DSR is one line of arithmetic — annual principal and interest on all household debt, divided by annual income. And yet five points along that line catch people out again and again. None of them are hard. All five are cases of reading the formula differently from the way it is written.
Each one is shown with figures. Every number below came out of our own DSR calculator, so any of these cases can be re-run there with your own inputs.
1. The biggest gap. Counting interest only. A KRW 30m credit loan at 5.50% over five years costs KRW 1.65m in interest but enters DSR as KRW 6.88m — 4.2 times as much.
2. The second biggest. Ignoring the term. That same KRW 30m consumes 21.7% of DSR over three years and 7.8% over ten.
3. The one that runs backwards. Believing a regulated area tightens DSR — the Commission’s own note says “the same in regulated areas”.
One — count interest only and you have counted a quarter
The name is debt service ratio: principal and interest. The numerator includes repaid principal. Yet the calculation that starts “my interest is X a month” is extremely common.
| The same KRW 30m credit loan (5.50%, 5 years) | Per year | Against income of KRW 50m |
|---|---|---|
| Counting interest only | KRW 1.65m | 3.3% |
| What actually enters DSR | KRW 6.88m | 13.8% |
| Difference | KRW 5.23m | 10.5 points |
Ten and a half points is a quarter of the entire 40% allowance. Counting interest only and concluding “I have room” is precisely what produces the surprise at the branch. How and when principal actually leaves, by repayment method, is laid out instalment by instalment in the loan repayment calculator.
Two — the heavy thing is the term, not the balance
| Term | Repayment | Share | New mtg |
|---|---|---|---|
| 1 year | 30.9m | 61.8% | Nil |
| 3 years | 10.87m | 21.7% | 129.7m |
| 5 years | 6.88m | 13.8% | 186.4m |
| 10 years | 3.91m | 7.8% | 228.6m |
Figures in the table are in KRW, and the balance is KRW 30m in every row. Stretching the term from three years to ten adds KRW 98.9m of mortgage capacity. DSR does not ask how much you owe; it asks how much you must repay this year.
Do not read this table as “so stretch the term”. For credit and other loans the bank converts the balance to a set assessed period rather than the contractual term. Article 29-2(3) of the supervisory regulation hands the method of computation to the supervisor, and we have now opened that rulebook — set out below under the assessed terms are a table in the rulebook. What this table establishes is still only how far one box can move the answer.
The assessed terms are a table in the rulebook
Where Article 29-2(3) of the supervisory regulation points is Chapter 3 of Appendix 18 to the Detailed Enforcement Rules on Banking Supervision. Its <Table 3> — debt computation for DSR fixes, by loan type, over how many years principal is spread. It is not a figure each bank picks.
| Loan type | Years over which principal is spread |
|---|---|
| Credit loan (amortising) | 5–10 years |
| Credit loan (other) | 5 years |
| Interim payment · relocation loan | 25 years |
| Loan secured on a jeonse deposit | 4 years |
| Non-housing or securities-backed loan | 8 years |
| Other secured loan | 10 years |
| Card loan | 5 years (3 if not amortising) |
| Other loans | The next year |
What bites in this table is the notes.
· An overdraft line counts at its «full limit», not what you have drawn (note 1). Not a won drawn, and the whole limit still counts.
· “Amortising” has its own definition (note 3) — no grace period, and at least 40% of principal repaid in instalments by maturity. A product called amortising with a grace period does not qualify.
· An officetel interim loan with more than a year’s grace counts as bullet repayment (note 4).
· Cash advances count interest as an average of two months (note 8).
So the “3 / 5 / 10-year term” in the table above is, for a credit loan, not a number the bank chooses — 5 to 10 years if the amortisation test is met, otherwise a flat 5 years. Stretching the contractual term to 20 years still gets computed as 5. Lengthening the term to raise the limit works on the mortgage side, and there item 12-2(g) of the rulebook caps it at 40 years. Item 11(f) says outright not to “circumvent the rules by lengthening maturities”.
Item 12(b) of the rulebook lists eleven kinds of loan excluded from the DSR computation. “Is my loan an exception” is not the bank’s discretion but whether it is on that list — if it is not listed, it is not excluded. Where spouses’ incomes are combined to raise the limit, item 12-1(d) is explicit: “the spouse’s debt is combined as well.”
Three — the LTV limit is not your limit
“LTV is 40% in a regulated area, so a KRW 1bn house means KRW 400m.” True as far as it goes. The question is whether that KRW 400m gets through DSR.
| Annual income | What DSR 40% allows | LTV 40% of KRW 1bn | The actual limit |
|---|---|---|---|
| KRW 50m | KRW 284.1m | KRW 400m | KRW 284.1m (DSR) |
| KRW 70m | KRW 397.7m | KRW 400m | KRW 397.7m (DSR) |
| KRW 70.4m | KRW 400m | KRW 400m | the crossing point |
| KRW 90m | KRW 511.3m | KRW 400m | KRW 400m (LTV) |
Computed at 4.30% plus an assumed 1.50 point stress rate over 30 years; a capital-area or regulated-zone mortgage carries 3.00 points, which pushes the crossing point further right. Below an income of about KRW 70.4m the binding constraint is DSR; above it, LTV. Planning the deposit and balance around LTV alone, on an income below that line, is how a hole opens up at the balance payment.
There is a fourth ruler on top: the cap by house price — in regulated areas, KRW 600m up to a KRW 1.5bn house, KRW 400m from 1.5 to 2.5bn, KRW 200m above that. Your limit is the smallest of LTV, DTI, DSR and that cap. The order in which money leaves, and where it stops being reversible, is in the order to check before buying.
Four — the stress rate is not money
“The stress DSR put rates up 1.5 points” is a common sentence. Rates did not go up. The Commission’s own document settles it in a bracket — “not charged on the actual loan rate”.
| Aspect | What the stress rate does | What it does not do |
|---|---|---|
| Computation | Inflates the rate while DSR is measured | — |
| Result | Shrinks the limit | — |
| Monthly payment | — | Unchanged |
| Total interest | — | Unchanged |
In the earlier case — income KRW 50m, a KRW 30m credit loan over five years — the add-on at zero allows KRW 221.0m and at 1.50% allows KRW 186.4m: 1.50 points removes KRW 34.6m. The interest paid is identical in both. What changed is only how much you may borrow.
Which surfaces a fifth mistake — memorising “the stress rate is 1.5%”. That was the first half of 2025. The 15 October package raised it to 3.0% for capital-area and regulated-zone mortgages only, so for the second half of 2026 there are three figures.
| Category | Add-on | Limit | Versus 0% |
|---|---|---|---|
| Capital-area / regulated | +3.00 pts | KRW 159.5m | −KRW 61.5m |
| Credit / other (over 100m) | +1.50 pts | 186.4m | −KRW 34.6m |
| Regional mortgage | +0.75 pts | KRW 202.6m | −KRW 18.4m |
Buying in the capital area while planning on 1.5% means expecting to lose KRW 34.6m when the real figure is KRW 61.5m — KRW 26.9m of headroom that is not there, and it shows up at the balance payment. Two refinements worth knowing: credit loans attract the add-on only above a KRW 100m balance, and a longer fixed-rate period attracts a smaller share of it. Why those branches exist is set out against the rule text in the DSR rules explained.
Five — a regulated area does not tighten DSR
Designating an area tightens several things at once, so it is natural to assume DSR drops below 40% as well. The Commission’s own note says the opposite.
FSC, FAQ on the 15 October housing market measures — “Financial-sector lending is subject to DSR regulation (40% at banks, 50% at non-banks, the same in regulated areas)”.
What actually tightens is LTV (70% to 40%), DTI, and the cap by house price. The same document keeps first-time buyers at 70% LTV even in a regulated area and lower-income owner-occupiers at 60% — working out which category you fall into is the faster move.
The rule runs the other way too: DSR does not loosen outside regulated areas. Moving the search to a different district buys headroom on LTV, not on DSR.
The five, in one table
| The common belief | What is actually true | In figures |
|---|---|---|
| Interest is what counts | Principal counts too | KRW 1.65m → 6.88m |
| A small balance is fine | A short term is heavy | 3 yrs 21.7% vs 10 yrs 7.8% |
| The LTV limit is my limit | The smallest of four | They cross at KRW 70.4m income |
| You pay the stress rate | It only cuts the limit | Same interest, −KRW 34.6m |
| The stress rate is 1.5% | It splits by region and loan type | Capital-area mortgage +3.00 pts · regional +0.75 |
| Regulated areas tighten DSR | DSR is unchanged | Banks 40% · non-banks 50% |
If the plan is to repay something early to free up DSR, an early repayment fee comes with it — the early repayment fee calculator gives the figure. Jeonse loans are treated differently in the DSR computation and sit outside all of this; what banks charge on them is measured in the jeonse loan rate article.
None of the five come from not knowing the rule. They come from reading the formula differently: principal is in the numerator, the numerator is set by the term rather than the balance, and DSR is only one of several rulers.
Questions that remain
Doesn't DSR just count the interest?
The name is debt service ratio — principal counts too. Repaying a KRW 30m credit loan at 5.50% over five years costs KRW 1.65m in interest, but KRW 6.88m goes into the DSR: 4.2 times as much. On an income of KRW 50m that difference alone is 10.5 percentage points, more than a quarter of the 40% cap. Count interest only and you have counted less than a quarter of it.
Surely a small balance barely affects my DSR?
What decides the effect is the term, not the balance. The same KRW 30m consumes 21.7% of the DSR over three years and 7.8% over ten — a gap that moves the mortgage you can add by KRW 98.9m. DSR asks not “how much do you owe” but “how much must you repay this year”. Do not read that as “just lengthen the term”, though: banks convert credit and other loans to a set assessed term rather than the contractual one, and that basis sits in the supervisor's rulebook.
LTV in a regulated area is 40%, so a KRW 1bn home means KRW 400m — right?
Right up to that point; whether the KRW 400m clears DSR is the open question. On a KRW 1bn home with no existing debt, the two lines cross at an income of KRW 70.4m. Below that, DSR sets the limit; above it, LTV does. At an income of KRW 50m, LTV says KRW 400m but the real limit is KRW 284m. Plan the deposit and balance around LTV alone, on an income below the crossing point, and a hole opens at the balance payment.
Isn't the stress rate 1.5%?
That was the first half of 2025. The 15 October package raised it to 3.0% for capital-area and regulated-zone mortgages only, so for the second half of 2026 there are three figures: +3.00 points on a capital-area mortgage, +1.50 on credit and other loans, +0.75 on a regional mortgage. On the inputs above those give limits of KRW 159.5m, KRW 186.4m and KRW 202.6m. Planning a capital-area purchase on 1.5% means expecting to lose KRW 34.6m where the real figure is KRW 61.5m — KRW 26.9m of headroom that is not there.
Does a regulated area tighten DSR as well?
No. The Commission's guidance states it plainly — “identical in regulated areas”: 40% at banks and 50% at second-tier lenders, wherever the property is. What tightens in a regulated area is LTV (down 30 points), DTI (down 10 to 20), and the cap by property price band. The misconception points the wrong way: the thing to check is not DSR but LTV and the band cap.
Sources
- Financial Services Commission — press release — April 2026 household lending review. Carries, in footnotes, both the definition “DSR = annual principal and interest on all household debt / annual income” and the definition of the stress DSR. Confirmed in a 2026 document.
- Financial Services Commission — policy Q&A — FAQ on the 15 October housing market measures. Source for “40% at banks, 50% at non-banks, the same in regulated areas”, the 40% LTV in regulated areas, the price-band caps, and the 70% and 60% categories.
- Financial Services Commission — press release — stage 3 stress DSR confirmed. Source for the 1.50% add-on, the rule that credit loans attract it only above a KRW 100m balance, the reduced share for longer fixed-rate periods, and “not charged on the actual loan rate”.
- Financial Services Commission — “Stage 3 stress DSR — notice of amended administrative guidance” (FSC notice 2026-362, 18 June 2026), carrying the “stress DSR operating direction for H2 2026”, table 1. Source for 3.0% at a 100% base ratio on capital-area and regulated-zone mortgages, 1.5% at 50% for regional mortgages and 1.5% at 100% for everything else, running 1 July to 31 December 2026. These figures are set by administrative guidance, not a press release — which is why they are absent from the FSC’s press-release listing and sit instead in the financial regulation portal (better.fsc.go.kr), issued under article 9 of the Regulation on Financial Regulatory Operation.
- Korea Ministry of Government Legislation — administrative rule — Regulation on Supervision of Banking Business, article 29-2. Paragraph 3, delegating the method of computation and the detailed criteria to the supervisor, is the basis for “the bank converts the term”.
- National Law Information Centre — administrative rule — Detailed Enforcement Rules on Banking Supervision, Appendix 18, Chapter 3. Where article 29-2(3) points. <Table 3>’s periods by loan type (credit loans 5–10 years amortising, 5 otherwise; interim 25; jeonse-secured 4; non-housing and securities 8; other secured 10; card loans 5 or 3; other, the next year), note 1 (overdrafts at full limit), note 3 (amortising = no grace + at least 40% of principal), note 4 (officetel grace over a year), note 8 (cash advances, two months), item 11(f) (no circumvention by lengthening maturities), item 12(b) (eleven exclusions), item 12-1(d) (spouse’s debt combined with spouse’s income), item 12-2(g) (40-year cap) and <Table 3-1> transitional terms all sit in this appendix (checked 18 September 2026).
Where to check further
- Whether your credit loan is 5 years or 10. The rulebook sets amortising credit loans at “5–10 years” — a range. Which point in it applies is a question for your bank.
- How much of your income is recognised. Documented, imputed or declared income — and at what figure — changes the denominator. Ask your bank first.
- Whether your loan is one of the eleven in item 12(b). We have read the list, but which item a particular loan falls under depends on its terms — the Financial Supervisory Service and your bank can say.
- Whether transitional terms apply to your loan. <Table 3-1> of the same appendix sets different periods by drawdown date (credit loans drawn by 30 June 2021: 10 years; 1 July 2021 to 2 January 2022: 7 years, and so on) — your drawdown date is what decides it.
Written as at August 2026. The stress rates are those in the operating guidance for the second half of 2026 (1 July – 31 December), and the worked examples assume 1.50 points. The ratio caps and stress rates were read directly from Financial Services Commission documents, and every figure in the tables is our own calculation on an equal-total-payment basis. Assessed terms and the exception list were filled in from Appendix 18 of the supervisor’s rulebook (read 18 September 2026; this English page updated 20 September) — before that, the terms here were assumptions. A real limit also turns on how the bank recognises income and on the product terms — confirm at the branch immediately before you draw down. This is general information, not financial advice.
