Solar Production Guarantee: What a Shortfall Actually Pays

Inside a 20-year residential solar contract on file with New York's Department of Public Service there is a four-column table headed Example Guaranteed kWh / Example Actual kWh / Example Agreed $/kWh Energy Price / Example Payment to You. It has one row: 10,000, 9,500, $0.10, $50.00.

That is the contract explaining its own arithmetic. Fall 500 kilowatt-hours short over two years and a cheque for fifty dollars arrives. It is a perfectly honest illustration, and reading it is the fastest way to understand that a production guarantee is not a promise about production. It is a formula with four terms — a threshold, a price, a measurement method, and a settlement period — and every one of those terms is set by the party that also produced the estimate.

Two contracts are taken apart below. Both are standard-form agreements filed with a state regulator and downloadable by anyone. Both are completed specimens rather than one household's signed paperwork — the customer is "John Smith" — but the system sizes, contract prices and guarantee tables are filled in, and every figure below is read straight off them. Both are also old: version 5.1.0 of April 2018 and version 1.4.0 of December 2019. Newer versions of these forms exist and are not on this docket, so read what follows as how the clause is built rather than as the terms any seller is offering today. The four things to check do not change.

The three warranty documents behind the hardware are a separate matter entirely; this is only about the clause that says what happens when the kilowatt-hours do not show up.

Reading the threshold out of a table that never states it

The first document is a Solar Home Improvement Agreement, version 5.1.0 dated 30 April 2018, for a 4.25 kW DC system at a contract price of $15,938, with a stated first-year estimate of 4,563 kWh (filed with the NY Department of Public Service, read 10 September 2026). Exhibit 2 carries the Power Production Guarantee, and its table runs in two-year steps:

Year   Total kWh   Agreed Price/kWh
  2      7,738         0.08756
  4     15,399         0.08756
  6     22,983         0.08756
 ...
 20     73,998         0.08756

Nowhere in that clause is there a percentage. So derive it. The document states annual degradation of 0.5%, which makes the first two years of the estimate 4,563 + 4,540 = 9,103 kWh. The guaranteed figure for the same window is 7,738.

7,738 / 9,103  =  0.850

Exactly 85%. Check it again on the next step: years three and four estimate 4,517 + 4,495 = 9,012, and the table's increment is 15,399 − 7,738 = 7,661, which is 0.850 of it. The deadband is a clean 15 percentage points, applied consistently, and stated nowhere.

Run a shortfall through it. If that array delivers 90% of its own estimate over 24 months — 8,193 kWh, a real and unremarkable miss — actual production sits above the guaranteed 7,738 and the payment is zero. Push it to 80%, or 7,282 kWh, and the claim finally opens:

shortfall  =  7,738 − 7,282        =    456 kWh
payment    =  456 × $0.08756       =  $39.93   (per 24 months)

The homeowner is 1,821 kWh short of what the proposal projected. The guarantee returns $39.93 against it. Valued at the U.S. average residential price for the twelve months to June 2026, those missing kilowatt-hours were worth about $326, so the cheque covers roughly 12% of the loss; on a 25-cent tariff it covers under 9%. Which of those you get is set by the next term, not by the guarantee.

The agreed price is a second discount, applied quietly

$0.08756 per kWh, flat, for two decades. Compare it to what the missing kilowatt-hour would have been worth: the U.S. average residential price was 12.87 cents per kWh in 2018, when this contract was written, and 17.90 cents over the rolling twelve months ending June 2026 (EIA Electric Power Monthly, Table 5.3, data for June 2026, released 26 August 2026, read 10 September 2026). The agreed price started at roughly two-thirds of the retail rate and is now worth about half of it, because retail moved and the contract term did not.

It is easy to model this the wrong way round. A payback spreadsheet that settles shortfalls at whatever the utility charges turns the clause into something resembling insurance, and a spreadsheet that forgiving will wave through a soft production estimate. The clause is not insurance. It is a fixed and undisclosed fraction of the loss, and the fraction shrinks every year the tariff rises. Use the number written in the contract rather than your own rate — and if the contract prints no number at all, that absence is the finding.

The second document shows how far this term can move. A Solar Subscription & Home Improvement Agreement, version 1.4.0 dated 2 December 2019, covers an 8.235 kW DC system estimated at 7,274 kWh in year one (also filed with NY DPS, read 10 September 2026). Same company, same mechanism, different settings: its first 24-month guarantee is 13,787 kWh against an estimate of 14,512 — a 95% threshold on the same division, not 85 — and the price starts at $0.21040 and steps up to $0.23026 by year 20. The cover page of that document states the threshold as something else again, which is worth a section of its own below.

Normalise the two, so the different array sizes stop mattering. A 20%-below-estimate run over 24 months pays:

Nearly six to one, for the identical words on the cover page. And note the last row of the middle column against the payment escalator in the same subscription document, which permits the monthly bill to rise by up to 2.9% a year: over 18 years that compounds to +67%, while the price paid for a missing kilowatt-hour compounds to +9.4%.

One document, two thresholds: 85% on the cover, 95% in the table

Deriving the subscription's 95% turns up something the arithmetic cannot settle. The threshold above comes from Section 21, the Power Production Guarantee, whose table sets Guaranteed kWh at 13,787 for the first 24 months — 95.0% of the 14,512 kWh the same document estimates for those two years.

The summary page at the front of that same agreement describes the identical remedy differently. Under the heading Performance guarantee / System production guarantee, it reads: "If at the end of each successive 24 month anniversary of your first monthly payment the cumulative Actual kWh generated by the System is less than 85% of the estimated kWh, then we will send you a refund check equal to the difference between the cumulative Actual kWh and the Guaranteed kWh multiplied by the Guaranteed Energy Price per kWh."

Read that sentence twice. The trigger is 85% of the estimate. The amount is measured from the Guaranteed kWh, which Section 21 puts at 95%. One sentence, two different reference numbers, and a ten-point gap between them where it is genuinely unclear whether a claim exists.

Put a system in that gap. Suppose the array returns 88% of its own estimate over the first 24 months, or 12,771 kWh:

Section 21 reading   13,787 − 12,771 = 1,016 kWh × $0.21040 = $213.77
Cover-page reading   12,771 > 12,335 (85% of 14,512)        =   $0.00

The same shortfall is either a $213.77 cheque or nothing at all, depending on which sentence governs. Nor does the document resolve it by hierarchy in the direction you might expect: its own Overview box states that "in the event that the terms in this statement conflict with terms appearing elsewhere in your contract, the terms in this statement are controlling" — pointing at 85% — while the payment formula inside that very statement points back to the Section 21 table for the number it multiplies.

There is a hint about how the 85% got there, though not a resolution. In the companion filing, the Solar Lease Disclosure Statement uses word-for-word the same sentence with a different figure: "less than 95% of the estimated kWh." Same drafter, same box, same year, and the number that matches the subscription's own table appears on the lease form rather than on the subscription's.

Which controls is a question of contract interpretation, and nothing in these two documents answers it. That is the point worth carrying to a sales appointment. Ask for the trigger percentage in writing, ask the salesperson to point at the sentence it comes from, and ask what happens at 88%. A guarantee whose threshold moves by ten points depending on which page you read is not a number you can put in a spreadsheet, and if nobody will put it in an email, that is your answer about how often it pays.

The measurement clause, and who owns the meter

A threshold and a price are worthless without an agreed number to compare against. Both contracts define Actual kWh as the AC electricity "measured and recorded by Tesla" — through the PowerGuide monitoring service in the purchase agreement, "our gateway monitoring device" in the subscription — and both add that where the service is not available, the provider "will estimate the Actual kWh by reasonable means." The subscription document goes further in its cost summary: without a maintained high-speed internet connection, the company cannot monitor the system and therefore cannot provide the guarantee, and the homeowner may be required to supply annual production figures from the inverter.

So the claim is measured by the counterparty, on equipment the counterparty installed, over a link the homeowner is contractually responsible for keeping alive. That is not automatically unfair — somebody has to read a meter — but it decides where the burden of proof sits during a dispute, and it is worth knowing before rather than after. Ask which device produces the number, whether you can export the raw interval data yourself, and what happens to a claim for a period when the gateway was offline.

One clause runs the other way and is worth knowing before you conclude the whole thing is one-sided: in both contracts, production above the guaranteed figure in any 24-month window "will be carried over and will be used to offset any deficits that may occur in the future." A strong first two years therefore buys down a weak third and fourth. The subscription adds a matching restriction — cancel partway through a 24-month period and no cheque is issued for it at all, with no pro-rating for the months already served.

Then read the exclusions, which is where the shortfall you are most likely to have gets removed. Exhibit 2 of the purchase agreement lists eleven of them, numbered (i) to (xi), covering lost power production as well as repairs. Among them: materials and equipment already covered by manufacturers' warranties, "shading from foliage that is new growth or is not kept trimmed to its appearance as of the date the System was installed," loss that follows from the customer's own request — the contract's examples are a system removed for roof repairs and an inverter the homeowner insisted be sited out of the shade — and any Force Majeure Event, a term the same document defines to include "abnormal weather condition or actions of the elements," along with wind, lightning, drought and animals. A closing line caps everything: "Any cash payout to you is limited to the payouts described in Section 2(b) of this Limited Warranty." One sentence in the guarantee clause itself goes further than any exclusion: "Your cumulative Actual kWh is dependent on a shading percentage of 12.00% on your Home. If this shading percentage increases, your Guaranteed Actual kWh will be reduced proportionately."

The guarantee is indexed to a shading figure printed in the contract. Trees growing is not a claim; it is a reduction of the guarantee. That makes the shading assumption the load-bearing number in the document, which is the same conclusion the twelve required proposal items reach from the other direction.

One heading, four different promises

The purchase filing is a bundle, and it carries three disclosure statements — purchase, lease and PPA — each with a Guarantees box under the identical heading, each saying something different. Add the subscription from the other filing and there are four. The purchase box points at the Guaranteed kWh table; the lease box says 95% of the estimate; the subscription box says 85%. The fourth is the shortest. The Solar PPA Disclosure Statement's Guarantees box lists one item — "Performance guarantee" — with the remedy stated in full as: "Tesla will repair or replace any defective part and restore System performance." No kilowatt-hour table, no price, no payment.

That reads like the fake version of the clause, and in a PPA it very nearly is not, because under a power purchase agreement you pay per kilowatt-hour delivered. Underproduction settles itself on the invoice. The structure, not the promise, is doing the work. Reverse it and the picture inverts: under a cash purchase or a loan you have paid up front for the whole of a system the contract puts at a 30-year estimated lifetime, so a production shortfall is a pure loss with no automatic offset anywhere — and that is exactly the arrangement whose guarantee carries the widest deadband and the lowest agreed price.

Regulation lines up on the same side, and the manual's own headings say so. NYSERDA's NY-Sun Program Manual splits the requirement in two: section 3.14 is titled "Production Warranty for PPA/Leases" and section 3.13 "System Warranty for Purchase Agreements." Section 3.14 obliges the contractor to offer a production guarantee "for the initial term of the agreement (at minimum)" on leases and PPAs, and caps the degradation the guaranteed output may assume at 1% per year. Purchasers get 3.13 instead — at least a full five-year transferable warranty against breakdown "or degradation in electrical output of more than 10% from the original rated electrical output," with warranty requests answered in 72 hours and repairs completed within 30 days (NY-Sun Program Manual, April 2026 revision, read 10 September 2026). That is a test of the hardware against its nameplate, not a promise about annual kilowatt-hours, and the two are easy to confuse in a sales conversation.

California does not close the gap either. The Solar Energy System Disclosure Document that must be printed on the front or cover page of every residential solar contract is one page long, and it has three things on it: the total cost including financing, the CSLB complaint route, and the three-day right to cancel where the contract was not signed at the contractor's place of business. There is no fourth item. Production, kilowatt-hours and performance guarantees appear nowhere on the form, which is developed jointly by CSLB and the CPUC under Business and Professions Code section 7169 (CSLB Solar Energy System Disclosure Document, form 13L-6 (05/2018), read 10 September 2026).

Four divisions to do before you sign

None of this requires a lawyer. It requires the guarantee table, the production estimate, and a calculator.

1. Threshold. Guaranteed kWh for the first 24-month period, divided by the first two years of the proposal's estimate with degradation applied. Anything below 0.95 is a deadband you are funding.
2. Price ratio. Agreed price per kWh divided by your own marginal rate off the tariff sheet. Then check whether the agreed price escalates at all across the table's twenty years.
3. Worst honest case. Multiply the final cumulative row by the agreed price. On the purchase agreement above, 73,998 × $0.08756 is $6,479 — the most this guarantee can ever pay, against a $15,938 contract, if the array produced nothing for two decades.
4. Basis of the estimate. NY-Sun's own annual output figure is generated in NYSERDA's portal using NREL's PVWatts from system size, location, and the total solar resource fraction from a shading report, with projects below 70% TSRF receiving a reduced incentive. Ask which tool produced the number your guarantee is indexed to, and then rebuild it yourself before deciding whether 85% of it is worth anything.

If the seller cannot supply the threshold, the price, the measurement device, and the exclusion list in writing, there is no arithmetic to check, and a guarantee you cannot compute is a paragraph rather than an obligation.