In July I wrote on this blog that the Fed was weighing a hike rather than a cut. Two months later it hiked — 25 basis points on 16 September, the first increase since 2023. I am not writing this to claim a call. What actually surprised me was something else: the Fed raised rates and US Treasury yields fell.
The mortgage rate outlook for Korean borrowers turns on exactly this point. It is natural to reason that if the policy rate rose, your loan rate follows. The bond market moved the other way. This piece is my attempt to set down how that mismatch should be read.
What the Fed did, and what the market read
The decision itself is simple. The target range moved to 3.75–4.00% on a unanimous vote. On the dot plot, 12 of 18 members projected one more hike during 2026, four projected two, and two projected none. The median points to one more.
The market reaction, however, did not follow the textbook.
| Indicator | Day of decision |
|---|---|
| US 2-year Treasury | about -1bp |
| US 10-year Treasury | -4bp |
| US 30-year Treasury | -5bp |
| S&P 500 | +0.3% |
| Nasdaq | +0.7% |
Long yields fell and equities rose on a hike. The market appears to have read this not as the start of tightening but as a one-shot credibility move — and therefore as a signal that hiking ends soon. To my eye that reading has grounds.
The word that disappeared from the statement
References to supply shocks and energy-driven inflation were dropped from this statement, replaced by “resilience in domestic spending.” That substitution tells you the character of the decision.
US CPI for August came in at 3.4% headline and 2.4% core. The energy index rose 16.3% year on year and gasoline 27.4%. The Bureau of Labor Statistics noted that gasoline accounted for over one third of the monthly all-items increase. The fact that supply, not demand, produced this inflation is written plainly into the statistics.
A principle I learned in uniform came to mind here. When a commander cannot act on a cause, the mission gets redefined around managing the symptom. Oil is a variable the Fed cannot control. Yet responsibility for missing the inflation target still sits with the Fed. So the uncontrollable cause was deleted from the statement, the seemingly controllable domestic demand was put in front, and the rate was raised. Chair Warsh’s line — “inflation is too high and has been for too long” — reads less like policy logic than like a declaration of credibility.
Why oil crossed $100 I covered in last week’s piece: a conflict over who administers passage through Hormuz, sitting on top of the oil price. The cause the Fed deleted from its statement is still running.
The starting point of the mortgage rate outlook — Korean market rates are already ahead
These figures come straight from the Bank of Korea’s ECOS system, as of the 17 September close.
| Indicator | Value | As of |
|---|---|---|
| BOK policy rate | 3.00% | 2026-09-15 |
| CD 91-day | 3.20% | 2026-09-17 |
| KTB 3-year | 4.063% | 2026-09-17 |
| KTB 5-year | 4.262% | 2026-09-17 |
| Corporate 3-year (AA-) | 4.726% | 2026-09-17 |
| KRW/USD (close) | 1,382.2 | 2026-09-17 |
| Household credit | KRW 2,019.8tn | 2026 Q2 |
The number I looked at longest is the 3-year KTB at 4.063%. The policy rate is 3.00%. The spread is 106bp. That reads as a bond market which has already priced coming hikes.
This matters for a specific reason. In a market where the pricing-in is done, an actual hike does not push market rates up by the same amount again. The news cycle reacts to “the BOK will hike too”; loan rates move along a curve that has already risen. Being surprised by an announcement and judging from a price are different activities.
Is a 0.75%p Korea-US gap a warning sign?
The US is at 3.75–4.00%, Korea at 3.00%. The inversion is 0.75–1.00%p. Read on its own the number looks like capital flight, but I would not conclude that. In episodes where the gap widened to 2%p, capital outflow did not operate as the decisive variable; foreign holdings in the bond market and the current account acted as buffers.
Not staking a conclusion on a single indicator is basic to intelligence work. The rate gap is one of several variables explaining currency pressure, and in this phase oil looks like the variable that comes first. Oil lifts import prices, import prices erode the current account, and that path into the exchange rate is thicker than the rate gap.
Is the currency’s direction turning?
The won closed at 1,382.2 per dollar on 17 September. Its intraday low of 1,334.7 on 7 September was the strongest since October 2024, and it has given back more than 47 won in ten days. From 1,368.6 on the day of the decision, another 13.6 won was added in a single session.
Widen the frame to three months, though, and the picture inverts. Monthly averages ran 1,528.0 in June, 1,488.9 in July, 1,404.4 in August and 1,350.7 in early September. The won had been strengthening throughout. This hike is an event that braked that trend; calling it the opening of a new weakening phase would be premature.
So I watch the persistence of the rebound rather than the absolute level. Whether ten days of rebound is a trend reversal or a correction becomes input for the October rate-setting meeting.
The mortgage rate outlook — two paths ride on different things
Assuming floating and fixed rates move together will throw your judgment off. They ride different benchmarks.
| Type | Benchmark | Current level | Transmission |
|---|---|---|---|
| Floating | COFIX (new-origination) | 3.18% (August, stopped rising) | deposit rates → COFIX → reset every 6 or 12 months |
| Fixed | 5-year bank debenture | 4.577% (highest since Mar 2023) | US long yields → bank debenture → immediate |
On the floating side the issue is lag. COFIX stopped rising in August, but banks raised deposit rates across the board in September. COFIX reflects what banks actually paid to fund themselves, so that enters from the next release. And existing borrowers reset every 6 or 12 months. Even in a month when COFIX is flat, anyone whose reset date comes due takes the earlier increase.
On the fixed side the issue is direction. Bank debentures track US long yields. On 15 September the US 10-year stood at 5.030%, the highest since July 2007. Then, on the day of the decision, that yield fell. If the move persists, upward pressure on fixed rates may ease first.
To put it plainly: a policy-rate hike reaches floating rates with a lag, while a fall in US long yields reaches fixed rates immediately. There can be a stretch where the two move in opposite directions. Bank spreads move separately, however, so estimating a loan rate from benchmark rates alone is unsafe.
Checking the mortgage rate outlook — the actual numbers at the bank counter
These are Korean bank mortgage products pulled from the Financial Supervisory Service’s product-comparison API: 34 products, all 108 rate options.
| Type | Rate range | Average floor rate | Options |
|---|---|---|---|
| Floating | 3.62 – 6.87% | 4.76% | 62 |
| Fixed | 4.74 – 7.66% | 5.48% | 46 |
Fixed costs 0.72%p more on average. I consider this the most practical piece of information in the article. The price tag on stability is posted at 0.72%p a year.
On a simple calculation for a KRW 300 million, 30-year equal-payment loan, 4.76% gives roughly KRW 1.57m a month and 5.48% roughly KRW 1.70m. That is about KRW 130,000 a month, or KRW 1.56m a year. The figure shifts with each bank’s spread and preferential terms, so treat it as indicative.
The 22 October meeting — three variables I am watching
The real branch point is not the Fed but the Monetary Policy Board on 22 October. I watch three things.
Oil. Brent and WTI have both passed $100 a barrel. The cause is the US-Iran conflict, which monetary policy cannot address. In June I described the gap between inventories and prices as a compressed spring. That spring is now releasing not as a supply cut but as a repricing of passage.
The currency. Above 1,382, further weakness passes into domestic prices through import costs. Conversely, a return to the strengthening trend lightens the BOK’s burden.
Household debt. Household credit passed KRW 2,000tn in the second quarter, at 2,019.8tn. Every hike lifts the interest burden on that stock. I take this to be the single biggest reason the Bank of Korea cannot mechanically follow the Fed. Korean CPI in August was 3.1%, with the index excluding food and energy at 3.4%.
- Path A — oil settles. Hormuz tension eases and oil slips below $90, weakening the case for further Fed hikes and leaving the BOK on hold. The most favourable path for floating-rate borrowers.
- Path B — one more hike, then stop. The path the dot-plot median points to. The BOK follows once in October or November and then stops. I consider this the most likely.
- Path C — oil rises further. If $120 a barrel gives way, inflation re-accelerates and the number of hikes grows. In that case the 0.72%p fixed-rate premium is justified after the fact.
When the mortgage rate outlook is uncertain — how I set the criterion
The substance of the question is that the mortgage rate outlook is not clearly one-sided. So I think it is better not to look for the answer by getting the rate call right.
When we planned operations we did not stake the plan on guessing the adversary’s intent. We first built a posture that could hold whatever choice they made, and judged from there. Rates can be handled the same way. Calculate the monthly payment if your floating rate rose another 1%p. If you can carry that figure, there is room to stay floating and manage the reset dates. If you cannot, the fixed rate’s 0.72%p is better understood as an insurance premium than as the cost of a forecast.
If you are weighing a refinance, early-repayment fees belong in the calculation — typically 1–1.5% within three years. There is a range in which the fee offsets the 0.72%p difference.
The limits of this analysis
What I treated as fact and what I left as judgment
Treated as fact: the Fed’s statement and economic projections, the Bureau of Labor Statistics inflation release, the Bank of Korea ECOS API responses, the Financial Supervisory Service product API responses, and the Korea Federation of Banks COFIX release. Secondary financial-media figures were cited in one place only — the five-bank rate ranges — and attributed.
Left as judgment: reading the statement’s substitution as “deleting an uncontrollable cause,” the assessment that the market’s dovish reaction has grounds, and the view that Path B is most likely. These three are my judgment, not fact. The next COFIX release and the October decision will test them.
I should add that AI tools were used alongside the research and drafting of this piece, because of the speed they add when cross-checking API responses from several institutions against one common reference date. Every figure was re-verified against the original release and the API response, and the interpretation and judgment are my own.
Frequently Asked Questions (FAQ)
Not mechanically. The BOK had already moved pre-emptively to 3.00% on 27 August, and the Korea-US gap is 0.75–1.00%p. In past episodes where the gap reached 2%p, capital outflow did not operate as the decisive variable. Household credit of KRW 2,019.8tn also acts as a constraint on further hikes. The 22 October meeting is the real branch point.
Because the bond market prices coming hikes in advance. The spread is currently about 106bp, which reads as the market expecting a good part of further tightening. It also means that, with the move already priced, an actual hike may not lift market rates by that much again.
Two reasons overlap. First, floating rates reset every 6 or 12 months, so an earlier COFIX increase can be reaching you now. Second, a loan rate is COFIX plus the bank’s spread, and the spread is adjusted separately. Banks raising deposit rates in September is a factor that will enter the next COFIX release.
Not directly, but there is a path. More than a third of the August US price increase came from gasoline, and Korea takes the same effect through import prices. If oil eases, price pressure lightens and the case for further hikes weakens. Oil, however, depends on the US-Iran conflict and is hard to forecast.
📚 References
- Federal Reserve — FOMC statement and Summary of Economic Projections, 16 September 2026
- US Bureau of Labor Statistics — Consumer Price Index, August 2026, released 11 September 2026
- Bank of Korea ECOS Open API — 100 key statistics (policy rate, KTB, CD, FX, household credit), retrieved 18 September 2026
- Financial Supervisory Service product-comparison Open API — full mortgage query (34 products, 108 options), retrieved 18 September 2026
- Korea Federation of Banks — COFIX disclosure, August 2026 (released 15 September 2026)
- Statistics Korea — Consumer Price Trends, August 2026
- Five-bank rate ranges and 5-year bank debenture level: Money Today, 15 September 2026