

The datasets

DatasetRecordsVerifiedWhat it decides
Electricity prices
cents per kilowatt-hour
162026-10-14Every bill in the model, in every year of the projection.
Solar yield
kilowatt-hours per kilowatt per year
252026-09-05System sizing and every year of production, including the size pages.
Installed cost
dollars per watt, before incentives
112026-09-05Gross price, net cost, payback and every 25-year figure.
Export rules
export credit as a share of the retail rate
132026-09-05The bill with solar, in every scenario, and the whole battery case.
State incentives
credits, rebates and payments for production
102026-09-05Net cost, and the payments that arrive over the years after installation.
Federal policy
credit percentage, by year and by who owns the system
92026-09-05Every net cost, and the difference between buying and leasing in 2026.
Modelling assumptions
the numbers we chose, not the numbers we read
232026-09-04Everything. These are the values a reader is most entitled to disagree with.

How fresh it is

A source read once and never revisited is a source that will eventually be wrong, so the read date travels with every record rather than sitting in a footer. The oldest file here was verified 4 September 2026; the newest 14 October 2026.

Two of these age differently from the rest. Electricity prices move monthly and are refreshed from an API by a script, so they are the least likely to be stale. Export rules move when a commission issues an order, which can happen at any time and is announced nowhere we can subscribe to — those are the records most likely to be out of date without us knowing, and the ones where a reader's correction is worth the most.

What is missing

A data index that lists only what it has is marketing. These are the gaps we know about, taken from the files that record them rather than from memory:

HI
Not served. Net metering closed to new customers in 2015; export compensation must come from the HECO tariff / PUC docket for the current DER program, and installed cost from a primary source, before HI gets a record here. Do not add a value without a source.
CO, GA, NC, NV
Not served. No export rule and no installed-cost record on file; the same two fallbacks that produced the wrong Hawaii number.
Illinois, Pennsylvania: no reference price
IL and PA have no LBNL state median: the slide shows a state only with at least 20 observations.
The export ceiling
The engine credits every exported kilowatt-hour at export_ratio with NO annual ceiling. That is correct for a system sized to the household's own use, where annual production is about equal to annual consumption and the year-end surplus is near zero - which is how the engine sizes by default. It is NOT correct for a system that generates materially more than the home uses: several states settle the year-end surplus at a lower rate than the in-year export credit (see annual_surplus), and some utilities will not approve a system larger than the customer's recent usage at all (see sizing_cap). Do not use these figures for a deliberately oversized system.

Every state we publish has its own rate, production, cost and export-compensation record. A state missing any one of them is not served at all, and the build fails rather than fall back to a national average. Why we publish eleven states and not fifty.

Using this data

Take it. Each dataset page carries a JSON file — the exact file the engine reads, metadata and all — and a CSV of the same records flattened for a spreadsheet.

What you may reuse
Our compilation: which records we chose, the values we read, the source URLs, the read dates and the notes. Use them freely, with a link back so a reader of yours can check them too.
What we cannot give you
The underlying data. We do not own an EIA series, a laboratory report or a marketplace's price average, and we cannot license to you what is not ours. Every record carries the source it came from; that source's own terms apply, and following the link is always the more authoritative move.
What you should not assume
That a record is current. It was true on the date beside it and no later. Where we have been wrong, the corrections log says so, with what changed and when.

How a record gets here

  1. Read from a primary source. A commission order, a tariff sheet, a statute, a tax department page, or an API run by the agency that collects the series. Never an aggregator, and never another calculator.
  2. Recorded with its source and the date. Both travel with the value in the same record, so a figure can never appear on a page without them.
  3. One source per dataset, across every state. A file that takes the best available source per state produces a ranking that measures the sources rather than the market.
  4. Published only if it is complete. A missing or unmodellable record raises an error, and an error fails the build. Understating a saving is the same failure as overstating one.

Rates: EIA 2026-07, refreshed by script. Production: PVWatts v8 (verified). Federal policy verified 2026-09-05.

Published 6 September 2026. Last verified 14 October 2026. A figure here that looks wrong is the mail we most want: tell us, and we publish corrections. The formulas that consume these records are on the methodology page.