Walk down almost any side street in Seoul and you will pass a small storefront with three red letters above the door: MG. Most visitors read it as a phone shop, a gym, or another convenience store. In reality it is a lender, and 3,198 of these storefronts are scattered across the country. For three years now, they have sat at the heart of the Saemaul Geumgo crisis — a slow-burning stress test on ₩273 trillion, or roughly $195 billion, of Korean savings.
Foreign coverage of Korean finance tends to stop at the big five commercial banks, Kakao Bank, and the occasional chaebol credit story. Meanwhile, the country’s largest cooperative lender has been quietly shrinking, merging, and writing off bad property loans at a pace no Korean bank has matched. As a result, an institution most expats have never heard of now offers the clearest window into what Korea’s property hangover actually costs.
This is the story of how a 1963 village savings scheme became a ₩273 trillion balance sheet, why a ministry that also runs local government supervises it, and what any of that means if your salary lands in a Korean account every month.
First, the headline numbers. As of the end of June 2026, the nationwide Saemaul Geumgo network held ₩273.3 trillion in total assets, ₩243.2 trillion in deposits, and ₩181.8 trillion in loans. Its delinquency ratio stood at 6.34%. In the first half of 2026 alone, the network posted a net loss of ₩676.8 billion.
Those figures look bad in isolation. However, the direction matters more than the level. A year earlier, the same six-month period produced a net loss of ₩1.33 trillion — roughly double. In other words, the bleeding has slowed by about half.
At the same time, the delinquency number moved the wrong way. It closed 2025 at 5.08%, then climbed 1.26 percentage points in six months. Consequently, the cleanup that looked nearly finished in December suddenly looked unfinished again in June.
Here is the four-year arc in one line:
| Date | Delinquency ratio |
|---|---|
| End of 2024 | 6.81% |
| June 2025 | 8.37% |
| End of 2025 | 5.08% |
| June 2026 | 6.34% |
Notice the sawtooth. Korean mutual finance institutions tend to push bad-loan sales into the fourth quarter, which flatters the December print. Afterwards, the ratio rebounds as fresh arrears accumulate through the first half. Therefore, the honest read is not “5.08% then 6.34%” but rather a structural range somewhere in the six-to-eight percent band, with December figures dressed for the annual report.
To understand the numbers, you first have to understand the institution. A geumgo (금고) is a member-owned credit cooperative, not a branch of a national bank. Each one is legally independent. Each one has its own members, its own board, its own chairman, and its own balance sheet.
The first village fund opened in 1963 in Sacheon, a small city in South Gyeongsang Province. At that point it was a mutual-aid pot for farmers who had no access to formal credit. A decade later, in 1973, the movement got a central body — the Korean Federation of Community Credit Cooperatives, better known as KFCC, which still represents the sector internationally.
The name itself causes confusion abroad. Saemaul means “new village,” and it comes from the Saemaul Undong, the rural modernization campaign Park Chung-hee launched in 1970. Nevertheless, the two are not the same thing. The savings network predates the campaign by seven years; it simply absorbed the branding afterwards.
In practice, membership buys you three things. Firstly, you can open tax-advantaged deposits that ordinary banks cannot offer. Secondly, you receive an annual dividend on your share capital rather than pure interest. Thirdly, you get access to promotional rates that routinely beat the commercial sector — mutual finance institutions were advertising deposit rates near 4% in early 2026 while the big banks paid about 2.8% on a one-year term.
That gap is not generosity. Instead, it is the price of funding an institution that markets cannot easily price.
Now for the structural oddity that makes the Saemaul Geumgo crisis different from every other Korean financial story.
Banks answer to the Financial Services Commission. Savings banks answer to the Financial Services Commission. Insurers, brokerages, card issuers, and the internet-only banks all answer to the Financial Services Commission. Saemaul Geumgo, by contrast, answers to the Ministry of the Interior and Safety — the same ministry that oversees provincial governments and civil registration.
The reason is historical. Village funds began as community development projects, so they were filed under community affairs rather than finance. The filing never changed, even as the network grew past a quarter of a trillion dollars.
The consequences are concrete. Financial supervisors build examiner teams, stress-test models, and early-warning systems as a matter of routine. A general-purpose ministry does not. Reporting in August 2026 put the Financial Supervisory Service’s dedicated Saemaul Geumgo staff at roughly ten people — for more than 1,200 independent institutions.
Regulators have started patching the gap rather than closing it. In January 2026, four bodies — the Interior Ministry, the Financial Services Commission, the Financial Supervisory Service, and the Korea Deposit Insurance Corporation — launched a joint special management task force. Every week, the four share management indicators for troubled geumgo and coordinate joint inspections. Originally the task force was scheduled to wind down in June; instead, it was extended to the end of 2026.
Meanwhile, the Interior Ministry rewrote its supervisory standards by ministerial notice on 9 February 2026, tightening the criteria under which a geumgo is flagged. Bills to move credit-business supervision to the financial authorities are sitting in the National Assembly. Ultimately, the political question — who should own this ₩273 trillion problem — is expected to reopen once the task force reports at year-end.
A headline delinquency ratio hides more than it reveals. Break it apart and the picture becomes much sharper.
As of June 2026, the corporate loan book carried a delinquency ratio of 10.17%, up 2.40 percentage points from 7.77% at the end of 2025. The household book carried 1.90%, up a comparatively trivial 0.12 percentage points.
Read that again. One in ten corporate won is late. Fewer than one in fifty household won is.
Korean households, in short, are not the problem here. Ordinary borrowers — the people taking out jeonse loans, car loans, and small credit lines — are paying roughly on schedule. The damage sits almost entirely in the corporate column, and inside that column it sits in property.
Specifically, it sits in land-collateral lending and managed land trusts: the financing that funds a site before construction begins. When the Korean property project financing bust hit, those sites stopped moving. Nationwide, project financing exposure stood at ₩169.8 trillion as of March 2026, with the sector-wide project loan delinquency ratio at 4.65%. For land-collateral loans at savings banks and mutual finance institutions, however, the delinquency ratio reached a staggering 31.88%.
Almost a third. That is the real engine of the Saemaul Geumgo crisis. It also explains why a lender with a healthy household book can still lose money for three straight years.
Commercial property has offered no rescue either. Even in the capital, where office vacancy and rents are rising together, the recovery has been narrow and uneven.
Loan balances tell the same story from the other side. Corporate lending fell from ₩100.8 trillion to ₩97.5 trillion in six months, a 3.3% contraction. Household lending, by contrast, grew ₩2 trillion. Consequently, the network is deliberately walking away from the business that broke it.
Korea does not usually let financial institutions fail. Instead, it merges them.
In the first half of 2026, 21 troubled geumgo were absorbed into healthier neighbours, while 35 more underwent examination. Additionally, another 30 mergers were scheduled for the second half. Altogether, that is roughly 51 institutions disappearing from the map in a single year.
The cumulative effect is easy to miss because the storefronts often stay open under a new parent. At the end of 2025, Korea had 1,251 geumgo. Six months later, 1,239 remained. Since the deposit run of July 2023, 42 institutions have been consolidated away.
Supervisors have also been selling the bad assets rather than nursing them. During the first half of 2026, the network disposed of ₩3.48 trillion in non-performing loans. Notably, MG Asset Management — the network’s own bad-bank vehicle — absorbed ₩3.19 trillion of that, or 91.7%. The remainder went to NPL funds, the state asset manager KAMCO, and securitization structures.
That concentration is worth flagging. When a group sells nine-tenths of its bad loans to a subsidiary it owns, the loss does not leave the group; rather, it changes address. Analysts tracking Saemaul Geumgo’s bad-asset burden have made exactly this point.
How weak is the tail? As of June 2025, 159 of roughly 1,250 geumgo — 12.7% — carried a composite management rating of 4 or 5, the two lowest grades. Put differently, one geumgo in eight was classified as weak or at risk.
Here is the question that separates a crisis from a cleanup.
Between December 2025 and June 2026, total assets fell ₩13.4 trillion. Deposits fell ₩12.1 trillion. Loans fell ₩1.3 trillion. For most financial institutions, a balance sheet contracting at that speed would be an emergency.
There is a benign reading. Under it, management is shedding risky exposure on purpose, refusing to chase deposits it cannot profitably lend, and letting expensive promotional accounts run off. Certainly the net capital ratio, at 7.96%, remains above the regulatory floor. Similarly, the halving of net losses year on year fits a recovery narrative.
There is also a darker reading. Deposits do not shrink by ₩12 trillion because savers feel reassured. Korean depositors have spent three years reading headlines about mergers, embezzlement cases, and ratings downgrades. Some of that money has moved to the big commercial banking groups; some has moved into stocks and deposit-linked products elsewhere.
Tax policy is pushing in the same direction. Historically, mutual finance deposits enjoyed a powerful perk: interest on up to ₩30 million was exempt from the 14% income tax, subject only to a 1.4% rural development levy. From 2026, however, members earning more than ₩70 million in salary face a 5% separate tax on that interest, rising to 9% from 2027. For high earners, in other words, the single biggest reason to hold a geumgo account is being phased out.
Either reading can be defended with the same data. Nevertheless, the deposit number is the one to watch. Assets falling while loans hold steady is deleveraging; deposits falling faster than loans is something else.
If you live in Korea, this is where the Saemaul Geumgo crisis touches your own money.
Bank deposits in Korea are insured by the Korea Deposit Insurance Corporation under the Depositor Protection Act. Saemaul Geumgo deposits are not. Instead, they sit behind a separate statutory fund created under the Saemaul Geumgo Act. That scheme dates to 1983, and it was the first legally mandated depositor protection fund anywhere in the Korean cooperative sector.
The practical coverage is now the same. Since 1 September 2025, Korea’s deposit protection ceiling has been ₩100 million per depositor, up from ₩50 million, and the increase applied simultaneously to banks and to mutual finance institutions including Saemaul Geumgo, Shinhyup, and the agricultural cooperatives. Coverage runs per geumgo, not per network; branches of the same geumgo are aggregated. Furthermore, the fund provides emergency living funds of up to ₩20 million per person on an expedited basis when an institution fails.
So the protection is real. Even so, the plumbing differs, and the plumbing is under strain.
Consider how much the fund has been spending to finance mergers:
| Year | Drawdown for mergers | Share of the fund |
|---|---|---|
| 2023 | ₩40.4 billion | 1.5% |
| 2024 | ₩92.3 billion | 3.1% |
| 2025 | ₩360 billion | 11.8% |
The 2025 figure is almost four times the 2024 figure. Moreover, it means roughly one-eighth of the entire protection fund was consumed in twelve months. Should the second-half 2026 merger programme run to plan, another substantial draw follows.
None of this means deposits are at risk today. Rather, it means the buffer behind them is thinner than it was, which is precisely why four agencies now meet weekly about it.
One more practical note. Those 3.5% to 4% promotional deposits are genuinely attractive, and roughly 460 geumgo branches operate in 89 designated population-declining regions where no commercial bank bothers to stay. Before you chase the rate, however, check the individual geumgo’s management rating and capital ratio — each institution discloses them, and they vary enormously across the network.
Governance deserves its own section. Above all, it explains how so many small institutions made the same mistake at the same time.
Scandal has been a constant background noise across Korea community credit cooperatives. Embezzlement cases, lending irregularities, and losses tied to voice phishing and account takeovers have surfaced repeatedly at individual branches.
For most of its history, the network chose chairmen through delegate assemblies. Critics called the result hereditary. Accordingly, reformers pushed for direct elections, and in March 2025 Korea held the first nationwide simultaneous Saemaul Geumgo chairman election across 1,101 institutions.
The results were sobering. Of the 534 geumgo large enough to require a direct vote — those with assets above ₩200 billion — 326 attracted only a single candidate. Therefore, a genuine contest occurred at just 208 institutions, fewer than four in ten.
Turnout matched the enthusiasm. Of 1,752,702 eligible members, 451,036 voted: a turnout of 25.7%. By comparison, Korea’s simultaneous agricultural, fisheries, and forestry cooperative elections drew 79.6% — roughly three times higher.
Incumbents, predictably, did well. In Gyeonggi and Incheon, 93 of 143 winners were sitting chairmen. In Gangwon the figure reached 43 of 52, or 83%. Across Gwangju and South Jeolla, 50 of 85 incumbents held their seats.
The lesson is not that the reform failed outright; a first election is a first election. Nonetheless, an owner-members model only disciplines management when the owners show up. At 25.7% turnout with 61% of seats uncontested, the discipline is mostly theoretical — and a cooperative without effective owner oversight is exactly the kind of institution that lends aggressively into a property boom.
As the former Reuters correspondent Yoo Choon-sik argued in a Korea Herald column, the steady drip of small scandals at individual geumgo carries a cost that extends well beyond the network itself: trust in the financial system as a whole.
For foreign residents, five things follow.
First, the ₩100 million ceiling applies per geumgo, not per branch or per network. Two accounts at the same institution are aggregated; two accounts at different geumgo are not.
Second, check the individual institution, not the brand. MG is a shared signboard covering 1,239 legally separate lenders whose health ranges from solid to grade-5. Ratings and capital ratios are published.
Third, understand what you are buying with a promotional rate. A 4% term deposit at a small regional geumgo is a legitimate product and a legitimate risk premium. Both halves of that sentence are true.
Fourth, the tax perk is narrowing. If your Korean salary exceeds ₩70 million, the tax-free interest allowance that made these accounts compelling is being phased down from 2026.
Fifth, watch share capital separately from deposits. Membership shares (출자금) pay dividends and are treated differently from insured deposits. Read the product sheet before you assume equivalence.
Beyond personal banking, the network matters for anyone modelling Korean credit risk. Saemaul Geumgo is a large, concentrated, pro-cyclical lender to exactly the segment — regional developers and small commercial property — that Korea’s housing market correction has hit hardest. Its corporate delinquency number is arguably a better real-time gauge of regional property stress than anything published by the commercial banks.
Four markers will define the next twelve months of the Saemaul Geumgo crisis.
The first is the December merger count. Thirty institutions were slated for consolidation in the second half of 2026; whether that number lands, overshoots, or slips tells you how deep the weak tail really goes.
The second is the year-end delinquency print. Given the seasonal pattern, a December figure near 5% proves nothing on its own. A December figure above 6%, on the other hand, would confirm that the corporate book is still deteriorating faster than it is being cleaned.
The third is supervision. When the joint task force reports at the end of 2026, the argument over moving credit oversight from the Interior Ministry to the financial authorities returns to the National Assembly. Transferring it would be the single largest structural change to Korean mutual finance in decades — and, notably, it would require the financial regulator to staff up far beyond its current ten-person team.
The fourth is the protection fund. Another year of double-digit drawdowns would move the conversation from “which geumgo merges next” to “who recapitalizes the fund,” which is a fundamentally different and more political question.
For now, the storefronts stay open. The signs still say MG. Behind them, though, a quarter-trillion-dollar experiment in community banking is being rebuilt in public, one merger at a time. Korea, meanwhile, has not yet decided who should be holding the blueprints.
The year Vietnam passed China In April 2026, the number of Korea international students crossed…
For thirteen straight years, road deaths in South Korea fell. Then, in 2025, the line…
The Korea Housing Lottery: Apartments You Win, Not Buy On the first day of April…
In the third week of September 2026, the Korean stock market had one of its…
Walk into any Emart in mid-September and you will hit a wall of boxes before…
Every Chuseok, the country empties out in one direction at once. This year the government…