Who Deserves Korea’s Housing Credit?

As South Korea expands state-backed guarantees for housing project finance, weak absorption in Busan is putting greater weight on the judgments that determine which developments receive public credit.

Busan’s private apartment market produced an unusually weak reading in the second quarter of 2026. According to Korea Housing & Urban Guarantee Corporation statistics carried by KOSIS, the city’s average initial sales rate fell to 17.2 percent from 71.0 percent in the previous quarter, while Seoul recorded 100 percent. The measure is narrower than a citywide gauge of housing demand: it covers a changing cohort of privately developed apartment projects that had been on the sales market for more than three months but no longer than six months and had received housing-sale guarantees from HUG. Because the composition changes from quarter to quarter, a small number of poorly performing launches can exert an outsized influence on the result. Even with that limitation, the figure describes a market in which forecasts about future buyers carry unusual weight, since slower absorption leaves less room for optimistic assumptions to be corrected by stronger demand later.

Against that backdrop, South Korea is preparing to place substantially more state-backed credit behind residential development. Measures announced on Aug. 13 call for 23 trillion won in public project-finance, or PF, guarantees in 2026 and 33 trillion won in 2027, alongside larger restructuring funds and private-sector lending facilities for viable developments that have lost access to financing and for troubled projects already in need of reorganization. The announced guarantee capacity establishes how much credit public institutions are prepared to support, while the path to construction depends on which sites qualify, whether banks ultimately lend against the guarantees and whether the apartments that reach the market attract enough buyers to sustain repayment. Long before work begins, projected land costs, construction expenses, sale prices, presales, interest burdens and the timing of future cash flows must be converted into a judgment about whether a development can support its debt.

Korea’s shrinking PF market has made that judgment more consequential. Total exposure fell from 231.1 trillion won at the end of 2023 to 169.8 trillion won at the end of March 2026, yet lenders have continued to extend new money selectively. New PF lending reached 16.8 trillion won in the first quarter of 2026, 50 percent more than a year earlier, with financial regulators saying the flow was concentrated in developments with stronger business prospects and more advanced progress. Capital has continued to circulate inside a smaller market, increasing the importance of the institutions that decide which sites remain eligible for financing and which no longer do.

State guarantees alter that process by reducing the risk borne by private lenders. A sound development can lose access to credit when banks retreat broadly from a region or asset class even though enough buyers are likely to exist once the homes reach the market, and a public guarantee can make financing possible again. The same mechanism can prolong a weak project when the forecasts supporting the guarantee overstate future demand. The policy challenge is to separate developments whose economics remain sound after credit disappears from those whose prospective customers are insufficient to support the debt, a judgment that becomes especially demanding in markets such as Busan, where recent apartment absorption has been markedly weaker than in Seoul.

From guarantee capacity to project selection

The Aug. 13 package combines financial tools that perform different jobs. HUG and the Korea Housing Finance Corporation, or HF, provide guarantees for qualifying developments. The Korea Asset Management Corp. operates a normalization fund for troubled PF sites, while syndicated lending by banks and insurers and other financial-industry funds provide refinancing or restructuring capital under separate arrangements. The government places the broader PF guarantee and financing framework at 47.8 trillion won or more, a headline number that spans public guarantees, private credit and restructuring vehicles. Treating the full amount as a single government cash injection would obscure how capital and risk are distributed and erase the difference between financing a viable development and repairing one already in distress.

Housing production depends on the sequence that follows. A developer submits a proposal, the relevant institution assesses the business and financing structure, a guarantee may be approved and issued, and the lender must complete its own credit decision before funds reach the site. Groundbreaking, presales, completion and repayment arrive later and on different timetables, with each stage revealing something different about the original decision. A guarantee ceiling records the amount of support available, an approval records an institutional judgment, construction shows that financing reached the physical project, and sales and repayment eventually test the assumptions used to justify the deal.

Korea encountered the distance between capacity and approval during an earlier expansion. In September 2023, the government raised HUG’s PF guarantee capacity from 10 trillion won to 15 trillion won while easing several eligibility conditions. According to HUG data submitted to the National Assembly, the corporation received 109 applications and approved 10 during the first nine months of the year; from October through December, another 154 applications produced 11 approvals. Those figures do not constitute a controlled before-and-after comparison because reviews take time, the pool of applicants changed and cases submitted late in the year could have been decided later. They nevertheless show why a larger guarantee ceiling cannot be read as an automatic increase in financeable housing developments. Policy can expand capacity immediately, while individual sites still have to clear a review process built around uncertain future revenues and costs.

HUG’s PF business has grown sharply since then. Guarantee approvals rose from 1.642 trillion won in 2022 to 2.983 trillion won in 2023, 8.372 trillion won in 2024 and 11.633 trillion won in 2025, when 73 projects received approval. By the end of June 2026, outstanding PF guarantees had reached 20.4319 trillion won, roughly twice their level at the end of 2024. Available figures for the period do not show a comparable surge in PF guarantee claims, so the expanding balance cannot be treated as evidence of an existing loss crisis. What has grown instead is the volume of future outcomes tied to judgments made before the commercial performance of recently approved sites can be observed.

The 33 trillion won goal planned for 2027 adds an institutional burden to the financial expansion. A larger program brings more sales forecasts to test, more construction schedules to reconcile, more cash-flow models to interrogate and more local housing markets to understand. Balance-sheet capacity can be raised through a policy decision, whereas experienced review, internal challenge and the ability to identify implausible assumptions depend on people, procedures and evidence that are harder to expand at the same speed. The durability of the program will depend in part on whether a much larger flow of applications can be processed without allowing pressure to deploy credit to weaken scrutiny of individual sites.

Inside the credit decision

Housing PF begins with projections about assets that do not yet exist and cash that has not yet been earned. Developers estimate land and construction costs, schedule payments across the life of the project, forecast apartment prices and assume a pace at which buyers will sign contracts and deliver proceeds. Lenders and guarantors use those assumptions to judge whether debt can be serviced as money moves through construction and sales. Ratios and grades give the process structure, but their apparent precision rests on forecasts about events that remain uncertain when the guarantee is issued.

HUG sits between the developer and the lender when it evaluates a site for a PF guarantee. Its review considers the economics and financing structure of the undertaking as well as the companies responsible for carrying it out, and measures including cumulative debt-service coverage and initial sales performance enter the assessment. Contractor strength still matters because lenders need confidence that a building can be completed, yet completion capacity cannot establish that households will purchase the finished units at the price or speed assumed by the financing model. A large builder can execute construction efficiently and still deliver homes into a weak market.

Recent changes to HUG’s guarantee structure have shifted more weight toward the economics of the development itself. A special PF guarantee for projects involving smaller construction companies reduced the weighting assigned to the contractor from 35 points to 30 and increased the weighting assigned to project viability from 65 to 70. The design seeks to prevent an otherwise workable development from losing access to credit merely because its builder lacks the balance-sheet strength or industry standing of a major group. Giving greater weight to project economics places correspondingly greater demands on the reliability of projected revenue, costs and demand.

A development can generate enough revenue over its lifetime and still run short of cash if large construction payments fall due before sufficient sales proceeds arrive. Moving a payment several months backward or forward changes the cash position visible at a particular stage without changing the eventual construction bill. Faster assumed presales pull revenue into the model sooner, while a higher projected selling price improves expected receipts before any buyer has signed a contract. A coverage ratio can therefore be arithmetically correct and still rest on commercial assumptions that prove too generous.

The surrounding housing market places limits on those assumptions. A plausible sales forecast has to account for competing supply, unsold inventory, recent transaction prices, household demand and the relationship between the proposed selling price and nearby offerings. Citywide statistics cannot determine the fate of an individual address, but the local market constrains what can reasonably be expected from that address. A sales pace that looks conservative in a fast-moving part of Seoul may be ambitious in a Busan submarket where comparable units are taking longer to clear, and a model that ignores the difference can appear financially coherent while embedding a weak revenue assumption.

HUG’s own research gives empirical weight to the role of local conditions. A 2026 study examined 94 developments that had received HUG PF guarantees between 2021 and July 2025 and found a statistically significant negative relationship between initial sales performance and the number of unsold homes in the surrounding market. Researchers divided the sample into three groups; the weakest-performing cluster of 12 developments recorded an average initial sales rate of 40.24 percent and was surrounded by an average of 2,506 unsold homes, while local price conditions were also weaker. The contractors attached to those sites were not unusually weak: their average construction-capacity ranking was 33rd, compared with 41st in the most stable group.

Every site in the sample had already passed the PF guarantee process, and the researchers cautioned against extending the results beyond that population without further validation. The study therefore cannot establish that HUG should have rejected those developments. It does show that sales outcomes can diverge sharply after approval and that surrounding inventory and market direction are associated with part of that variation. Corporate strength alone was insufficient to neutralize weak local demand, leaving the quality of the underlying market assumptions as an important part of any credit decision that relies increasingly on project economics.

Expected prices, sales velocity, construction costs and payment schedules must be tested before they are compressed into a score. Contracts can provide evidence about construction obligations, comparable transactions can anchor pricing assumptions, competing supply can inform expected absorption and the physical sequence of construction can be checked against the timing of projected cash needs. The final number acquires credibility only to the extent that the chain of evidence beneath it has survived those tests.

When timing changes the score

A July 30 audit showed how vulnerable that chain can become when one of its assumptions is insufficiently tested. According to the Board of Audit and Inspection, some developers submitted financing models in which future construction payments were scheduled in ways that allowed cumulative debt-service coverage ratios to remain at or above the required level. HUG accepted the schedules without sufficiently verifying whether the assumed timing corresponded to the projects’ actual obligations. Three developments that would otherwise have fallen below the relevant screening standard received a combined 382 billion won in PF guarantees.

The total construction costs remained in the models, but their placement in time was enough to change the financial picture. Deferring a major payment improved the cash position visible at an intermediate stage and made projected debt service appear more manageable even though the eventual construction obligation had not changed. A formal ratio could therefore move from failure to acceptance because the chronology beneath it had shifted, demonstrating how closely the apparent strength of the model depended on the assumptions used to build it.

Forward-looking finance cannot eliminate that uncertainty because a guarantor deciding whether to support construction cannot wait until completion to discover whether sales forecasts and payment schedules were correct. The work occurs earlier, when uncertainty has to be disciplined through evidence. Construction-payment assumptions can be reconciled with contracts and expected progress; sales forecasts can be tested against nearby supply and transaction data; unusually favorable departures from observed market conditions can be examined before they become embedded in a favorable score.

The audit also found grading errors in two other projects that resulted in 1.64 billion won in guarantee fees being undercharged. Those cases involved a different mechanism, but they reinforced the same operational vulnerability. Formal criteria work only when they are applied consistently to individual files, and weaknesses that appear manageable in a small number of cases become more consequential as the value and number of guarantees increase.

A larger guarantee program therefore requires greater review capacity as well as greater financial capacity. HUG and HF are being asked to move more financing toward developments capable of reaching construction while continuing to screen out sites whose repayment prospects do not justify support. Excessive conservatism can strand sound projects when private lenders have retreated too broadly, while weakly tested forecasts can allow state-backed credit to sustain developments whose sales assumptions depend on buyers who may never appear. Protecting viable sites while screening out demand-deficient ones depends on the same review process that the new program is now asking to operate at substantially greater scale.

Korea is enlarging that support while its statutory framework for development-project oversight is still being phased in. The law’s business-feasibility evaluation provisions took effect in May 2026, while project-level reporting requirements covering such information as financing structures, approvals, construction progress and sales are scheduled to take effect in May 2027. HUG, HF, private lenders and financial regulators already hold substantial information, so the system is not operating in a data vacuum. The unresolved institutional issue is how consistently information held across agencies, lenders and individual sites can be applied as the number of public credit decisions increases.

State guarantees may also acquire greater regulatory significance from 2027. Financial regulators plan to raise required equity ratios for PF lending in stages, using capital structure more explicitly in prudential treatment. Projects backed by HUG or HF are among those the government has identified as possible candidates for exceptions where effective risk is considered lower, although the scope has yet to be finalized. A favorable guarantee decision could consequently influence both a lender’s exposure to loss and the regulatory conditions attached to the financing, increasing the value of the status assigned by a public underwriter and the consequences of getting that assignment wrong.

A national model meets Busan

Busan provides a demanding environment in which to observe whether national criteria remain sensitive to local market conditions. Its 17.2 percent initial sales rate in the second quarter does not describe the entire housing market, and it does not cover every development carrying a PF guarantee. The quarterly sample changes, making direct comparisons across periods vulnerable to shifts in project composition. Reading the fall from 71.0 percent in the first quarter as a 53.8-percentage-point collapse in citywide housing demand would therefore claim far more than the statistic can support.

The quarter nevertheless captured a cohort of recently marketed apartments selling into weak conditions. Slower presales delay cash receipts and keep a larger share of funding dependent on debt for longer, while limited pricing power leaves less revenue available to absorb cost overruns or higher interest expenses. Forecasts that could survive modest error in a faster market become more exposed when buyers arrive slowly, giving local absorption a direct bearing on the cash-flow assumptions used to finance construction.

Weak metropolitan data cannot determine the viability of every site. Demand across Busan varies by neighborhood, redevelopment status, price point, competing supply and buyer profile, and a well-located development with limited competing inventory may remain commercially sound even when citywide indicators are poor. Lenders responding to broad regional risk can also become overly conservative toward individual projects, creating precisely the kind of financing gap that a public guarantee is designed to bridge.

Regional weakness can also coexist with developments whose financial models depend on assumptions the market is unlikely to support. Two sites with similar leverage, construction budgets and contractor quality do not carry the same revenue risk when one enters a neighborhood with limited competing inventory and another relies on comparable prices amid accumulated unsold units and weakening transactions. Their financial ratios can look alike while the probability of converting homes into cash differs materially, forcing a national underwriting framework to interpret the numbers through the market in which repayment will actually occur.

The HUG study and Busan’s quarterly sales statistic describe different populations and cannot be compared numerically as though they measured the same phenomenon. The 94-project research covered sites already approved for PF guarantees and employed its own definition of initial sales performance, whereas Busan’s 17.2 percent comes from a separate housing-sale-guarantee population observed under a quarterly methodology. HUG’s study nevertheless associates local inventory and market direction with weaker outcomes among guaranteed projects, while the Busan series shows that some newly marketed apartments have entered a sharply slower sales environment. The two datasets support the same underwriting concern without supplying a direct numerical comparison.

Several modest forecasting errors can become consequential when they move in the same direction. A slightly inflated selling price can lift projected revenue across hundreds of units, a presale schedule that proves several months too fast postpones receipts and extends the period over which interest accumulates, and a construction payment assumed to occur later can improve an intermediate coverage measure even though the bill eventually falls due. Forecasting is unavoidable in development finance, but the credibility of the resulting model depends on how closely those assumptions remain tethered to observable market conditions and contractual reality.

Weak demand can expose gaps between a model and the market more quickly than strong demand because buyers are less likely to arrive fast enough to erase an optimistic forecast. The same regional weakness can cause lenders to withdraw too broadly, denying credit to developments whose individual economics remain intact. Busan therefore tests both functions of the guarantee system: its ability to reject projects whose revenue assumptions are unsupported and its ability to preserve financing for sound projects that private lenders have abandoned because of conditions surrounding them.

Measuring what public credit changed

The government can count guarantees almost immediately, whereas the economic result takes much longer to emerge. A development already capable of obtaining ordinary commercial financing may receive public backing and proceed exactly as it would have otherwise. The homes can be completed, sold and fully repaid while generating little additional construction attributable to the guarantee. Public support can also push a weak project into construction even though the underlying market cannot absorb the finished units, producing apparent additional supply before poor sales, refinancing pressure or losses reveal that the original problem lay in demand rather than credit.

Public backing adds the most housing activity when enough demand exists to support construction and repayment but private lenders have become unwilling to finance the site because of broader market risk, temporary liquidity pressure or tighter credit conditions. In such a case, the guarantee changes the financing decision without changing the commercial logic of the development itself, allowing a viable project to proceed without using state credit to override a market signal that the units are unlikely to sell.

Judging the Aug. 13 measures against that standard requires evidence that unfolds over time rather than a single performance indicator. The first question is whether the guarantee materially altered access to finance. Construction then shows whether credit reached the site, sales reveal whether households appeared at prices and speeds broadly consistent with the original forecast, and completion answers whether the physical development survived the building phase. Eventual repayment provides a different test, showing whether the financing structure endured without transferring an avoidable commercial loss onto the guarantor.

An individual Busan refinancing case illustrates why those stages have to remain separate. Doosan We’ve the Zenith Central Yangjeong received 80 billion won through a bank-and-insurer PF syndicated loan after financial institutions concluded that the development faced a temporary liquidity problem and retained a credible path to completion. The site was completed in April 2025. Publicly available information reviewed during this reporting has not established the final repayment status or ultimate recovery on the syndicated financing, so the case demonstrates completion after refinancing rather than a fully documented successful rescue.

Approval, issuance, outstanding balance, groundbreaking, completion, repayment and guarantee loss describe separate stages and answer separate questions about policy performance. Combining them can make a program appear reassuring or alarming without revealing what happened economically. The same caution applies to the national guarantee book. A low level of claims in 2026 says something about losses realized so far, but it cannot validate every project approved while exposure was rising rapidly because many of those sites remain under construction, in sales or inside financing structures whose ultimate outcome will emerge later.

Korea already publishes substantial information about PF exposure, guarantees, construction and housing conditions, yet an outside observer cannot easily follow an individual publicly supported site from underwriting through financing, construction, sales and final repayment. Commercial confidentiality legitimately limits access to some of the documents needed to reconstruct a credit decision. The resulting information gap still matters because the state can measure how much support it supplied more precisely than the public can measure what that support changed, making it easier to judge the program by the volume of credit deployed than by the quality of the housing activity produced.

By 2027, the most visible scorecard will be straightforward: HUG and HF will either come close to the planned 33 trillion won in guarantee supply or they will not. A more consequential record will accumulate more slowly across sites where forecasts about future buyers turn into observable sales, refinancing needs and repayments. Construction starts will show whether credit reached the physical economy, buyer behavior will test assumptions about demand, and debt performance will reveal whether the financial model survived contact with the market.

Busan’s recent sales environment is likely to expose those outcomes more readily than markets where strong demand can absorb forecasting errors. The data do not establish that residential development in the city should stop, nor do they imply that every publicly supported project carries excessive risk. They do require more convincing evidence from forecasts that depend on buyers arriving quickly and at prices sufficient to carry a development through repayment.

State-backed credit can preserve a sound development when financing disappears, but a guarantee cannot create the household demand required to support a project whose market has been misread. As Korea expands the reach of public housing finance, the program’s quality will be determined less by the size of its headline commitment than by the evidence used before that commitment becomes concrete and steel: evidence capable of showing which financing gaps deserve to be bridged and which reveal that the underlying market cannot support the project.