The Secured Overnight Financing Rate — SOFR — is the volume-weighted median rate on roughly a trillion dollars of overnight Treasury repurchase agreements. The Federal Reserve Bank of New York publishes it every business morning at 8:00 a.m. Eastern. It took over from dollar LIBOR when that benchmark retired in June 2023, and it now anchors one-month term SOFR, the floating leg of most dollar interest-rate swaps, and the resets on the funding facilities behind consumer credit. Your credit card does not borrow at SOFR. Card APRs are indexed to the prime rate, which sits at 7.50% against the Fed’s 4.25–4.50% target range as of this writing. But SOFR’s behavior in the last three business days of December — when its volatility clusters into a spike — is the cleanest daily read there is on how stressed bank balance sheets are. That stress does not reprice a card directly. It shows up in credit lines, APR margins, and January statements, on lags you can count in days and weeks.
So here is the pattern, the mechanics underneath it, and the transmission chain — how a 4.30% money-market print travels, step by step, to the interest line on a card statement.
SOFR, Prime, and the Index That Actually Sets Your APR
SOFR prints inside the Fed’s target range almost every day. Through 2025 it has spent most of the year between 4.28% and 4.31% — a few basis points below the 4.40% the Fed pays on reserve balances, and about 20 basis points under the 4.50% ceiling. Prime is simpler. Take the upper bound of the range and add 3.00 points: 4.50% plus 3.00% is 7.50%. It moves only on the days the FOMC shifts the range, which happens at most eight times a year.
Cards price off prime: APR = prime + margin. The average card account carries an APR of about 21.5%, which works out to a spread of roughly 14 points over prime. New offers average around 24%. The accounts actually assessed interest — the people carrying balances — run near 23%, per the Federal Reserve’s G.19 consumer credit release. The margin is where issuers park funding costs, loss expectations, and risk pricing. The index is just the Fed.
That distinction is the whole question. A SOFR spike is not an APR event — SOFR is not the card index. It is a funding-stress reading, and funding stress moves the margin side and the credit-supply side on slower clocks.
What Volatility Clustering Looks Like in the Data
Volatility clustering just means big moves arrive in runs instead of one at a time. Statisticians model it with GARCH. The practical version: variance piles up on specific dates, and SOFR’s version is calendar-driven. An ordinary day moves it a basis point or less from the prior print. Then the reporting dates arrive, and the elevated prints stack on consecutive days:
- December 31, 2018: SOFR jumps from 2.41% to 3.03% — 62 basis points in a single print.
- September 17, 2019: SOFR prints 5.25%, roughly 280 basis points above where it sat two business days earlier, at the collision of a quarter-end and a corporate tax date.
- Recent year-ends: the last business days of December print 5 to 20 basis points above mid-December levels — smaller, but still stacked on the reporting dates rather than scattered.
The clustering is the tell. Random news scatters; a structural cause stacks. I have kept a log of every year-end turn since 2018, and the elevated prints land, consistently, on the last one to three business days of December. That points at one mechanism: the balance-sheet snapshot.
Why the Spike Lands on the Last Business Days of December
Banks measure regulatory capital against snapshot balance sheets. The supplementary capital ratio that applies to the largest banks, plus the G-SIB surcharge, leans on quarter-end data, and the December 31 snapshot carries the most weight in the annual calculations. So dealers pull back from repo lending around the snapshot — right when government money funds would rather hold bills to maturity than lend overnight into the turn. Collateral supply meets shrinking intermediation capacity, and the overnight rate pops.
Since July 2021, the Fed’s standing repo facility has capped the pop. Banks can borrow against Treasuries at the top of the target range — 4.50% at present — and take-up concentrates in those final days. That is why recent turn premiums run 5 to 20 basis points instead of 2018’s 62. The spike got smaller. The clustering did not go away, because the snapshot dates did not move.
The Transmission Chain to a January Statement
Step 1: The Fed decision, prime, and the 7-to-45-day lag
The index side of a card APR moves only with the target range. The December 2024 cut of 25 basis points put the range at 4.25–4.50% and prime at 7.50% the same day; a card priced at prime + 14 reindexed from 21.75% to 21.50% at the start of its next billing cycle. Full reflection takes one to two statements — 7 to 45 days from the FOMC decision to the interest line. When the Committee holds, the index side just sits there. I laid out what a hold does and does not change for cardholders in What a Rate Hold Actually Means for Credit Card Borrowers.
Step 2: The turn premium as a credit-supply signal
The size of the turn premium is a stress gauge. A 62-basis-point turn like 2018’s says dealer balance sheets are pinned, and that kind of pressure shows up as credit-line decreases, stingier credit-limit increases, and tighter new-account approvals 30 to 90 days later — the same constrained balance sheets price every form of consumer credit. A 10-basis-point turn says the plumbing works, and the signal value for cardholders is close to zero. Recent prints put current conditions in the second category. That is a cautious read of the data, not a forecast.
Step 3: What the spike itself costs — about eight cents
Run the arithmetic on a $2,000 card balance carried through a 20-basis-point, seven-day turn: $2,000 × 0.0020 × 7 ÷ 365 ≈ $0.08. Eight cents. Even if a funding bump passed straight through — and it does not — the spike itself is a rounding error on a household balance. The durable version of the signal sits in term rates. One-month term SOFR typically prints 10 to 20 basis points above overnight SOFR in the second half of December, because the turn falls inside the term window. Issuers fund card receivables with a mix of deposits, fixed-rate securitizations, and SOFR-linked facilities, so a persistently elevated term structure is already priced into next year’s funding plans — and it shows up in margin decisions one to two quarters later.
Where the Real Year-End APR Increases Come From
January statements carry higher effective rates for reasons that have nothing to do with the turn. Four mechanisms, each on its own clock:
1. Promotional expirations — a 12-month lag, to the day. Zero-percent intro APRs opened during last year’s holiday spending expire on their 12-month anniversaries, in December and January. The card converts to prime + margin at the standard rate, around 24% on the average new-offer account. A $4,000 promotional balance that converts at 24% accrues roughly $80 a month from that point forward.
2. Deferred-interest cliffs — retroactive to the purchase date. Retail and medical cards with deferred interest run APRs near 27%. If any balance remains when the promo window closes, interest is billed on the original purchase amount from the purchase date. A $1,200 purchase from last December that comes up $200 short at expiry generates about $324 in back-billed interest — 27% on the full original amount, not on the $200 left.
3. Holiday utilization and risk-based reviews — a 30-to-60-day lag. November and December spending lifts statement balances, and the statement balance is what reports to the bureaus. Reported utilization jumping from around 10% to 40% typically costs 20 to 40 score points, and issuers that reprice on risk review pick that up within one to two cycles.
4. Margin resets tied to losses — a one-to-two-quarter lag. Card charge-off rates at the large issuers run in the mid-4% to low-5% range in 2025, up from roughly 2% in 2021. That loss curve is already priced into new-offer margins, and existing-account reviews apply it on the issuer’s calendar — annually or at trigger events, not on the Fed’s.
The pattern is hard to miss. The index moves on the Fed’s at-most-eight dates a year. Everything that actually bites in January moves on the issuer’s calendar — anniversaries, statement closes, review dates. And the margin is where the money is: the average 14-point spread costs a $4,000 balance about $560 a year, while a 25-basis-point index move is worth all of $10.
A Five-Minute Year-End Rate Check
- Pull the last three SOFR prints. The New York Fed publishes the series each morning, and FRED carries the daily history. Compare each print to the top of the current target range — 4.50% as of this writing, which is also the standing repo facility’s cap. Prints at or a few basis points below the ceiling: normal. Prints 20-plus basis points above: a tight turn, which points to tighter credit-supply signals in the first quarter, not a direct APR move.
- Check the December FOMC decision. Prime follows the range’s upper bound within a day, and your statement reindexes within one to two cycles.
- List every promotional end date. A 12-month no-interest window opened last December ends this month. Deferred-interest plans first — those bill retroactively.
- Pay before the statement close, not the due date. The statement balance is what reports. Bringing the balance down two days before the close is the cheapest credit-score protection available in December.
- Ask for a margin review. Issuers grant APR reductions on request for accounts with six to twelve months of on-time payments. A three-point margin cut on a $4,000 balance is about $120 a year; the phone call costs ten minutes.
Frequently Asked Questions
Does a SOFR spike directly raise my credit card APR?
No. Card APRs are indexed to prime, and prime moves only when the FOMC changes the target range. A 4.75% SOFR print on December 31 leaves a 24% card at 24%. The chain from the Fed to a statement runs: decision → prime the same day → reindex at the start of the next billing cycle → 7 to 45 days to full reflection.
What is the year-end “turn,” and how big does it get?
The turn is the balance-sheet-driven spike in overnight rates on the last business days of December, when banks shrink repo activity ahead of the December 31 regulatory snapshot. The 2018 turn put SOFR 62 basis points higher in one print — 2.41% to 3.03%. Since the Fed’s standing repo facility went live in July 2021 and began lending at the top of the target range, recent turn premiums run roughly 5 to 20 basis points.
How long after a Fed rate cut does my card APR drop?
One to two billing cycles. The issuer applies the new index at the start of the cycle, so the lag runs 7 to 45 days from the FOMC decision to the interest calculation. A cut announced in mid-December shows up in full on February statements.
Why does my APR feel highest in January?
Because the January increase usually comes from the issuer’s calendar, not the Fed’s: promotional APRs opened during last year’s holiday season expire on their 12-month anniversaries, deferred-interest plans bill retroactively if the balance was not cleared, and holiday utilization lifts reported balances, which risk-based reviews pick up within 30 to 60 days.
What I’m Watching Into January
Three prints and one release. The last three SOFR prints of December, against the 4.50% ceiling. Standing-repo-facility take-up on the final business day — a large number there says the cap is doing the work. And the one-month term SOFR spread over overnight, where 10 to 20 basis points is routine and 30-plus says the turn premium is leaking into 2026 funding costs. Then the G.19, released in early February, closes the loop on December revolving balances — a five-to-six-week lag from the swipes to the official print. I run this check every year-end, and the readings go into a follow-up column in early February. If the index side is what you need before the next FOMC week, the rate-hold primer linked above covers the half of card pricing the turn does not touch. Reader questions are welcome — the recurring ones go into this column’s FAQ each December.
