1
Tuesday morning, nine o'clock. The day begins in a tenth-floor conference room of a savings bank in Gyeonggi Province.
The building stands in the middle of an industrial park on the outskirts of the Seoul metropolitan area. Exit the expressway interchange and a four-lane road stretches straight ahead, flanked by the corrugated steel roofs of auto-parts factories. Wedged between the factories are convenience stores, kimbap shops, and labor attorneys' offices. The lettering on the signs is stained with grime, and the roadside trees have turned gray with fine dust. Rising out of this landscape is a single ten-story glass building — rather clean by industrial park standards. The savings bank's name is displayed on a sign affixed to the exterior wall, and above the ground-floor entrance hangs a banner reading: "We will safeguard our customers' precious assets."
Step into the first-floor lobby and the marble floor is polished clean. Two potted tropical plants sit to the left. Past the security gate, the deposit counter is on the right. Morning business hours have not yet begun, but three or four customers are already sitting on the lobby benches clutching numbered tickets. Term deposit maturity withdrawals, new installment savings accounts. A man in his sixties has come to open a five-million-won term deposit; a couple in their thirties wants to put part of their lease deposit into an installment savings plan. Their money will be discussed on the tenth floor. The only ones who know that are the people on the tenth floor.
Take the elevator to the tenth floor and the conference room is on the left side of the corridor. It is too small to properly call a conference room. A twelve-seat rectangular table fills the space almost wall to wall, and through the windows you can see a parking lot and a cluster of gray apartment blocks. Two rows of fluorescent tubes are recessed into the ceiling; the third light in the right-hand row flickers faintly — it was reported to facilities management but has gone unreplaced for two months. The air conditioning is on, but the airflow is concentrated on one end of the table, so the head of the risk management team, who always sits there, drapes a thin cardigan over her shoulders. On the whiteboard along one wall, numbers someone wrote during last week's agenda item linger in faint smudges. "LTV 62%," "Presale rate 68%." The erased numbers are traces of decisions already made. Today, new numbers will go up on this whiteboard.
The thin smell of instant coffee from the vending machine drifts through the conference room. In my private equity days, it was Nespresso capsules; at the investment bank, hand-drip pour-over. At the savings bank, coffee is a 200-won coffee-mix from the vending machine at the end of the corridor, dispensed into a paper cup with brown sediment settling at the bottom. After enough years in this business, you learn that the quality of the coffee has nothing to do with the quality of the decisions.
The head of corporate planning opens his laptop and pulls up an Excel file. In the center of the screen, a single cell highlighted in yellow catches the eye. It is the BIS ratio as of the end of last month.
The BIS ratio. This single number determines the fate of a savings bank. Its full name is the "Bank for International Settlements capital adequacy ratio," and in plain terms, it works like this. A bank does business with other people's money — depositors' money. If loans go bad, the losses must ultimately be covered with depositors' funds. The BIS ratio shows "for every 100 won in risky assets, how much of the bank's own capital has been set aside." Think of it this way: if you lend out one million won and have 80,000 won of your own money on hand, your BIS ratio is 8 percent. That 80,000 won is the buffer that absorbs losses when loans go sour. For savings banks with total assets exceeding one trillion won, once this ratio drops below 8 percent, the Financial Supervisory Service (FSS) — Korea's financial regulator — begins to take notice. Drop further, and the interventions arrive in sequence: a management improvement recommendation, then a management improvement demand, then a management improvement order. Three warning lights, each harsher than the last. At the final stage comes suspension of business.
That is how Busan Savings Bank shut its doors in 2011. On February 17, the Financial Services Commission ordered a suspension of operations. The month before, in January, the designation of Samhwa Savings Bank as a distressed institution had been the opening shot. Led by Busan Savings Bank, three affiliated savings banks shuttered simultaneously, and later that year Daejeon, Bohae, Jeonju, and Domin savings banks followed one after another. The dominoes did not stop there. When the FSS launched a full inspection of all 85 savings banks at the time, sixteen were hit with business suspensions in that single year alone. Twenty-four savings banks ultimately closed, and approximately 38,000 depositors suffered losses. The total number of victims reached an estimated 100,000. Of the 76 trillion won in total savings bank deposits — a record high as of the end of 2010 — 32 trillion won was withdrawn. It was a bank run. The Korea Deposit Insurance Corporation injected over 27 trillion won in public funds, yet by the end of 2022, only half — 13.8 trillion won — had been recovered. The other half was covered by taxpayers.
From the second half of 2022, the real estate project finance (PF) market began freezing over again at an alarming pace. The Legoland crisis was the detonator. In September 2022, Gangwon Province declared that it would not honor its payment guarantee on asset-backed commercial paper (ABCP) issued for the Legoland Korea development project, sending a wave of panic through the entire real estate PF market. The very fact that a local government could break a guarantee undermined the credibility of every PF guarantee in the country. Then in December 2023, Taeyoung Engineering & Construction — ranked 16th in construction capability — filed for a workout. It could not bear 3.2 trillion won in PF guarantee liabilities. When a construction firm in the top twenty went down, it meant the financial safety net of dozens of project sites for which that firm had pledged completion guarantees was shaking all at once.
This savings bank's BIS ratio, too, has been moving in an unsettling direction. Factor in contingent liabilities and the capital soundness is at a worrying level. The head of corporate planning updates this number every week. The official figure is finalized at quarter-end, but there is no luxury of waiting until then. If the ratio comes in at 7-something percent at quarter-end, the FSS will send in inspectors. Once they come in, it takes three months for them to leave. In three months, there is no telling where the market will go.
"Let's start with the PF delinquency status."
The risk management team leader opens the discussion. Adjusting the sleeves of her cardigan, she places a single-page A4 summary on the table. Five real estate PF loans are listed. Three are printed in black, one in orange, and one in red. Orange means caution; red means danger. The color coding requires no further explanation.
"The Yongin project has had unpaid interest for three months running. The developer's CEO has stopped answering calls."
In this industry, when a developer stops answering the phone, it's over. Once a developer cuts off communication, all that remains is collateral disposal. But in a frozen real estate market, there is no guarantee that auctioning off unsold apartment land will recover the principal. Unsold inventory in the Yongin area already runs into the thousands of units. If the property sells at auction for 60 percent of appraised value, that would be fortunate. If it goes for 40 percent, the remaining 40 percent is pure loss. And simply getting the property to auction takes six months to a year. During that time, not a single won of interest comes in.
"I visited the construction site with the general contractor," another team member says. "The progress reports don't match what's actually on the ground. The paperwork says 40 percent, but the site is still at the foundation stage. There's no sign of any structural framing going up. I checked the concrete-pouring records — they're at half the documented volume, and there's no tower crane on site."
Silence.
This silence carries weight. When progress reports do not match the site, it suggests that milestone billings may have been inflated. Loan proceeds may not have gone toward construction at all, but been siphoned elsewhere. Developers inflating milestone claims to divert loan funds is one of the most common patterns of PF fraud. It was the very heart of the 2011 savings bank crisis — the gap between paperwork and reality. Park Yeon-ho, the majority shareholder of Busan Savings Bank, set up over 120 special purpose companies (SPCs) and executed more than 4.5 trillion won in loans. Of approximately 7 trillion won in total loan assets, PF loans accounted for 65 percent. At the time, the average PF exposure across the savings bank sector did not exceed 20 percent — his was more than triple the norm. On paper, everything looked clean. The audit reports raised no flags, and the PF feasibility reviews were all in order. Behind those documents, money was flowing to places it was never meant to go.
"Has the FSS inspection schedule been announced?"
"I've heard it's set for the middle of next month."
The sound of a pencil rolling across the table fills the room. The head of corporate planning has a habit of absently spinning his pencil. An FSS inspection date being confirmed means someone is already looking at this savings bank.
Even after the briefing ends, the tension in the room does not lift. Everyone is thinking the same thing: in this state, they still have to proceed with the new PF loan review scheduled for this afternoon. Approving a new loan would increase risk-weighted assets and push the BIS ratio down further. But cutting back on lending would reduce interest income and erode profitability. Cut back and it is dangerous; expand and it is dangerous. This is daily life for a savings bank on the brink of a management improvement recommendation — hands caught in the gears of real estate PF, unable to stop the machine, unable to keep it turning.
Through the window, smoke rises from a factory chimney in the industrial park. From this building's tenth floor, the landscape of outer Gyeonggi Province is gray in every direction. Factory roofs, apartment blocks, the asphalt of parking lots. Within that gray landscape, the flow of hundreds of billions of won is being decided.
2
Two o'clock in the afternoon, the same conference room. The scene on the table has changed since morning. A binder sits in front of each seat — three centimeters thick, at least. Color-coded index tabs protrude from the sides: "Business Plan," "Appraisal Report," "Financial Statements," "General Contractor Credit Rating," "Sales Feasibility Review." Each time someone opens a binder, the plastic tabs click against one another. The cover of each binder is stamped with a date and agenda number. Today's item is a mixed-use residential-commercial development on the outskirts of Gyeonggi Province.
The Credit Committee — or yeoshin simsawiwonhoe. In legal terminology, it is "the collegiate body that deliberates and resolves the appropriateness of credit extensions." In plain language: the group that decides whether this money gets lent. At least two-thirds of the seated members must be present, and at least two-thirds of those present must vote in favor for a motion to pass. These procedures are prescribed by the Mutual Savings Bank Act, Article 12, and its enforcement decree. The FSS's "Detailed Supervisory Regulations for Savings Banks" adds a further emphasis on the committee's independence: majority shareholders and senior management must not influence the review process. The minutes must record the names of all attendees, a summary of the agenda item, a synopsis of each member's remarks, and the result of the vote. These minutes are among the very first documents the FSS examines during an inspection. That is the principle, at any rate.
Five people are present today.
The chairman — held by the head of the lending division. He sits at the head of the table. Early fifties, though his hair is already half gray, and silver-rimmed glasses perpetually rest near the tip of his nose. He spent twenty-five years in corporate finance at a major commercial bank before moving to the savings bank. His role is to preside, not to speak. For most of the meeting he turns pages and listens. He has a habit of clicking his ballpoint pen cap open and shut with his right hand. Click, click. When the review drags on, the clicking quickens. When a presentation does not sit well with him, the clicking stops, and he gazes at the RM over the top of his glasses. That gaze is sharper than any question. His single remark can steer hundreds of billions of won, but that remark does not come until the meeting is nearly over. He learned long ago that waiting is power.
The senior credit analyst — the person who analyzed the deal's risk and authored the report. Early forties, lean build, glasses. A red ballpoint pen and a highlighter are lined up in front of him. Post-it notes dot his binder, and the appraisal report pages are underlined in red. He knows the deal's weaknesses better than anyone. He is the room's examiner. He has read the binder's numbers enough times to recite them from memory, and he has prepared a separate A4 sheet compiling actual sales performance data from nearby complexes. His questions are terse and incisive. He always prepares his documents with zero margin for error. If he is wrong, his name will be on the report when the FSS comes.
The risk management team leader — she looks not at the individual deal but at the entire portfolio. Not "this one loan" but "what happens to the whole when this one loan is added." Her laptop always has a portfolio simulation spreadsheet open — a sheet where PF exposure, BIS ratio changes, and the liquidity coverage ratio update in real time when a new deal is entered. When the numbers turn red, she shakes her head. Today she has not shaken her head yet, but the hand adjusting her cardigan is tense.
The compliance officer — he looks only at regulatory violations. LTV limits, single-borrower exposure caps, aggregate PF ceilings. If it runs afoul of the law, the merits are irrelevant — he votes no. There is no compromise. Early sixties, the oldest person at the table. He is a former FSS official. From a brown document envelope, he produces printed copies of the relevant regulations, and as each agenda item comes up, he highlights the applicable provision with a marker. When this man says "That is not permissible," there is no room for discussion. His role is not to interpret the law but to apply its text as written. Occasionally his eyes meet the chairman's, and he conveys his position with the slightest shake of his head. His dissent does not automatically defeat a motion, but his opposing opinion is recorded in the minutes. And the FSS reads the minutes.
And the relationship manager (RM) — the person who sourced and nurtured this deal. Late thirties. His necktie is neatly knotted, but there is a ballpoint ink smudge on his dress-shirt sleeve. He is on the developer's side — he has to be; that is the only way the deal survives. He sits in the chair closest to the door. A position that says he can step out at any time once his presentation ends. At the same time, it is also the seat most directly exposed to the room's crosswinds. Beside his rolled-up sleeve, a smartphone lies on the table. This morning, the developer's CEO sent him a KakaoTalk message: "Please let me know the result today." That message weighs on him. Months of visiting project sites, reconciling appraisal values, securing the general contractor's completion guarantee, requesting a legal opinion from a law firm, running credit evaluations — all of that time is riding on these two hours in this room.
The RM opens his laptop and switches on the projector. The ceiling-mounted projector hums softly, and a blue glow spreads across the white wall. Twelve PowerPoint slides. The first shows the project overview. The RM's voice is clear, the tempo slightly fast — he has clearly rehearsed the presentation several times. Each time the laser pointer in his right hand picks out a number on the slide, the red dot trembles faintly against the wall. He is nervous.
Project: Mixed-use residential-commercial development, OO-dong, OO City Scale: 2 basement levels, 29 above-ground floors; 280 apartments + 120 officetels (studio units that double as small offices) + ground-floor retail Total project cost: 185 billion won Requested loan amount: 68 billion won (bridge loan to main PF conversion) Collateral: Land held in collateral trust + general contractor's completion guarantee Appraised value: Land at 102 billion won (LTV basis: 67%) General contractor: Mid-tier construction firm (credit rating BBB+, ranked approximately 40th by construction capability) Presale price: 16.5 million won per 3.3 square meters
Sixty-eight billion won. To convey the weight of this number to a non-specialist: the average annual salary in South Korea is roughly 40 million won, so 68 billion won represents the combined annual earnings of 1,700 people. If the man in his sixties sitting in the first-floor lobby with his numbered ticket came to deposit five million won in a term account, this single loan equals the pooled deposits of 13,600 such people. It represents a substantial share of this savings bank's total equity capital. Five people in this cramped conference room are deciding whether to lend it — under flickering fluorescent lights, with vending-machine coffee going cold.
The RM's presentation continues for ten minutes. Flipping through slides, he lays out the project's strengths. A subway extension as a positive catalyst; presale prices competitive relative to surrounding market values; the general contractor's completion guarantee; the developer CEO's track record — two successful sales campaigns in the neighboring area. "The specialized floor plan achieves a three-bay layout." "The school district is excellent — two elementary schools and one middle school within a one-kilometer radius." The numbers change with each slide, but the narrative is singular: this project can work.
On the numbers, it does not look bad. Total projected sales revenue of 234 billion won. Break-even presale rate of 73 percent. Project IRR of 11.2 percent per annum. General contractor ranked in the top 40 by construction capability. Financial ratios are sound as well — debt-to-equity of 180 percent, operating margin of 6.3 percent. A completion guarantee has been secured. By the numbers alone, it is a loan worth making.
But no one in this conference room looks at numbers alone.
3
The mood in the room shifts the moment the credit analyst begins to speak.
"Let me start with the presale rate assumption."
He produces a separate sheet of A4 paper. It is a table compiling the actual sales performance of comparable nearby complexes. The type is dense, with no margins. Key figures are circled in red ballpoint. To produce this single page, he cross-referenced the Ministry of Land, Infrastructure and Transport's real-transaction price database, Real Estate 114, and on-site visits.
"The break-even point is 73 percent. But look at the actual recent presale rates of comparable nearby complexes: Complex A at 64 percent, Complex B at 58 percent, Complex C at 71 percent. The average is 64 percent. The presale price of 16.5 million won per 3.3 square meters is roughly 8 percent above prevailing market levels. If you bring the presale price down to market level, the break-even point rises to 81 percent."
Numbers hang in the air. 73 percent, 64 percent, 58 percent, 71 percent, 81 percent. In the gaps between these numbers, the fate of 68 billion won is suspended. A nine-percentage-point discrepancy. A chasm between the business plan and reality.
The RM pushes back. "Complex A was not near a subway station. Our site is on a confirmed subway extension route, scheduled to open in 2027, which justifies a premium. And the specialized floor plan—"
"Is the subway extension confirmed?" The credit analyst cuts in.
"It passed the preliminary feasibility study."
"And groundbreaking?"
A brief hesitation. "It's scheduled for the second half of next year."
Scheduled. Not broken ground — scheduled. In real estate development, "scheduled" is another way of saying "not yet done." Between passing a preliminary feasibility study and actual groundbreaking lie the mountains of budget allocation, design, bidding, and land compensation. A 2027 opening becoming 2029, then 2031, is common in this industry. The Phase 3 extension of Seoul Metro Line 9 was originally meant to open in 2014; it was not completed until 2024. A ten-year delay. That is the scale of uncertainty packed into a single word. In sales brochures, "scheduled" is used as though it means "confirmed" — model-home pamphlets print the subway route map as if the line already runs. But at a credit committee table, "scheduled" is merely scheduled.
"Complex B's 58 percent presale rate is the bigger concern," the credit analyst continues. "Complex B is two kilometers from our project site. Same living zone. It has been a full year since completion and they're still at 58 percent. That means unsold units are not clearing. When unsold inventory becomes chronic, the developer cannot sustain the interest burden."
The RM pushes back again. "Complex B had issues with its view corridor. Our project site is predominantly south-facing—"
"RM." The credit analyst's voice is calm but firm. "Every underperforming project has an explanation. Complex A wasn't near a subway station, Complex B had a poor view corridor, Complex C had an uninspired floor plan. Yet the result is the same across all three — below 73 percent. The individual reasons differ, but the market's signal is consistent. Breaking through 73 percent in this area is not realistic in the current environment."
The RM's face tightens slightly. His hand, turning through the presentation materials, stops.
The credit analyst moves to his second point. "The general contractor's credit rating is BBB+. Before the Taeyoung crisis, that rating would have been adequate. But circumstances have changed. Taeyoung was rated A. When an A-rated construction firm enters a workout because of PF guarantee liabilities, how should we read BBB+? One notch down and it's BB+ — speculative grade. If the general contractor falls to speculative grade, the meaning of its completion guarantee is diluted."
The risk management team leader interjects.
"Let me address this from a portfolio perspective. Our current PF exposure stands at 18.3 percent of total credit extensions. The regulatory cap is 20 percent, so adding this deal leaves us with only 1.7 percentage points of headroom. If another deal comes in next quarter, we breach the cap."
She swivels her laptop screen to show the simulation results. A single cell in the spreadsheet has changed from yellow to orange. There is still room before it turns red, but that room is a single number: 1.7 percentage points.
This is a problem on an entirely different plane from the merits of the individual deal. Even if this deal were superb, if real estate PF accounts for too large a share of the total portfolio, systemic risk grows. It is putting too many eggs in one basket. In 2011, Busan Savings Bank's PF exposure exceeded 65 percent of its loan assets. The 20 percent regulatory cap itself was a fence erected after that catastrophe.
"And one more thing," the risk management team leader adds. "If this deal goes through, the PF exposure reported to the FSS inspection next month will not be 18.3 percent but 19.8 percent. I think everyone here knows what an inspector's reaction will be when they see a portfolio 0.2 percentage points from the cap."
The conference room falls quiet again. Only the hum of the air conditioning remains. Under the flickering fluorescent light, the expressions of the five people point in different directions.
The compliance officer studies his documents and asks, "Do you have the general contractor's completion guarantee — the executed original, not the draft construction contract?"
The RM flips through his binder. "... At this point, it's prepared as a draft construction contract, and the original guarantee will be obtained before loan disbursement—"
"Can we proceed with deliberation without the original?" The compliance officer looks at the chairman. In his hand is a copy of the regulations, a passage underlined in highlighter: "Deliberation shall proceed after obtaining the original completion guarantee from the general contractor" — the applicable provision of the Detailed Supervisory Regulations. His gaze over the top of his glasses is sharp.
Silence.
The chairman's pen-clicking has stopped.
This silence is the essence of the credit committee. No matter how high the numbers and logic are stacked, at the moment of decision, a gap opens that words cannot fill. Whether a 73 percent presale rate is realistic, whether the subway will be built on time, whether the general contractor will see it through to the end — no one has a definitive answer to these questions. The only certainty is that nothing is certain, and in the face of that uncertainty, someone must say "go."
Adding further weight: this deal is a consortium. This savings bank is not acting alone; it is providing capital alongside two other financial institutions. The other two have already passed their credit reviews. That fact is noted in the reference materials tab at the back of the binder. A rejection here would shake the entire consortium. The relationship with the developer would be severed. The deal the RM has spent months cultivating would dissolve into nothing.
That pressure never appears on the review table, yet everyone in the room feels it. The other consortium members will ask, "They already approved it over there — why is it being blocked here?" The developer's CEO will say, "This savings bank killed the project." The financial market is small. Once a reputation is damaged, it is hard to rebuild. The RM's performance review, the team's sales targets, the division head's management evaluation — everything is connected to the decision made in this small conference room. At a table where the only currency should be credit judgment, forces that have nothing to do with credit are at work.
When the majority voice says "go," saying "no" alone is not easy. When a gatekeeper's judgment becomes collective, the edge of that judgment is, paradoxically, dulled. In a structure where everyone shares responsibility, no one bears it entirely. This is the structural paradox of collegial decision-making. Autocracy is dangerous, but consensus suppresses bold refusal.
The chairman speaks for the first time. It is roughly one hour and forty minutes into the meeting. He sets his ballpoint pen down on the table with a small sound, and at that sound, every pair of eyes turns to him.
"Let's move it forward with conditions. Subject to obtaining the original completion guarantee from the general contractor, recalculation of the presale rate assumption at 65 percent, and an LTV re-review."
Conditional approval. Neither a full endorsement nor a clean rejection. But in practice, everyone in the room understands this is closer to "yes." If the conditions are met, the loan will be disbursed. And in most cases, conditions are met — because people make sure they are met. The condition "recalculate with a 65 percent presale rate assumption" will most likely yield a conclusion that profitability still holds. Once that conclusion is reached, the remaining conditions are processed procedurally. The original guarantee can be obtained from the general contractor; the LTV re-review can be handled by adjusting the appraiser's figures. A conditional approval is, in practice, an approval — except that it gives everyone in the room a line of defense: "I never said yes without conditions."
The vote proceeds. The chairman calls each name.
"Senior credit analyst."
A moment of hesitation. He rolls his red ballpoint pen once across the table. "... Conditional approval."
"Risk management team leader."
"In favor. With the proviso that re-submission be requested if the recalculated presale rate falls below a certain threshold."
"Compliance officer."
He looks at his documents one more time. His eyes linger on the highlighted provision. "I abstain until the original guarantee is obtained."
"The RM is a party of interest and has no voting rights."
In favor, in favor, conditional approval, in favor, abstention. The motion carries. These days, many institutions anonymize the votes so that it is impossible to tell who voted how — the rationale being that anonymity enables honest judgment, though it simultaneously blurs the locus of accountability. There was one abstention. The weight of that abstention will depend on what outcome this deal ultimately meets.
The minutes are drafted. The resolution is recorded. The moment that seals the flow of 68 billion won is quiet — disproportionately so for the size of the room.
4
Before I entered this world, I was in another one.
Private equity — PE. The business of pooling investor capital, investing in companies or assets, increasing value, and selling at a profit. What I did as CEO of my PE firm was, at its core, the same thing: judging "whether this money should go here."
But the atmosphere of the two worlds was entirely different.
A PE investment committee (IC) moves fast. An office on Teheran-ro in Gangnam, Seoul. Through floor-to-ceiling windows, the city skyline is visible. The conference table is wider than the one at the savings bank, but instead of binders, laptop screens are open. The IC is usually conducted in English. Even at Korean PE firms, the investment committee is called the "IC," and pitch materials are prepared as English-language PowerPoint decks. Around thirty slides. The first is an Executive Summary — the deal's essence compressed into a single page. "What's the thesis?" "What's the exit multiple?" "What's the downside case?" English and Korean mingle at a rapid clip. "What's the IRR?" "What's the deal-breaker?" Four or five key figures per slide. In thirty minutes, the direction is set. When there is conviction, they move fast. Miss the market's timing and the deal vanishes.
The PE PowerPoint is clean. McKinsey-style structured slides. Market Overview, Investment Thesis, Valuation, Key Risks, Exit Strategy. One message per slide. Numbers are visualized in charts and tables, with key metrics color-coded — green for positive, red for warning. "This company is worth 30 billion won right now — if we get involved, can it become 50 billion?" That is the core question. Of course PE considers downside risk. But the frame's center of gravity tilts toward opportunity.
The savings bank's credit review materials are different. Not PowerPoint but binders. Inside a three-centimeter-thick binder, densely printed A4 documents are separated by index tabs. Appraisal reports, land registry transcripts, building permits, business plans, financial statements, legal opinions. No design, only figures and text. Not glamorous, but the information density is higher than a PE deck. If the PE slide asks "why this deal is good," the savings bank binder asks "why this deal is dangerous." The same information, organized through diametrically opposite frames. The center of gravity tilts toward risk. "What happens if this loan goes bad?" is the starting point of every question.
There is a reason it must be this way. The money PE manages belongs to institutional investors and high-net-worth individuals. They understand risk and accept it. The investment prospectus says "principal loss is possible," and they have signed on the line. But the money a savings bank manages belongs to depositors. A retiree in his sixties who entrusted his pension. A young couple who parked their lease deposit. A small-business owner who faithfully contributes to an installment savings plan every month. They do not even know their money has gone into real estate PF. They do not think they need to know — the bank will manage it well on their behalf. The Depositor Protection Act covers up to 50 million won. Those are the people sitting in the first-floor lobby.
When I crossed from PE to the savings bank, the hardest adjustment was this difference in decision-making frames. The same capital allocation, but an entirely different grammar. In PE, you prove "why this deal should be done." At the savings bank, you prove "why this deal is safe." It is not so much the difference between offense and defense as a difference in the direction of one's gaze. In PE, you look out the window at the horizon. At the savings bank, you scan the cracks beneath your feet.
There is one scene I remember. During my PE days, we were reviewing the acquisition of a mid-sized company. The numbers were good. The acquisition price relative to EBITDA was reasonable, and the synergy scenario was convincing. I had finished my presentation to the investment committee and was fielding questions. The most senior partner asked: "What excites you most about this deal?" That was the language of PE. Excitement. Opportunity. Upside.
After I came to the savings bank, while reviewing a PF deal of similar scale, the question I received was different. "If this blows up, how much do we lose?" That was the language of the savings bank. Loss. Defense. Downside.
And one more thing. People protect themselves. In PE, bold judgment is praised; at a savings bank, conservative judgment is a virtue. The cost of being wrong is categorically different. If a PE fund takes a loss, investors are furious and the next fundraise becomes harder — but the general partner (GP) does not go to prison. If a savings bank loan goes bad, the BIS ratio drops, the FSS arrives, names appear in the newspaper, and executives face prosecution. After the 2011 Busan Savings Bank scandal, approximately seventy people were arrested: majority shareholders, executives, credit analysts, marched into courtrooms one after another. The weight of managing other people's money is something you cannot understand until you have carried it yourself.
At a PE investment committee, I reviewed deals worth hundreds of billions of won and asked, "What's the exit multiple?" At a savings bank credit committee, I asked, "Is the presale rate assumption realistic?" The questions took different forms, but in the end, they were the same question:
Can this person pay the money back?
5
That question is not new.
Six hundred years ago in Florence, a branch manager of the Medici Bank faced the same reckoning. Should he extend credit to the Duke of Burgundy? Should he accept the wool trader's bill of exchange? Before him lay the counterparty's reputation, past repayment history, and the uncertainty of Mediterranean shipping routes — and he weighed it all without a calculator, on the scales of experience, intuition, and personal networks.
Only the form of the documents has changed. The bill of exchange became a credit review report; the gold florin became the Korean won; a remittance that once took twenty-five days from Florence to Bruges now settles in real time. But the act of poring over a pile of documents and agonizing over "Can this person repay?" — the structure of that act has remained remarkably unchanged.
The Medici Bank had a secret ledger — the libro segreto. It was a leather-bound volume recording each partner's capital contributions, each branch's profit and loss, and outstanding loan balances. It was clasped shut with a metal lock, and only three people could open it. In 1950, when the economic historian Raymond de Roover discovered these secret ledgers in the Florentine state archives, the instrument for taking the temperature of a six-hundred-year-old financial empire was revealed. Total capital of the Venice branch: 12,000 florins. Headquarters' contribution: 10,500 florins. Branch manager's contribution: 1,500 florins. Each branch's profitability, changes in partner capital, the details of bad loans — what a modern management information system (MIS) does, they were doing with parchment and ink.
Today's savings bank has its own secret ledger — the BIS ratio estimation Excel file. The official figure comes out quarterly, but the people in the trenches update the number every week, taking the bank's temperature. Yellow highlighting stands in for parchment ink; pivot tables stand in for ledger pages. The parallel bears closer examination. The Medici ledger recorded: "How much of other people's money are we managing, and how much of our own money do we have to survive if that money is lost?" The savings bank's Excel file records total risk-weighted assets, equity capital, and supplementary capital items. The form has evolved from parchment to spreadsheet, but the function is the same. A safeguard given to the gatekeeper. A dashboard that tells you: "How much danger are we in right now?"
And the way gatekeepers fail is also the same.
One of the core reasons for the Medici Bank's decline was that Tommaso Portinari, the branch manager in Bruges, escaped headquarters' control and extended excessive credit to the Burgundian court. Portinari had entered the Bruges branch as an apprentice at thirteen, and he managed it from 1465 to 1480. Fifteen years. Over that time, he became deeply enmeshed in the court of the Duke of Burgundy, Charles the Bold. Invited to the court's banquets, recognized as one of the Duke's inner circle, he grew intoxicated with his social standing. Lending money to a monarch was an act of purchasing political influence, and Portinari was seduced by that influence. He bought the wool monopoly at the port of Gravelines for 16,000 francs per year, but when the Duke banned English wool imports, the investment evaporated after 1471. The allure of the court, not the numbers in the ledger, governed his judgment. By the time the Bruges branch was liquidated in 1478, cumulative losses exceeded 70,000 gold florins. One man's lapse in judgment hastened the fall of Europe's greatest financial empire.
The majority shareholder of Busan Savings Bank in 2011 followed a similar pattern. He turned the credit committee into a rubber-stamp body for decisions already made, filled the compliance officer's seat with his own people, and channeled 4.5 trillion won in illicit loans through 120 SPCs. Where Portinari was blinded by the splendor of the Burgundian court, Busan Savings Bank's majority shareholder was blinded by the profits of real estate development. The tools were different — bills of exchange versus SPCs, wax seals versus corporate registrations — but the pattern was identical. A field manager who had slipped free of headquarters' control ignored the parent organization's rules and extended excessive credit.
It was not the system that failed. It was the human being operating the system.
The credit committee, the BIS ratio, the CAEL rating, prompt corrective action — all these mechanisms are safeguards placed upon human gatekeepers. Rules to assist judgment, standards to constrain discretion, penalties to deter deviation. The Medici Bank had its own safeguards — an internal policy of "avoid doing business with princes and nobles wherever possible," the requirement that branch managers invest their own capital, the supervisory apparatus of the headquarters. Over the centuries, the form of these safeguards evolved from parchment to legal statutes, from a family's internal code to a regulator's detailed supervisory provisions. But it is humans who create safeguards, and it is humans who neutralize them.
If six hundred years of history reveal a pattern, it is this: human gatekeepers work well most of the time, but when they fail, they fail together with the safeguards. When Portinari ignored the rules, Lorenzo de' Medici was indifferent to the bank's management. When Busan Savings Bank's majority shareholder distorted the credit review process, the FSS inspection arrived too late. Safeguards fail to function at the very moment they are needed most. That is the paradox.
6
It is past four in the afternoon. The credit committee has ended, and only cold coffee and binders remain in the conference room. Five paper cups. Only two still have coffee in them. The other three are completely empty. Two hours and forty minutes of deliberation.
The RM was the first to leave. By the emergency exit at the end of the corridor, he pulls out his phone and calls the developer's CEO. "Yes, it passed with conditions. We'll need the original guarantee and recalculated presale rate materials. Yes, it's doable. Thank you." His voice is bright. The tension from the conference room has lifted. After hanging up, the RM gazes out the window for a moment. Afternoon sunlight falls on the parking lot. To him, "conditional approval" is effectively "approval." Meeting the conditions is his job, and he is confident he can do it. He ends the call with the developer, then immediately texts the general contractor's project manager: "Is it possible to have the original guarantee by the end of this week?" The deal's momentum is pushing him forward.
The credit analyst is organizing the minutes. He summarizes each committee member's key remarks, clearly documents the conditions attached to the conditional approval, and double-checks the abstention notation with his ballpoint pen before placing the file in its folder. He knows that these minutes could become a shield protecting him if this deal later becomes a problem. The single line — "The senior credit analyst raised objections based on actual sales performance of comparable nearby complexes regarding the presale rate assumption" — could prove important down the road.
The risk management team leader has returned to her desk and opened her laptop. She updates the PF exposure Excel file. From 18.3 percent to 19.8 percent. Another yellow highlight is added. On her face, one reads tension rather than relief. She is likely thinking about how 19.8 percent will be interpreted during next month's FSS inspection.
The compliance officer sighs while waiting for the elevator. It troubles him that the deliberation proceeded without the original guarantee. His abstention is recorded in the minutes, so he believes he has done his part, but the unease lingers. The phrasing of the supervisory regulation loops through his mind.
The chairman remains alone at the head of the table, closing his binder. He gathers the empty paper cups into one spot and slips his ballpoint pen into his breast pocket. He glances once out the window. The industrial park sky is tinged with the afternoon light. Sixty-eight billion won. The number lingers like the weight of the pen in his breast pocket.
The flow of 68 billion won has been decided. After five people debated for more than two hours.
This is what gatekeepers do. It is not glamorous. It is not dramatic. They read documents, scrutinize numbers, pose questions, and render judgment in the face of uncertainty. And they bear responsibility for the consequences of that judgment. Get it wrong and your name stays in the minutes; get it badly wrong and the FSS arrives; get it catastrophically wrong and prosecutors come. Get it right? No one remembers. The credit analyst who prevented a bad loan does not become a hero. The ones who make the news are always the credit analysts who failed to prevent one.
For six hundred years, capital allocation has operated this way. The tools have changed, but the structure has remained the same. The Medici bill of exchange, the Bank of England's government bonds, the modern savings bank's PF loan — different names, all running on the same engine: "Humans judge, humans decide, humans bear the consequences."
That this engine is robust is beyond doubt. But this afternoon, in this conference room, an uncomfortable truth also revealed itself. The engine is slow. Reviewing a single loan took two hours. It is expensive. Five specialists had to be deployed simultaneously. And it is vulnerable to bias. Consortium pressure, the temptation of hitting targets, deference to superiors — forces that cloud pure judgment are always present inside the conference room. That "abstention" exists as an option is itself a testament to the system's limitations. Neither "yes" nor "no," but a third choice that sidesteps accountability.
Until now, these costs — the slowness, the expense, the vulnerability to bias — were worth bearing. There was no one besides a human being who could make this judgment.
But at this very moment, something else is doing the same work.
At the same hour that five people in this conference room spent two hours deliberating over 68 billion won, a server in New York is rebalancing a $14 trillion portfolio in milliseconds. On the Ethereum blockchain, smart contracts are automatically approving hundreds of loans every twelve seconds. No documents, no meetings, no seals. Code sets the rules, and code enforces them. No fingers turning appraisal reports, no chairman clicking a pen, no one casting an abstention vote.
All three are doing the same work: capital allocation. But the methods are as different as three things on this planet can possibly be.
Whether this savings bank's conference room is the final form of capital allocation, or whether it remains the best we have — to find that answer, we must first understand how the other two methods arrived at this point.
In 1973 in Chicago, two mathematicians published a formula that put a price on uncertainty. From there, the story begins to diverge.