Mizrahi Tefahot: Is the Housing and Real-Estate Book as Strong as It Looks?

The main article argued that Mizrahi Tefahot ended 2025 with very strong profit, but also with a CET1 cushion that does not leave much room for mistakes. This continuation isolates the issue that ties those two things together: whether the housing and real-estate book truly came out of the last two years stronger, or whether net profit simply benefited from an unusually benign credit-cost year.

That question matters now because 2025 was not a clean year of housing-market euphoria. Apartment sales in Israel fell 11.9% to roughly 90.7 thousand units, home prices rose only 0.4%, and yet mortgage originations to the public still increased 13.4% to NIS 105.7 billion. In other words, the improvement in Mizrahi Tefahot's credit quality did not depend on an overheated market or on aggressive home-price inflation. That supports the view that the retail mortgage book genuinely looks better.

But that is only one layer of the story. In the same year, group provisions for housing credit swung from an expense of NIS 64 million in 2024 to an income contribution of NIS 119 million in 2025, while total public credit risk in Israel's construction and real-estate sector rose to NIS 96.3 billion from NIS 76.5 billion. The right reading is therefore not that "the risk is gone," but something more precise: households look cleaner, while exposure to the system around them, developers, project finance, guarantees and commitments, has grown.

The point an initial read can miss is that the bank itself holds both statements at once. On one hand, mortgage-book risk indicators are described as stable, the risk level remains low to moderate, and most customers that deferred payments have already resumed paying. On the other hand, in the very same discussion the bank highlights close monitoring of non-linear purchase contracts and cancellations in residential projects. So the 2026 question is not whether a problem has already broken out. It is whether today's clean picture is truly durable.

What Actually Improved In Housing

The good news in the housing book is not cosmetic. Average LTV fell to 54.7% from 55.1%, risk metrics were defined as stable, and the mortgage-book risk level remained unchanged at low to moderate. This is not one attractive slide number. It is a cluster of indicators pointing to underwriting discipline that held up even after the difficult conditions of 2024.

The strongest signal is the move in provisions. When housing credit swings from a group expense of NIS 64 million to a group income contribution of NIS 119 million, that is more than a qualitative statement that the book is "good." It is evidence that the bank sees lower expected loss in the mortgage portfolio than it did before. The investor presentation reinforces the same message from a wider angle: total credit-loss expense fell to 0.06% of total loans in 2025, down from 0.14% in 2024.

Still, that number should not be read as proof that risk disappeared. In the financial report, the bank says the decline in credit-loss expense also reflected a lower group provision related to the war, and in the same breath notes that the balance-sheet allowance remains sizable. Part of the improvement, then, is the release of caution built in earlier periods, not proof that Mizrahi Tefahot has entered an era in which the housing book is almost immune to surprises.

Customer behavior supports that interpretation. Deferred payments in housing fell to only NIS 11 million, out of a modified balance of roughly NIS 238 million. A year earlier, deferred payments stood at NIS 115 million out of roughly NIS 5.2 billion of modified credit. That gap is too large to dismiss as noise. It says that, at the household cash-flow level, 2025 looks much less like a bridge year and much more like a return to normal.

This is the strongest bullish case on the housing book: the housing market was not perfect in 2025, yet Mizrahi Tefahot's retail borrower behaved better, with less stress, fewer payment deferrals and a slightly lower average LTV. If that were the whole story, it would be enough to conclude that the housing book is strong.

Where The Risk Still Sits

The problem is that Mizrahi Tefahot's housing exposure does not stop at household mortgages. Around it sits a much larger construction and real-estate finance book, with project finance, apartment-buyer guarantees, commitments and other off-balance-sheet lines. By year-end 2025, public credit risk in Israel's construction and real-estate sector stood at NIS 96.3 billion, up from NIS 76.5 billion a year earlier, a 25.8% increase. Within that, project-finance exposure rose to NIS 62.3 billion from NIS 48.3 billion, up 28.9%.

That changes the frame. An investor who sees a negative housing-credit provision may conclude that the entire housing story has turned calm. That is too strong. The reassuring data point belongs mainly to the retail borrower. The bigger risk book, the one tied to apartment sales pace, project financing costs and execution, actually expanded.

It is also important to state what supports that book. The bank does most of this business through closed project finance, relies on external construction supervisors, typically holds full real-estate collateral, and for parts of the exposure not secured by real estate it holds other collateral such as deposits and securities. In addition, 59.2% of on-balance-sheet credit risk and 71.4% of off-balance-sheet credit risk in the sector is designated for closed project finance, overwhelmingly in residential construction in demand areas. That is exactly why the book has looked so resilient so far.

But the same structure that reduces risk also explains why the bank is not relaxing. It says explicitly that, during underwriting and ongoing monitoring, it examines the share of apartment-sale contracts using non-linear payment methods, including their impact on project financing costs. In the same context, it says it is closely monitoring cancellations and has not identified risk realization in those areas so far. That wording matters. It does not say the risk is gone. It says there are specific risk nodes under close watch that have not yet turned into loss.

This is the real 2025 paradox. On one hand, credit quality in construction and real estate actually improved: total problematic credit fell to NIS 333 million from NIS 594 million even as total exposure increased. Within project finance itself, problematic credit fell to NIS 146 million from NIS 359 million. On the other hand, a book approaching NIS 100 billion does not need a dramatic deterioration to matter again for the provision line. A sustained change in sales pace, financing cost or contract quality would be enough to make 2025 look, in hindsight, like a peak-clean year rather than a new baseline.

Why The Positive Provision Does Not End The Debate

To judge whether 2025 marks a structurally lower risk level or simply a very benign year, the bank's housing and real-estate exposure has to be split into two books. The retail mortgage book is diversified, carries a relatively low average LTV, and shows clear evidence that payment-deferral stress has largely disappeared. The construction and real-estate finance book is inherently more concentrated, with heavier dependence on sales velocity, contract structure and budget discipline at the project level.

Put differently, the retail book is already giving the bank breathing room. The developer book still demands discipline. That is why the move from expense to income in housing credit is a very positive data point, but it should not, on its own, settle the question around the quality of Mizrahi Tefahot's full housing and real-estate exposure.

The bank almost writes this continuation thesis between the lines. It presents a stable mortgage book, stresses that most customers who took a payment break are paying again, and at the same time leaves the active warning signs in place: lingering uncertainty around the long-term effects of the war, ongoing scrutiny of non-linear contracts, and monitoring of cancellations. Put together, those points lead to a simple conclusion: 2025 was a year in which the risk did not materialize, not a year in which the risk ceased to exist.

The market backdrop around the sector supports that balanced reading. Apartment sales fell 11.9% in 2025, yet mortgage originations to the public rose to NIS 105.7 billion, and the bank says further rate cuts and continued stabilization in the security situation could gradually bring demand back to the sector. That is the positive counter-thesis: if demand returns without a deterioration in contract quality, the large project-finance exposure could prove to be a growth engine rather than a risk pocket. But that is still a 2026 test, not a settled 2025 fact.

What Needs To Hold In 2026

The first check is whether the housing book can stay around zero, or even below zero, in credit costs without the bank having to rebuild a heavier allowance. Mild normalization is not a problem. A fast move back to a meaningful expense would be a sign that 2025 was cleaner than the underlying reality.

The second check sits in the more valuable book: project finance. As long as non-linear sales contracts remain a monitoring point rather than a source of pressure on project financing costs, and as long as cancellations remain manageable, Mizrahi Tefahot can keep carrying a large real-estate book without paying for it through the provision line.

The third check is growth versus quality. In 2025 the bank got both rapid sector exposure growth and lower problematic credit. That is close to an ideal outcome. If exposure keeps rising in 2026 but the quality ratios stop improving, the market will start asking whether the 2025 improvement was simply too comfortable to extrapolate.

The fourth check links back to the main article. The cleaner this book stays, the more the bank can keep growing without absorbing capital through higher provisions. If disappointment shows up here, the damage will not stop at earnings. It will also flow into the question of how much real room remains against the CET1 constraint.

Conclusion

The fair reading of 2025 is that Mizrahi Tefahot's housing book does look stronger than it did a year earlier. Average LTV is lower, deferred payments have almost disappeared, and the housing segment moved from group provision expense to group provision income. Those are not accounting decorations. They reflect a real improvement in retail borrower quality.

But the more important test sits one layer below the surface. Construction and real-estate exposure grew quickly, the project-finance book expanded, and the bank itself is still not giving up close monitoring of non-linear contracts and cancellations. The precise thesis, then, is not that the housing and real-estate book is simply "as strong as it looks." It is that the household side looks stronger, while the project-finance side still has to prove its durability.

If 2026 passes without a renewed jump in credit costs and without fresh stress signs in residential projects, 2025 will look in hindsight like a year of genuine credit-quality improvement. If the risk markers the bank is highlighting now begin to move, the income contribution in housing credit will look like evidence of temporary cleanliness rather than immunity.