Do expensive electricity prices actually drive solar adoption?

The claim being tested

“Solar makes most sense where electricity is expensive” is the industry’s standard framing, and it is intuitive: the value of a solar kilowatt-hour is the retail kilowatt-hour it displaces, so the higher the retail price, the better the arithmetic. It is repeated in sales material, in policy submissions and in press coverage, almost always without a number attached.

It is testable. EIA-861 gives us, for every state in the same year, the average residential price and the number of households with a net-metered solar system. We computed both from the same filings and correlated them, then did the same for five other candidate explanations, so the price result could be judged against alternatives rather than in isolation.

Why the two numbers differ so much

Hawaii pays 42.86¢/kWh, 2.6 times the national average, and 22.0% of its households have solar. California pays 31.97¢/kWh and sits at 14.7%. Both are extreme on both axes, which is precisely the configuration that inflates a Pearson correlation.

The Spearman rank correlation, which is not sensitive to how far out an outlier sits, is 0.466 on the full set, already much weaker than the Pearson figure, and a hint that the raw 0.677 was doing more work than it should. Removing the two states brings the two statistics into agreement at around 0.319, which is the honest read: a real but modest relationship.

We are not saying Hawaii and California are wrong or should be discarded. They are real states and their adoption is real. The point is narrower and more useful: you cannot generalise from them. A pitch that reasons “electricity is expensive here, so adoption will follow” is reasoning from two jurisdictions with two decades of aggressive solar policy behind them.

The same plot without them

Remove the two and the cloud loses most of its slope. States between roughly 12¢ and 24¢/kWh show adoption anywhere from a fraction of a per cent to nearly 7%, with price explaining very little of the difference.

The four highlighted points make the case. Connecticut and Massachusetts pay 28.8¢ and 29.4¢ and reach 6.7% and 6.5%. Utah and New Mexico pay 12.2¢ and 14.2¢ (well below the national average) and reach 6.4% and 6.5%. Roughly the same adoption at half the price.

Five rival explanations, tested against the same data

A weak result for price is only interesting if the alternatives are tested too. We ran the same cross-section against five other variables available from the same federal form: the share of households on a time-varying tariff, interval metering penetration, grid reliability, annual household consumption, and average installed system size.

None beats price, and the strongest after price is annual consumption per household (kwh) at r = -0.462, which is negative. States whose households consume more electricity have less rooftop solar, not more. That is not a paradox once you see what it is measuring: high-consumption states are overwhelmingly hot, cheap-power southern states, and the variable is picking up the same regional pattern from a different angle.

Grid reliability comes out at r = -0.205, also negative. The intuition that people buy solar because the power goes out is not supported: the states with the worst outage records are southern and coastal, and they are near the bottom of the adoption table.

The states that break the pattern

If price drove adoption, the expensive-state table would be a list of solar leaders. It is not. Alaska pays 24.82¢/kWh and sits at 0.91%; Michigan pays 19.30¢/kWh and sits at 0.52%; Pennsylvania pays 17.77¢/kWh and sits at 1.47%; Wisconsin pays 17.18¢/kWh and sits at 0.65%. These are among the most expensive electricity markets in the country with adoption at or below the national average.

The reverse case is just as clear. Arizona pays 14.91¢/kWh and has reached 10.1% of households; Colorado pays 14.92¢/kWh and has reached 7.0% of households; New Mexico pays 14.20¢/kWh and has reached 6.5% of households. Cheap power has not stopped them.

What separates the two groups is not price. It is whether the state built a functioning residential solar policy: net metering that pays retail value, an incentive that a household or a third-party owner can actually claim, and an interconnection process that completes. Price sets the ceiling on what solar can be worth. Policy decides whether anyone gets to collect it.

The price map, for reference

Residential electricity price is itself strongly regional: expensive in New England, Hawaii, Alaska and California; cheap across the South, the Plains and the Pacific Northwest. Comparing this with the adoption map is the quickest way to see how imperfectly the two overlap.

US residential prices rose from 12.52¢/kWh in 2014 to 16.48¢ in 2024, 32% in nominal terms, with the sharpest increases in 2022 and 2023.

The measurement problem nobody mentions

There is a structural reason to expect this correlation to be weak, and it is worth stating because it cuts against our own result as much as against the industry’s.

Average price is not marginal price, and solar economics run on the marginal rate: the price of the last kilowatt-hour, often a top tier or an on-peak period well above the average. A state with steeply tiered residential tariffs offers better solar economics than its average price implies. California is the extreme case: 36% of its households are on a time-varying tariff.

Nationally, though, only 11.1% of residential customers are on any time-varying tariff, 15,832,411 households. For the remaining 88.9%, the average price is a reasonable proxy for the marginal one, because there is no time variation to capture. So the measurement problem is real but bounded: it distorts a handful of states badly and most states hardly at all.

This is one reason we do not present the weak correlation as proof that price is irrelevant. It is evidence that price alone is a poor predictor across states, not that price does not matter to a household.