Solar when you can’t pay cash: what a lease or PPA is really worth in PG&E territory

Our PG&E study modelled 56 homes across PG&E territory and found solar still pays there. It priced every home on PG&E’s standard residential rate, which is the right basis for a general sample: most PG&E households are on it.

This study asks a narrower question of the same data. Twenty-nine of those homes sit in census tracts where the median household income is below 80% of the county median — the usual low-income line. A household in that bracket may well be enrolled in CARE or FERA, paying a discounted rate, and may have no realistic route to a cash purchase. Neither of those things was in scope before. Both change the answer.

So we re-run those 29 homes on the rates and the financing that bracket actually faces. No new roofs were modelled; the measured shading, orientation and production are the ones already archived. What changes is the price of electricity, the way the system is paid for, and the question being asked.

1. CARE is the variable nobody models

PG&E’s California Alternate Rates for Energy programme discounts electricity by an at least 35% for income-qualified households — PG&E’s own published figure is “20% or more on gas and 35% or more on electricity”; FERA, for slightly higher incomes and households of three or more, discounts by 18% (CPUC).

Solar does not pay you. It saves you the cost of electricity you no longer buy. So a discount on that electricity is a discount on solar’s value, one for one:

Rate the household actually pays	$/kWh	25-year value, cash purchase
Standard E-1	$0.3263	$50,177
FERA (−18%)	$0.2676	$38,140
CARE (−35%)	$0.2121	$26,538

Median across the 29 homes. CARE cuts solar’s value by 47%. And because the programme guarantees “35% or more”, a household whose real discount is larger gets less than this, not more: every figure we show a CARE household is an upper bound. This is not an argument against solar for low-income households — $26,538 is still real money. It is an argument against quoting them the other number, which is what almost every calculator does, including, until this study, our own.

2. Cash and loans: mostly theoretical

The median system here costs $14,080 after incentives. Since the Section 25D residential credit expired for expenditures after 31 December 2025, there is no federal credit on a purchase, so that figure is simply the price. For a household at 80% of area median income, it is not an available option.

The industry’s answer is a no-money-down loan. We modelled a realistic subprime one: 20-year term, 8.99% APR, and a 25% dealer fee inside the price. Berkeley Lab puts dealer fees at 5–50% of the up-front price; the weaker the credit, the longer the term and the larger the fee.

Route, CARE household	Year 1, net	25-year value
Cash	—	$26,538
Subprime loan, no money down	−$628	$3,484

The loan turns a $26,538 asset into a $3,484 one and costs the household money in year one. The dealer fee is the mechanism: it is charged as a percentage of the price, it is never itemised on the quote, and it scales with exactly the weakness that made the borrower take the loan. It is a regressive product, sold hardest to the people it hurts most.

Credit is also a gate in its own right. Solar loans generally want a FICO around 700; leases and PPAs generally start near 650, some lenders from the low 600s. Third-party ownership is a lower bar, not an open door.

3. What a lease or PPA actually costs the company

Here is the part the industry does not publish. A third-party system is an investment: the fund pays for the hardware, claims the 48E commercial credit the homeowner can no longer claim, takes the depreciation, and sells the output. There is a price below which that investment stops working — a floor. Our calculator solves for it directly, because the same engine prices the fund’s side of the deal as prices yours.

For the median CARE household in this sample:

	What the fund needs	What is typically pitched	Multiple
PPA	8.5¢/kWh	18.0¢/kWh	2.1×
Lease	$45/month	—	—

“Typically pitched” is the industry’s own standard offer: 15% below your utility rate. For a CARE household that lands at 18¢ — about just over twice what the system costs to finance. The discount is real. It is just measured against the wrong number.

4. The line, and which side of it you are on

Pricing every contract as a multiple of that floor gives the whole answer at once:

Priced at	Lease	PPA	25-yr value	Homes it works for
1.00× floor	$45/mo	8.5¢	+$11,600	29 of 29
1.25×	$57/mo	10.7¢	+$6,700	28 of 29
1.50×	$68/mo	12.8¢	+$2,300	22 of 29
2.00×	$91/mo	17.1¢	−$7,400	0 of 29
2.50×	$113/mo	21.4¢	−$17,000	0 of 29

The collapse between 1.5× and 2× is the finding. It is not gradual: a lease that works for 22 of 29 homes at $68 a month works for none of them at $91. The median household goes from keeping about $2,300 to losing about $7,400. A 35% difference in monthly payment is the whole difference between a good decision and a bad one, and nothing on the contract tells you which one you are being handed.

Why the cliff is so sharp here: under NEM 3.0 exported power earns roughly a quarter of retail, but a PPA charges for every kWh the system makes, exported or not. The household pays full price for power worth a quarter. That gap is fixed, so once the contract price rises past a certain point there is nothing left to absorb it.

5. Illinois writes the test into law. California does not.

Illinois Solar for All serves households at or below 80% of area median income — the same bar this study samples. Where one of its residential contracts includes monthly payments to the solar company, the programme requires those costs to be outweighed by the value the panels generate.

That is the test in this study, enforced as a rule. We derived the line from 29 roofs and a cash-flow model; Illinois arrived at the same place from consumer-protection law. A household in PG&E territory signing a private lease or PPA gets no such guarantee, which is the entire reason we published the numbers above.

Illinois also shows what it looks like when it works. Jaouad Barmaki, a father of five in Champaign, expects to save more than $14,000 over fifteen years. In California, GRID Alternatives has served over 10,000 low-income families, 36 MW in disadvantaged communities, with an expected $290 million in household savings. Those systems were donated outright — a different and better product than anything priced here, and the first thing any household in this bracket should check eligibility for.

6. Method, and what this does not tell you

• Sample. The 29 homes from our September 2026 PG&E study whose census tract median household income is below 80% of the county median (ACS 2023 5-year B19013_001E). Tract income is a neighbourhood proxy, not a household test. A comfortable family in a modest tract is still comfortable.
• Roofs. Re-used from that study: measured shading, orientation and tilt from Google’s 3-D model, production from NREL PVWatts. No roof was re-modelled, so this study cost nothing in new measurement.
• Engines. Ownership from the same cash-flow model behind our calculator; lease and PPA from the same dual-ledger model, which prices the fund’s side and the household’s side in one pass. Every figure here is reproducible in the Solar Design Studio.
• Assumptions. 25 years, NEM 3.0 exports at 25% of retail, 2.5% utility inflation, 2.9% contract escalator, 0.5%/yr panel degradation, no federal credit to the homeowner, no battery.
• What it does not tell you. Whether you qualify for CARE, what a specific company will actually quote you, or whether a given contract has an escalator, a buyout, or transfer terms that change the answer. It tells you what the deal has to cost to be worth signing.

A note on what we can and cannot show you: we looked hard for named low-income households who went solar on a private lease or PPA and were glad they did. The documented stories are overwhelmingly from programmes like Illinois Solar for All and GRID Alternatives, not from the private market. We have not invented any, and we would rather show you a smaller number of real people than a page of plausible ones.