Bank Jerusalem's Credit-Sale Machine: Real Capital Release Or Convenient Earnings Support
This follow-up shows that Bank Jerusalem has already built a real sale, securitization, and syndication engine: it sold NIS 1.07 billion of credit in 2025, then added another NIS 680 million deal after the balance sheet. But as long as disclosure does not cleanly separate sale gains, servicing economics, and the effect on core banking margins, part of the profit improvement still looks too convenient.
Why This Follow-Up Exists
The main Export article argued that Bank Jerusalem did not improve 2025 only through spread income. Part of the story was fees, portfolio sales, and capital release. This follow-up isolates that thread and asks a narrower question: has the bank built a genuine lighter-capital growth platform, or a convenient way to smooth profitability through transactions that look very good on the closing date.
The short answer is that it is both, but not in equal proportion. The capital release is real. The bank's strategy explicitly includes credit sales, securitization, risk-transfer tools, and bringing partners into selected activities as part of capital-efficiency improvement. In 2025 it sold NIS 1.0698 billion of public credit, originated NIS 790.6 million of housing loans through syndication, and ended the year managing sold or jointly originated credit portfolios totaling NIS 4.4148 billion. That is no longer tactical.
But there is another side. The accounting profit is cleaner than the economic picture. In 2025 total pretax profit from credit sales was NIS 31.3 million. Immediately after the balance-sheet date, the bank signed another roughly NIS 680 million transaction with expected pretax profit of about NIS 38 million in Q1 2026. That means one post-balance-sheet deal is expected to generate pretax profit about 21% higher than all of 2025's credit-sale profit combined. This is no longer only a capital-release story. It is also an earnings-quality story.
So the right question is not whether portfolio sales are "good" or "bad." It is whether the market is separating three different things that now sit under one umbrella: one-off gain at the moment of sale, capital release that allows continued growth, and recurring income from servicing and operating credit that is no longer fully on the balance sheet.
What The Bank Has Actually Built
The story begins in strategy rather than in the P&L. The multi-year plan for 2025 through 2027 lays out four strategic paths, one of which is improving the financial infrastructure and the efficiency of capital usage. Inside that section, the bank explicitly lists building infrastructure for credit sales and securitization, developing risk-transfer tools, and bringing partners into certain activities. The 2026 targets repeat the same logic: sales, securitization, credit purchases, CRT, and activity carve-outs are not presented as cosmetic add-ons, but as part of the operating model.
That matters because it means the reader should not treat portfolio sales as a one-off disposal of legacy assets. The bank is building a permanent pipe. In 2025 it sold commercial and housing credit totaling NIS 1.0698 billion with NIS 31.3 million of pretax profit. In parallel, the syndication agreement signed in June 2024 with an institutional group, initially sized at up to NIS 1.8 billion over three years and later expanded to NIS 2.1 billion, generated NIS 790.6 million of housing-loan originations during 2025. The outstanding balance under that framework stood at NIS 956.0 million at year-end.
This is already a multi-lane machine:
| Layer in the model | What the local evidence shows | Why it matters |
|---|---|---|
| Upfront sale gain | In 2025 the bank sold NIS 1.0698 billion of credit with NIS 31.3 million of pretax profit | This is the part most visible in the filings, and therefore the easiest to overread |
| Ongoing servicing and operation | The bank continues to manage and operate sold or jointly originated portfolios, and in syndication also earns origination fees | That means the transaction does not end the economic relationship with the credit |
| Capital release and risk transfer | On the sold portion the bank bears no credit risk, and it remains inside regulatory limits | This is the layer that truly allows faster growth with less capital consumption |
The practical implication is that Bank Jerusalem is moving away from a pure originate-and-hold model toward one in which it originates, distributes, and keeps the servicing layer, the operating role, and part of the economics. That is closer to a credit platform than to a bank that fully funds every loan to maturity on its own balance sheet.
This chart tells the story better than the headline does. Even before asking what the profit quality looks like, the bank is already running meaningful credit volume outside the classic model of keeping the entire risk stack on balance sheet. That is structural change, not decoration.
Where The Capital Release Is Real, Not Just Cosmetic
This follow-up matters because the model really does change the balance sheet. At end-2025, sold housing-loan portfolios as a share of the housing-loan book, including the managed sold book and excluding commercial credit and joint origination, stood at about 22% against a 25% limit. Including syndication, the share stood at about 29% against a 35% limit. In other words, these deals are not just a bonus. They are already part of the way the bank manages its capital room.
That connects directly to 2026. The bank ended 2025 with a CET1 ratio of 10.8%, and in its 2026 targets it set a goal of not less than 10.25% from April 1, 2026, plus safety margins. In plain terms, capital release is no longer optional. If the bank wants to keep growing, widen activity, maintain payout policy, and continue investing in digital infrastructure, it needs a tool that generates credit without consuming the same amount of capital as the old model.
The February 2026 transaction makes the mechanism especially clear. The bank sold 90% of a roughly NIS 680 million real-estate-collateralized loan portfolio, retained only 10%, and kept that retained slice on a pari-passu basis with the buyer. At the same time, it signed a servicing and operating agreement, so it continues managing the sold portfolio for a fee. Put differently, the bank removes 90% of the exposure, does not leave the servicing layer, and does not give up the operating role around the portfolio.
Syndication works in the same spirit. In the housing loans originated under the institutional agreement, the bank funded only 10% of each loan, the institutional side funded 90%, and the bank continues managing the portfolio while also receiving origination fees. This is not just a case of selling an existing asset to create room. It is a mechanism that allows the bank to create more lending volume from the outset on the same capital base.
The point is not that the bank has endless room. The opposite is true. There is remaining room, but not carefree excess. That is why these transactions matter. They are a genuine capital valve. Without them, faster growth would require more capital, or much tighter restraint on loan-book expansion, or more conservative dividends.
Where Earnings Quality Gets Less Clean
This is where the story becomes less comfortable. Looking at pretax profit as a percentage of the credit sold shows a very wide spread between transactions:
| Channel | Credit volume | Pretax profit | Profit as % of volume |
|---|---|---|---|
| Commercial-credit sales in 2025 | NIS 599.5 million | NIS 23.9 million | 4.0% |
| Housing-loan sales in 2025 | NIS 470.3 million | NIS 7.4 million | 1.6% |
| Total credit sales in 2025 | NIS 1,069.8 million | NIS 31.3 million | 2.9% |
| February 2026 transaction | NIS 680.0 million | NIS 38.0 million | 5.6% |
This is the heart of the continuation thesis. The February 2026 deal is not just one more transaction. It is a transaction whose expected profit is larger than all of 2025's sale profit combined, and whose implied gain rate is much richer than the 2025 average. That does not prove there is a problem. It does mean the next report has to separate transaction profit from recurring banking profit. Otherwise the capital-release engine also becomes a quarter-smoothing engine.
There is another under-discussed side to asset quality. The risk-characteristics table for sold housing loans shows that the sold pools are tilted toward lower LTV than the housing portfolio left on the balance sheet. 51.5% of the sold book sits at LTV up to 45%, versus only 32.0% in the on-balance book. Only 0.4% of the sold book sits above 75%, versus 3.7% in the on-balance book.
The bank itself says that portfolio sales increase the average LTV of what remains on balance sheet, but that the average remaining LTV is still relatively low, and that the sales did not have a materially negative effect on net interest income. Both statements can be true at the same time. Even so, they create exactly the fog that needs to be illuminated. When the bank exports relatively conservative mortgage pools, leaves behind a somewhat higher average-LTV book, records a sale gain, and keeps collecting servicing fees, reported profit becomes less purely "core banking" and more a function of balance-sheet architecture.
The problem is not the transaction itself. On the contrary, if the move improves return on capital without materially weakening the risk profile, this is exactly what good banking management should do. The problem is that current disclosure does not let the reader quantify the split between the three engines: how much of the improvement comes from sale gains, how much from servicing and origination fees, and how much from genuinely better core banking economics after the cleaner credit has already moved out.
What Is Missing In Disclosure, And Therefore What Has To Appear Next
Precisely because this machine now looks real, the current disclosure standard is no longer enough. The bank does provide a useful economic skeleton: sale volumes, gains, regulatory limits, the fact that it continues servicing the portfolios, and the size of the managed books. But it does not provide the two answers the market now truly needs.
The first question: how much recurring income remains after the sale. The bank says there are servicing fees, and in syndication also origination fees, but it gives no number that allows a run-rate view. Without that, it is impossible to know whether the model is closer to a growing credit platform with an annuity-like service layer, or closer to upfront profit that pulls forward future economics.
The second question: what the margin-replacement cost is. The bank says the sales did not have a materially negative impact on net interest income, but it does not show how much interest income left the balance sheet, how much of that was replaced by fees and servicing income, and what happened to the yield on assets that remained. A reader trying to infer normal banking profitability from the consolidated numbers simply does not get enough layers here.
The third question: how much regulatory room is left after the February 2026 transaction. End-2025 still showed breathing room, 22% against 25% on sold housing and 29% against 35% including syndication. But the February 2026 deal already belongs to the world after the balance sheet, and the current filing does not yet show where the meter sits after that step. That will be one of the most important 2026 disclosures, because it will say how much runway the bank still has to keep using the machine at the same intensity.
In other words, the investor does not need another story about Bank Jerusalem being more "innovative." The investor needs three clean numbers: sale gains versus recurring post-sale income, the effect of sales on core spread economics, and the remaining regulatory headroom after the latest transaction. Without those three, the capital relief is visible, but the earnings quality remains unresolved.
Conclusions
Current thesis: Bank Jerusalem has already built a sale, securitization, and syndication machine that is real at the capital level, but the filing still does not separate real capital efficiency from profitability that looks too convenient at the headline line.
What works here is clear. The bank frees capital, keeps the servicing and operating layers, operates inside regulatory limits, and creates a path to grow without loading the full credit exposure onto its own balance sheet. That is not cosmetic. It is a model change.
What weighs on the story is equally clear. The post-balance-sheet deal already produces more profit than all of 2025's credit-sale gains combined, and the sold housing pools look cleaner on LTV than the housing book left on balance sheet. So anyone reading the coming quarters as if this were ordinary banking earnings may end up capitalizing the engine too generously.
The strongest counter-thesis is that the market may be overly suspicious. If the bank is genuinely transferring risk, keeping servicing income, and continuing to grow without materially damaging spreads or credit quality, this is exactly what healthy capital efficiency looks like. That is a legitimate argument. The problem is that the current disclosure still does not break the economics out clearly enough to settle the issue.
So the 2026 test is not whether another deal happens. Another deal is likely. The real test is whether the bank starts showing what the sale profit is, what recurring income remains after the sale, and what happened to the banking core after capital was already released. Only then will it be possible to decide whether this is truly a lighter-capital growth engine, or a too-convenient earnings bridge that gets ahead of the underlying economics.
Disclosure: Deep TASE analyses are general informational, research, and commentary content only. They do not constitute investment advice, investment marketing, a recommendation, or an offer to buy, sell, or hold any security, and are not tailored to any reader's personal circumstances.
The author, site owner, or related parties may hold, buy, sell, or otherwise trade securities or financial instruments related to the companies discussed, before or after publication, without prior notice and without any obligation to update the analysis. Publication of an analysis should not be read as a statement that any position does or does not exist.
The analysis may contain errors, omissions, or information that changes after publication. Readers should review official filings and primary sources before making decisions.