When the Federal Reserve raises rates, it does not raise the rate on your mortgage, your revolver or your corporate bond. It raises exactly five numbers that it controls by decree, and then waits. The FOMC lifted its target range by 25 basis points to 3.75-4.00% on 16 September 2026, unanimously, on a 12-0 vote, in the statement published that afternoon. Everything else in the chain is a separate decision taken by a bank, a borrower, a trader or a survey statistician, each on its own clock. Twenty-four hours after the decision, five rates had repriced by the full quarter point and the 30-year Treasury yield had fallen seven basis points below where it sat before the meeting began.
That gap is the whole subject. Most explainers of a Federal Reserve rate hike assert transmission and never measure it. So we measured it, link by link, against a single benchmark: the 15 September 2026 close, the last full session before the FOMC met. Of sixteen rates along the chain, five moved the full 25bp on day one, all of them administered rates the Fed or a bank simply announces. Six market rates moved by amounts ranging from plus four to minus seven basis points. Two of the most consumer-facing numbers in the chain, the average savings deposit rate and the credit card APR, are published so slowly that the hike cannot legally appear in them until late October at the earliest. Transmission is not one event. It is sixteen events with wildly different latencies, and knowing which is which is what separates a desk that hedges correctly from one that hedges the headline.
Key facts
- Target range raised 25bp to 3.75-4.00%, unanimous 12-0 vote — FOMC statement, 16 September 2026
- Interest on reserve balances set to 3.90%, primary credit to 4.00%, ON RRP offering rate to 3.75%, all effective 17 September 2026 — Fed implementation note, 16 September 2026
- Bank prime rate raised from 6.75% to 7.00% at eight large US banks, effective 17 September — Wells Fargo release, 16 September 2026
- 30-year Treasury yield 5.29% on 17 September, down from 5.36% on 15 September; 10-year 4.94%, down from 5.00% — US Treasury daily yield curve
- 30-year fixed mortgage 6.95% in the survey week to 17 September, from 6.76% a week earlier — Freddie Mac PMMS, 17 September 2026
- National average savings deposit rate 0.38% against a policy rate then at 3.50-3.75% — FDIC national rates, published 17 August 2026
- Credit card APR on all accounts 20.94%, a Q2 2026 observation — Federal Reserve G.19, released 8 September 2026
The five rates the Fed actually changed
The federal funds target range is a range, not a rate, and the Fed does not trade in the funds market to enforce it. The work is done by administered rates published in the implementation note that accompanies every decision. On 16 September the Board voted to move interest on reserve balances to 3.90%, the discount window’s primary credit rate to 4.00%, and the Desk’s standing overnight reverse repo offering rate to 3.75%, each effective 17 September.
IORB is the load-bearing one. A bank with a reserve account can earn 3.90% risk-free overnight, so it has no reason to lend fed funds below that for long, and the effective federal funds rate settles just under the ceiling. On 16 September, the last day of the old regime, EFFR printed at 3.63% against an IORB of 3.65%, a two-basis-point discount, on $90bn of volume, per the New York Fed reference rate data.
The ON RRP rate is the floor, available to money funds and government-sponsored enterprises that cannot hold reserves. It matters less than it did: take-up on 17 September 2026 was $276m, against a per-counterparty cap of $160bn. An instrument that once absorbed trillions is now a rounding error, which is precisely why the Fed keeps describing its framework as one of ample rather than abundant reserves.
The fifth rate is not the Fed’s at all. The bank prime rate is a private convention, conventionally the upper bound of the target range plus 300bp, and eight large US banks including Wells Fargo, JPMorgan, Bank of America and Citigroup moved it from 6.75% to 7.00% within hours of the decision. Prime is the reference for most HELOCs, small-business credit lines and variable consumer loans, which is why the first genuinely economic consequence of a hike usually lands on a revolving balance rather than a new loan. Chairman Kevin Warsh framed the decision in exactly those terms at the press conference: “We removed a dose of accommodation, so that financial and credit conditions would be more consistent with our ultimate objectives,” he told the Financial Times’ Claire Jones, in the transcript published by the Board.
What repriced, what did not, and who said so
A useful discipline is to separate rates that change because someone announces them from rates that change because someone trades them. The announced group moved on schedule. The traded group did something more interesting.
The published overnight benchmarks had not moved at all as of the evening of 17 September, and could not have. EFFR and SOFR are published the morning after the day they describe, so the most recent prints available on 17 September, 3.63% and 3.62% respectively, both describe trading on 16 September, the day before the new range took effect. Anyone reading a screen that evening was looking at the old policy regime. The first observation under 3.75-4.00% was scheduled for publication on 18 September.
Compounded benchmarks lag further by construction. The 30-day average SOFR stood at 3.64783% on 17 September, against 3.64850% on 15 September. A floating-rate loan resetting off 30-day average SOFR on the day of a hike captures none of it, and only reaches full pass-through around 30 days later. For a syndicated loan book that resets monthly, the effective date of a September hike is closer to mid-October.
Mortgages did move, though not because of the Fed. Freddie Mac’s Primary Mortgage Market Survey put the 30-year fixed at 6.95% in the week to 17 September 2026, up from 6.76% a week earlier and 6.26% for the 15-year. That 19bp weekly move tracked long Treasuries in the run-up, not the decision itself. Mike Fratantoni, SVP and Chief Economist at the Mortgage Bankers Association, was explicit about the sequencing in the association’s FOMC commentary: “Longer-term rates, including mortgage rates, had already baked in the expectation of hikes at this and future meetings. Thus, longer-term rates have not moved much in response to this news.”
Sixteen rates, one chart, and a curve that went the wrong way
Measured against the 15 September close, the Treasury curve did not tighten. It twisted. The 1-month bill rose 4bp to 3.97% and the 3-month rose 1bp to 4.12%, both reasonable responses to a higher overnight rate. The 2-year finished 17 September at 4.67%, exactly where it started. The 5-year fell 5bp, the 10-year fell 6bp to 4.94%, and the 30-year fell 7bp to 5.29%. A rate rise at the front and a rate fall at the back is what a credible tightening looks like from the bond market’s side: the curve is pricing the hike as effective against inflation rather than as the start of an inflationary spiral.
Credit agreed. Investment-grade and high-yield credit both closed higher on 17 September, with LQD at $105.145 and HYG at $78.73 per Nasdaq quote data, against 15 September closes of $104.28 and $78.38. Corporate funding conditions loosened on the day the policy rate rose. The dollar, by contrast, did tighten: the ICE US Dollar Index closed at 100.214 on 17 September and the euro weakened to 1.1481 from 1.1539 on 15 September, using ECB reference rates. For a broker or treasury desk, the practical reading is that the hike arrived through the currency and the front end, and nowhere else, in its first 24 hours. Our coverage of the 10-year at 5.025% going into the meeting shows how much of the long-end move had already happened beforehand.
| Link in the chain | Reset mechanism | Latency to full pass-through |
|---|---|---|
| IORB, primary credit, ON RRP | Board vote, announced | Next business day |
| Bank prime rate | Bank announcement | Same day to next day |
| EFFR, SOFR (published prints) | Traded, published T+1 | One business day |
| 30-day average SOFR | Backward compounding | About 30 calendar days |
| 2y to 30y Treasuries | Traded on expectations | Already priced, or never |
| 30-year mortgage | Tracks the 10-year plus spread | Weeks, and only via the 10-year |
| Savings and CD rates | Bank discretion | Months, often partial |
| Credit card APR | Prime-indexed, billing cycle | One to two statement cycles |
Two Dallas Fed papers put numbers on the parts of this table that intuition gets wrong. Matthew McCormick and Srini Ramaswamy, both financial economists in the Dallas Fed’s research department, decomposed mortgage rate dynamics over 20 years and concluded that “the mortgage rate exhibits a partial beta of less than 20 percent with respect to the fed funds rate, while exhibiting an 85 percent beta with respect to the 10-year rate,” in a paper published 7 May 2026. Combine that beta with what the curve actually did, and a mechanical estimate falls out: 25bp on funds contributes under 5bp to the mortgage rate directly, while the 6bp fall in the 10-year subtracts roughly 5bp. The two effects cancel. The 19bp weekly rise in PMMS came from the preceding fortnight’s repricing, not the meeting, which is exactly what Fratantoni described from the origination side.
The measurement problem nobody regulates
The structural tension in transmission today is not between innovation and regulation. It is between the speed of policy and the speed of official statistics. The Fed changes rates eight times a year at most, and the statistical apparatus that tells the public what happened is slower than that.
Take deposits. The FDIC’s national deposit rate series, published under the Section 337.7 interest rate restrictions, showed an average savings rate of 0.38%, interest checking at 0.07% and money market accounts at 0.63% in the file published 17 August 2026, based on data as of the prior month end. Against a policy rate then at 3.50-3.75%, the implied deposit beta at the average insured institution is close to zero. The series publishes on the third Monday of each month and reflects the last business day of the prior month, so the first observation that can contain a 17 September hike is the file due on 19 October 2026. Credit cards are worse: the Fed’s G.19 reports commercial bank credit card rates quarterly, and the release of 8 September 2026 carried a Q2 2026 reading of 20.94% on all accounts and 22.15% on accounts assessed interest. No published US statistic will show the September hike in a card APR before December.
There is a live policy argument underneath this. Sam Schulhofer-Wohl, Senior Vice President and Senior Adviser to the President at the Federal Reserve Bank of Dallas, built a formal measure of how precisely the funds target reaches other money markets and found that “monetary policy transmission has deteriorated in recent months but does not exhibit a breakdown comparable to those in 2018 through 2020,” in research published 16 December 2025. His diagnosis ties the drift to falling reserve balances, and his conclusion points at the Tri-Party General Collateral Rate, which “transmits more robustly, consistent with a recent proposal for the FOMC to adopt TGCR as its operating target.” The statement of 16 September repeated that the Committee “is continuing its policy of maintaining ample reserves in the banking system,” language that reads differently once you know the Desk’s own researchers are debating whether fed funds is still the right thing to target. Readers tracking the politics of this Committee will find the groundwork in our account of Warsh’s Jackson Hole turn.
What to watch, and when
Three dated checkpoints follow from the mechanics above, and each is falsifiable.
First, the overnight complex. EFFR should settle in the low 3.8s under an IORB of 3.90%, roughly preserving the 2bp discount observed on 16 September, with SOFR printing within a couple of basis points of it on most days and spiking at quarter-end. A persistent EFFR above IORB, or a SOFR-IORB spread that will not close, would be the first hard evidence that scarce reserves are degrading transmission in the way Schulhofer-Wohl’s measure anticipates.
Second, the compounding benchmarks. If EFFR holds near 3.88%, 30-day average SOFR rises from roughly 3.65% to roughly 3.87% by mid-October 2026. Borrowers on monthly SOFR resets should budget for the increase to arrive in the October reset rather than the September one, and hedges struck on the announcement date will look mispriced for about four weeks by construction.
Third, deposits. The gap between a 7.00% prime rate and a 0.38% average savings rate is the clearest arbitrage in retail banking, and it widens by 25bp with every hike that banks pass to borrowers and not to depositors. Watch the 19 October FDIC file for the first post-hike reading. A move of less than 5bp would confirm that the deposit channel remains effectively closed, which is bullish for net interest margin at deposit-funded banks and bearish for the assumption that a hike slows household spending through the saving decision. For the odds on whether a second hike arrives to test any of this, see our running coverage of the October meeting odds across venues and the September projections showing 16 of 18 officials penciling in another increase.
Frequently asked questions
What happens when the Fed raises rates, in order?
The Board raises interest on reserve balances, the primary credit rate and the ON RRP rate, effective the next business day. Banks raise prime the same day. Overnight benchmarks such as EFFR and SOFR drift to the new range and are published the following morning. Compounded benchmarks take about 30 days. Deposit and card rates move last, if at all.
How does a Fed rate hike work if the Fed does not lend at that rate?
It works by arbitrage. A bank can park reserves at the Fed and earn IORB, set at 3.90% from 17 September 2026, so it will not lend overnight materially below that. Non-banks can lend to the Fed at the ON RRP rate of 3.75%. Those two administered rates bracket the market, and the effective federal funds rate settles inside the corridor without the Fed trading at all.
Why did mortgage rates not fall when the Fed raised rates, or rise with it?
Mortgage rates track the 10-year Treasury, not the funds rate. Dallas Fed research published in May 2026 measured a partial beta below 20% to fed funds and 85% to the 10-year. When the 10-year fell 6bp over the meeting while the funds rate rose 25bp, the two forces roughly offset, which is why the 30-year fixed at 6.95% on 17 September reflected the prior fortnight rather than the decision.
When will a September 2026 rate hike show up in credit card and savings rates?
Card APRs are indexed to prime and reset on the next one or two statement cycles, so cardholders see it in October or November. Official confirmation is slower: the FDIC publishes national deposit rates monthly with a one-month lag, making 19 October 2026 the first possible post-hike file, and the Fed’s G.19 reports card rates quarterly.
What does the fed funds rate explained actually mean for a broker’s funding costs?
Repo and prime brokerage financing price off SOFR or off a spread to fed funds, so overnight funding repriced on 17 September 2026. Term funding did not, because the 2-year finished the week unchanged at 4.67%. Desks funding short and lending long saw carry compress by roughly the full 25bp, while those on 30-day average SOFR resets absorbed the change over the following month.
Does a rate hike always strengthen the dollar?
Not reliably, but it did here. The euro fell to 1.1481 on 17 September 2026 from 1.1539 on 15 September, and sterling and the yen weakened by similar magnitudes against the dollar on ECB reference rates. Currencies respond to the change in expected policy rather than the level, so a hike that markets had already priced can leave the dollar flat or weaker.