DER finance and due diligence

Almost every behind-the-meter project earns money by avoiding a utility charge: an energy charge, a demand charge, a time-of-use peak or an export credit. Grid-program pay sits on top, and it is usually shorter and less certain. The value of a DER investment is therefore a forecast of the tariff.

Nationally, that forecast points up. Residential prices rose from 13¢ to 17.3¢ per kWh between 2019 and 2025, a 33% nominal increase, with commercial up 26% and industrial up 27%, according to LBNL and Brattle’s 2026 price update. The Energy Information Administration expects 18.2¢ in 2026 and 18.7¢ in 2027 in its October outlook. Investor-owned utilities asked regulators for $18 billion in increases in 2025, and regulators approved 64% of requested dollars from 2021 to 2025, a signal of further near-term increases.

The fan is the point. A project priced at ten times first-year savings needs rates to rise about 2.7% a year to earn an 8% return, which sits inside the band. On the low path it falls short; on the high path it clears easily. Brattle’s own read is that nominal rates will keep rising but not much more than inflation if we are lucky. Long-term load forecasts don’t settle it either: EIA’s 2026 Annual Energy Outlook projects electricity use growing 0.9% to 1.6% a year through 2050 across its eleven scenarios, and growth can raise or lower rates depending on who pays for new wires.

Load growth is the new variable. PJM, the grid operator for 13 states and Washington, D.C., raised its 20-year summer peak growth rate from 2.0% to 2.4% a year and now expects roughly 82 GW of added peak demand over 15 years. A PJM filing puts data centers at about 30% of peak demand by 2036. Whether that growth raises or lowers rates for everyone else isn’t settled: LBNL’s review finds the effect depends on spare system capacity, the cost of new infrastructure compared with existing costs, and how accurate the load forecast turns out to be.

A model has to make two separate rate assumptions. The first is level: how fast the average price per kWh rises. The second is structure: which charges the asset can avoid and when. The first is the one most business cases debate. The second is the one most likely to break them.

Going from flat rates to 3.5% a year adds about 23% to present value. That’s similar in size to the 16% of value lost in the stranded-capital example below, where a rate redesign cuts the demand charge by 40% from year six. Escalation builds gradually; a redesign can arrive in a single rate case.

California shows how structure can shift under a stable average price. Starting in March 2026, PG&E bills most residential customers a Base Services Charge of about $24 a month and lowers the price per kWh to offset it, under the CPUC’s income-graduated fixed charge decision. Total revenue barely changes, but every kWh a battery or solar system avoids is now worth a little less. PG&E’s explainer notes that high-usage customers tend to pay less and low-usage customers more.

Fuel-driven systems swing in both directions. Hawaii’s monthly fuel adjustment moved the value of a stored kWh by roughly ten cents between January and October 2026, as our Hawaiian Electric residential deep dive shows.

Retail tariffs have no futures market, so behind-the-meter investors can’t buy a hedge the way a gas-fired plant can. The hedging happens in who signs what:

• Fixed payments. A lease or fixed-price power purchase agreement moves rate risk to the host. The developer then carries the host’s credit risk instead.
• Savings floors. Battery-as-a-service deals that guarantee the host a share of savings, such as the 15% floor we model on C&I pages, leave rate risk with the developer.
• Multi-year program contracts. A five-year term such as Hawaiian Electric’s BYOD Plus tariff is worth more to a lender than season-by-season enrollment.
• Grandfathering. Legacy periods for export credits or time-of-use windows protect existing systems when rules change. Diligence should check whether they are in the tariff or only in a decision.
• Conservative tails. Lenders size debt on contracted revenue and a P90 case, a level the project should beat nine years in ten. Anything after the contract ends gets a haircut.
• Diversity. Portfolios spread across territories, regulators and program types so one rate case can’t sink the fund.
• Change-in-law clauses. Tariff reopeners, termination payments and price resets decide who pays when the rules move.

Where real hedges do exist, for wholesale revenue and interest rates, our guide to hedging rate and revenue risk shows how tolls, floors, swaps and PPA escalators work.

Insurance is the cost line most often left out of simple paybacks, and it’s rising in the same places where rates are. Home premiums rose about 27% in nominal terms nationally from 2019 to 2024, roughly in line with inflation, but by 25% or more above inflation in parts of southern coastal states, according to the Government Accountability Office. The typical premium was 24% higher in early 2026 than three years earlier. Property insurance now averages $209 a month, or 9.6% of a mortgage payment.

Weather and wildfire costs show up twice. They raise the premium on the building, and utilities recover storm and wildfire spending through rates, which LBNL identifies as a major cost driver in California and in storm-hit states such as Maine. For a host, on-site backup can offset some of that through avoided outage losses, but only if the system is sized and configured for it.

Underwriters price storage on chemistry, spacing, fire testing and the operator’s record. Solarif’s underwriting guide puts annual premiums at 0.3% to 1.2% of project value and notes that lithium iron phosphate (LFP) cells usually get better terms than nickel manganese cobalt (NMC). Severe-weather zones add 20% to 50%. In the example, insurance can cost as much as routine upkeep, and diligence should treat it as a rising line, not a fixed one.

Our insurance guide for DER covers the full set of policies, the hail and fire losses behind premiums, and why small tax credits often go uninsured.

Incentives step down, run out of budget or get redesigned mid-life. The July 2025 federal tax law ended the residential clean energy credit for systems installed after December 31, 2025, and set new deadlines for solar and wind, according to the solar industry’s summary. Storage kept the full investment credit for projects starting construction through 2033, then 75% in 2034, 50% in 2035 and none after.

The bigger diligence item is the foreign-supplier test. Storage that begins construction after 2025 loses the credit unless enough of its component cost comes from suppliers outside prohibited foreign entities; the required share starts at 55% in 2026 and rises to 75% after 2029, as Plante Moran explains. That requires component-level cost tracking, and the IRS interim guidance is still being turned into regulations.

Our incentive risk guide goes deeper, with California net billing and PJM capacity prices as case studies and a split of locked versus expected value.

Program terms move too. Hawaii replaced its Battery Bonus with BYOD Plus, a five-year commitment with an upfront payment, and the terms differ enough to change payback. A model should separate payments that are locked in at signing from payments that are only expected.

Capital is stranded when the equipment still works but the revenue that justified it goes away. The chart below takes the 100 kW battery from our Hawaiian Electric C&I page and tests four ways the revenue can fail.

The base case covers its net cost 1.32 times. Either single failure still clears the line. A program ending plus a rate redesign drops it to 0.93, and a sale without an assignment clause to 0.89. That’s why contracts matter as much as forecasts: assignment rights, buyout schedules, termination payments and device portability decide whether a failure ends value or just moves it.

Aggregator exits are the newest form of the risk. When a platform leaves a market, enrolled devices may need a new operator before they earn again; our Hawaii residential analysis covers a 2025 example. Utilities face the mirror image: LBNL lists load forecast accuracy as a stranded-asset risk if expected large loads don’t arrive.

The stranded capital guide tests the same battery by the year each failure happens and covers provider bankruptcies, technology obsolescence and contract protections.

The same project looks different from each seat at the table. These are the checks each party usually runs.

Tax investors run a separate layer of checks on credits, recapture and supplier rules; our tax equity and fund capital guide walks through them with a worked capital stack.

Technical checks sit underneath every financial one: equipment certification and fire testing, warranty terms and fade curves, telemetry and control rights, and the measurement method that decides what a program pays; see our measurement and verification guide.

Our investor due diligence checklist ranks these checks by how much each finding can change value, and shows which party usually leads each one.

Green banks lend or guarantee where private capital is too expensive or won’t lend at all: small commercial hosts, multifamily buildings, low-income homes. Their tools are loan-loss reserves, subordinated debt and credit enhancement that bring private lenders in at lower rates. Their own funding can be a risk. The $20 billion federal Greenhouse Gas Reduction Fund awards made in 2024 were frozen after EPA terminated them in March 2025. An appeals court restored the injunction in August 2026, but the funds remain stayed through November 9, 2026, pending a possible Supreme Court petition, according to FindLaw’s case summary. A deal relying on that capital should model a no-funding case.

The green bank guide shows how loss reserves, subordinated debt and co-lending change debt service on the example battery, with leverage figures and the latest court status.