California NEM 3.0: How Export Credits Are Priced

Two kilowatt-hours leave the same California roof. One is exported at 3 p.m. on a Saturday in May and is worth $0.00001. The other is exported at 7 p.m. on a Saturday in August and is worth $1.19289. Both figures sit on one page — PG&E's Energy Export Credit price sheet for customers whose interconnection application year is 2026 (Solar Billing Plan EEC price sheets — that link downloads a ZIP holding four PDFs, one per interconnection application year; values here are from the 2026 sheet, read 5 September 2026). The gap between the cheapest cell and the dearest is about 119,000 to 1.

That page is what California's net billing tariff put in place of retail-rate netting. Every rate below is PG&E's: Southern California Edison and SDG&E run their own net billing schedules with their own export tables, and none of their numbers were opened for this piece, so do not carry a figure from here onto one of their bills. If you are outside California entirely, start with the three payment structures, then read the last section here.

The 576 cells your exports are priced against

Schedule NBT prices an exported kilowatt-hour by month, by hour, and by whether the day is a weekday or a weekend/holiday. Twelve months times twenty-four hours times two day types is 576 prices. Each cell splits again into a generation component and a delivery component, so the sheet carries 1,152 numbers.

The tariff names the ingredients. Generation components come from the Energy, Generation Capacity, Cap and Trade, Ancillary Services and Losses elements of the CPUC's Avoided Cost Calculator; delivery components from Distribution Capacity, Transmission Capacity, GHG Adder, GHG Rebalancing and Methane Leakage. Delivery values vary by climate zone inside the calculator, but PG&E averages them into one set, so where you live within PG&E territory does not move your export rate (Electric Schedule NBT, Sheets 10–11, Advice 7975-E, effective 28 August 2026, read 5 September 2026).

Weekday values from the 2026 vintage sheet, generation plus delivery:

The obvious shortcut is to average the table and carry one number into the model. Do that and you get $0.09292 per kilowatt-hour across all 576 cells, a figure that makes nearly any array look fine. But an array without storage does its exporting in the middle of the day, and those cells are the cheap ones: 9 a.m. to 4 p.m. averages $0.04664 on weekdays and $0.02407 on weekend days — half the flat figure, and about a quarter of it at weekends. The whole gap sits in summer evening hours a bare roof cannot deliver into.

On top of the table sits the ACC Plus adder, a flat credit per exported kilowatt-hour set by application year and held for nine years from Permission to Operate. The ladder is printed in the tariff itself: $0.02200 in 2023, $0.01760 in 2024, $0.01320 in 2025, $0.00880 in 2026, $0.00440 in 2027, with a higher parallel ladder for low-income customers ($0.03600 in 2026) and nothing at all for non-residential accounts (Schedule NBT, Sheet 11).

Two vintages, one table, one shrinking adder

Four price sheets sit in PG&E's download bundle, one per application year from 2023 to 2026, each headed 2026 Energy Export Credit (EEC) Values above its own application year — they show what each vintage collects during calendar 2026. Compared cell by cell across all 1,152 numbers, the 2023 and 2024 sheets are identical to each other, and so are the 2025 and 2026 sheets.

That is not a filing mistake. The tariff pins a vintage to the calculator "adopted as of January 1 of the calendar year of the customer's completed interconnection application date," and the CPUC's current adopted version is the 2024 ACC, approved in Resolution E-5328 on 13 November 2024 under D.24-08-007. A 2026 update is still in flight in R.22-11-013, with a staff proposal issued 23 April 2026 (CPUC DER Cost-Effectiveness, read 5 September 2026). So 1 January 2025 and 1 January 2026 pointed at the same calculator, and the two vintages inherited the same 576 cells.

Which leaves the ACC Plus adder as the only thing separating a 2025 applicant from a 2026 one: $0.01320 against $0.00880, a difference of $0.0044 on every exported kilowatt-hour for nine years — $17.60 a year on 4,000 exported kWh, $158 across the lock.

The older pair is not simply richer, either. Its flat 576-cell average is $0.09349 against the newer pair's $0.09292, six hundredths of a cent apart. But its single highest cell is $2.52139, a September weekday at 7 p.m., more than double anything on the 2026 sheet. Two vintages can average the same and still price a battery completely differently, which is why the averaging shortcut fails in both directions. Whether the 2027 vintage lands on a new calculator was still unsettled on 5 September 2026. Do not model it as though it were decided.

The import clock runs on different hours

Residential PG&E customers on Schedule NBT do not get to shop rate plans. The tariff says they "must take service on the E-ELEC (ELECTRIC HOME) rate schedule" and may add critical peak pricing (SmartRate) on top of it; that is the whole menu. The E-ELEC clock is short and blunt: peak 4 p.m. to 9 p.m. every day, weekends and holidays included, partial-peak 3 to 4 p.m. and 9 p.m. to midnight, everything else off-peak. Summer runs 1 June to 30 September, winter 1 October to 31 May (Schedule E-ELEC, Advice 7846-E, effective 1 March 2026, read 5 September 2026).

Total bundled energy rates on that schedule, from the sheet submitted and effective 1 June 2026 under Advice 7921-E and D.26-04-036:

Summer   peak (4-9 pm)          $0.55214 /kWh
         part-peak              $0.39026
         off-peak               $0.33358
Winter   peak (4-9 pm)          $0.32063
         part-peak              $0.29854
         off-peak               $0.28468
Base services charge, Tier 3    $0.79343 /day  = $289.60 /yr

Notice the mismatch. The import clock has one peak window and it never moves. The export table has 576 values and agrees with that window only part of the year. In August the 4-to-9 p.m. weekday hours average $0.90877 against $0.31956 for the whole day. That is 2.84 times the day's own average. In April the same five hours average $0.04783 against $0.04658 — a premium of 2.7 percent, in a month whose highest weekday cell falls at 2 a.m. A model treating "peak" as a single idea has already lost the thread.

A stored kilowatt-hour has two jobs, and August picks the odd one

The Commission said what the two clocks together were for. D.22-12-056 states that the high-differential import rates "in combination with the variable retail export compensation rates provided by the Avoided Cost Calculator send strong price signals to customers to shift their use of energy from the grid to mid-day and export electricity during the evening hours, which promotes the installation of storage with the solar systems," and that "to attain this sustainable growth, the market must transition to one focused on solar paired with storage" (Decision 22-12-056, pp. 3 and 147, read 5 September 2026).

The battery is not an accessory in this tariff; it is the mechanism. Which of its two jobs pays more flips with the season, and round-trip losses drop out of that comparison because both jobs pay them:

August, 7 pm weekday
  export         $1.15441 + $0.00880 ACC Plus  = $1.16321
  self-consume   $0.55214 summer peak avoided
  exporting wins by 2.11x

April, 6 pm weekday
  export         $0.07828 + $0.00880           = $0.08708
  self-consume   $0.32063 winter peak avoided
  self-consumption wins by 3.68x

Same battery, same house, opposite instruction. The other comparison — hold or sell now — is where round-trip efficiency bites, and here one input stops being a published number. Every rate in what follows comes off a tariff sheet; the 0.90 does not. It is a placeholder for the AC-to-AC round-trip figure printed on your own battery's datasheet, no source is claimed for it, and it is the first thing to overwrite. Take a month with 240 kWh exported between 10 a.m. and 4 p.m. on weekdays, and a battery you can push 80 kWh through:

August, nothing shifted
  240 kWh x $0.06849  (Aug weekday 10 am-4 pm avg)  = $16.44
  240 kWh x $0.00880  ACC Plus                      = $ 2.11
                                                      ------
                                                      $18.55

August, 80 kWh moved into the 7 pm hour  (round trip 0.90 ASSUMED)
  160 kWh x $0.06849 + 160 x $0.00880               = $12.37
   72 kWh x $1.15441 +  72 x $0.00880               = $83.75
                                                      ------
                                                      $96.12   (+$77.57)

April, the same 80 kWh, self-consumed in the 4-9 pm window
  as exported midday:  80 x ($0.00519 + $0.00880)   = $ 1.12
  as stored and used:  72 x $0.32063 winter peak    = $23.09   (+$21.97)

Replace the 240 and the 80 from your own interval download, the rates from your own vintage sheet, and the 0.90 from the datasheet. The totals are not the point. The point is that those two lines come from identical equipment and differ by a factor of three and a half, which no annual-average model can show you. Sizing the thing is a separate exercise off your own load profile.

The interconnection form decides whether you can reach the August hours

This part rarely survives into a proposal. Schedule NBT defines "NBT Integrity" as the requirement that a customer receive export credits only for renewably generated exports from an NBT-eligible generator, and it enforces that through the configuration you sign up for. Two definitions do the work, each enforced either by a Certified Power Control System — UL 1741 CRD, which the tariff dates to a Certification Requirement Decision issued 8 March 2019 — or by a PG&E-approved relay doing the same job in hardware.

• No Grid Charging. The battery cannot be charged from PG&E's grid, with inadvertent imports tolerated only up to ten seconds. A 48-hour PSPS notice is the one documented exception, and only for a system qualified for resiliency.
• No Storage Export. The battery cannot export to the grid "and therefore will not result in additional credits." Large DC-coupled paired storage above 10 kW has to meet this definition outright; the only choice there is which mechanism enforces it.

Read that second definition against the August column. The panels on such a system still export and still earn. What cannot reach $1.15441, or $1.13014, or $1.01459 is a kilowatt-hour that has been through the battery — not because the hardware is incapable, but because the interconnection agreement says it will not. Large AC-coupled systems instead pick one of four arrangements: a non-export relay on the storage, a three-point metering setup, an interval meter wired directly to the generator, or a Certified PCS.

Small paired storage — 10 kW or under, no Certified PCS, none of that metering — is billed by an estimation methodology instead. PG&E caps NBT-eligible exports for each billing month using a California Solar Initiative Expected Performance-Based Buydown production estimate scaled to the generator's size, and anything above the cap is forfeited. The tariff adds that forfeited exports "are assumed to have occurred during the highest price billing period during the Customers' billing cycle, regardless of when the excess energy was actually exported" — the loss lands on the best hours by definition. A footnote then suspends the whole mechanism: since 15 April 2024 residential customers who would be billed this way are "temporarily billed on NBT as if storage and REGF were all one generator, until the billing is operational for these customers." Suspended is not repealed. Ask which of the two applies on the date you sign, and get the answer in writing.

The charges your export credits cannot touch

Export credits are not money. They are a restricted offset, and the tariff says so plainly. They reduce volumetric import charges and nothing else — not minimum charges, customer charges, meter charges, facilities charges, base services charges, demand charges or surcharges. The base services charge alone is $289.60 a year at Income Tier 3, and no amount of exporting touches it. Nor may they reduce the Public Purpose Program, Nuclear Decommissioning, Competition Transition or Wildfire Fund charges. Only the ACC Plus credit may be applied to any charge on the bill.

Generation and delivery credits also accrue in separate buckets and offset only charges of their own kind, which matters because delivery is a rounding error in most hours and about a fifth of the value in the August evening cells — $0.24087 of the $1.15441 at 7 p.m. on an August weekday, against nine tenths of one percent of a January midday cell. Unused credits roll forward across your Relevant Period — twelve billing cycles from Permission to Operate, and each anniversary after — but only "while the customer is on the NBT." Stop being served under the tariff and the excess is forfeited at the last true-up on it, which is what closing the account to sell the house does.

One more figure. The nine-year payback quoted in California proposals is not a market observation but a design target, and the Commission wrote down both the target and its instrument. D.22-12-056 states that "the successor tariff targets a nine-year simple payback for a stand-alone solar system, which is equivalent to nearly $100 in monthly bill savings" (p. 77), and that ensuring that payback "allows customer-sited renewable generation to grow sustainably during the transition to a tariff that is focused on solar paired with storage" (p. 150). The instrument is the ACC Plus adder, and it has fallen 60 percent since — $0.02200 in 2023, $0.00880 now. The target was set once, in December 2022. Nothing has restated it against the prices anyone pays today.

Elsewhere: two questions that find your version of this table

Two readers need this section. One is on Southern California Edison or SDG&E, where the tariff structure is the same statewide decision but the export tables are each utility's own; SCE posts its schedules in its tariff books (checked 5 September 2026) and SDG&E posts a Solar Billing Plan page of its own. Do not reuse a cell from above. The other reader is out of state, where the mechanism travels even though the numbers do not. Both take the same two questions to the tariff library.

Is the credit a single rate or a schedule? If it is one cents-per-kWh figure, the arithmetic in the four numbers on your own bill covers it. If it points at an hourly table, a value-of-solar stack, or an avoided-cost model, find where that table is posted and which vintage applies to you.

What locks, and for how long? California locks nine years from Permission to Operate for applications completed by 31 December 2027, then pays averaged hourly values from whatever calculator is current, refreshed annually. Other states lock ten years, or five, or nothing. A proposal holding today's export rate flat for 25 years is asserting something your tariff probably never promised — the same failure mode everywhere.

Five numbers to lift off your own vintage sheet

Open the sheet for your application year and write down five things: the midday weekday average for your worst month and for your best, the highest evening cell anywhere in the table, your ACC Plus adder, and the date the lock expires. Then ask the installer which of the five produced their savings figure.

If the answer is one blended cents-per-kWh number, it sits somewhere between $0.02407 and $0.09292 depending entirely on which cells got averaged — and the distance between those two choices is the whole argument about whether a battery belongs in the quote. Which of the battery's two jobs the proposal is pricing is the next thing to pin down; they are not the same product.