Shoham Bizness: What Is Really Happening in the Part of the Book Already Past 180 Days
The most fragile layer in Shoham Bizness's book is not delinquency by itself, but the gap between balances already past 180 days and the pace at which they are classified as impaired and covered through accounting recognition. Collateral explains part of that gap, but it pushes the real test into collections and realization.
What This Follow-Up Is Isolating
The main article argued that Shoham Bizness's real test had shifted from book growth to credit quality. This follow-up isolates the most sensitive layer inside that argument: what happens once part of the book has already moved past 180 days, and how the company decides when deep delinquency becomes an impaired balance and when it remains, at least for now, a collectible claim.
The short answer is that loss recognition is happening, but not at the same pace as the deterioration in delinquency. In 2025 the company wrote off NIS 16.0 million of bad debts, added NIS 19.0 million of specific provisions, and ended the year with a total allowance of NIS 26.7 million. Even so, balances already more than 180 days past due rose to NIS 99.5 million, and roughly NIS 21.5 million of that layer was still not classified as impaired.
That is the core issue. Anyone looking only at the short duration profile and the wide drawer dispersion could conclude that the book remains under control. The deep-delinquency layer tells a different story: the problem is not dispersion or the original contractual duration of the book, but what happens after the debt stops paying on time.
| Key collection metric | 2024 | 2025 | Why it matters |
|---|---|---|---|
| Balances past 180 days | NIS 69.9 million | NIS 99.5 million | The troubled tail grew faster than the book |
| Of that, past 180 days but not classified as impaired | NIS 9.0 million | NIS 21.5 million | This is where management judgment matters most |
| Specific allowance against balances past 180 days | NIS 12.3 million | NIS 15.0 million | Recognition rose, but not at the same pace as delinquency |
| Debt arrangements | NIS 54.2 million | NIS 36.2 million | The pressure shifted away from formal workouts and toward deep arrears |
Where the Book Actually Worsened
The first finding, and it is not intuitive, is that the deterioration did not come from the short end of delinquency. Balances overdue by up to 180 days actually fell to NIS 17.3 million from NIS 21.6 million. At the same time, balances already past 180 days jumped to NIS 99.5 million from NIS 69.9 million. This is not a picture of mild slippage across the whole book. It is a picture of weight accumulating deeper in the tail.
Debt arrangements also fell to NIS 36.2 million from NIS 54.2 million. On the surface that looks positive. In practice it also means that by year-end 2025 the stressed layer was sitting less in intermediate restructuring channels and more in the harder zone of very deep delinquency.
That chart matters because it breaks a superficial read. If the issue were simply broader slippage, the short delinquency buckets should have swollen first. That is not what happened. Short delinquency shrank, while the part already past 180 days expanded sharply. In collection-quality terms, that is a less comfortable picture because it means the burden moved from the stage where a file may still be repaired quickly to the stage where repayment depends much more on negotiation, collateral, legal action, and realization speed.
In relative terms, balances above 180 days represented about 9.5% of the gross book at year-end 2025, up from about 7.4% a year earlier. Total overdue balances reached NIS 116.8 million, or roughly 11.2% of the gross book. This is no longer a marginal noise layer.
The Line Between Deep Delinquency and Impairment Is Still Flexible
The second and more important point is that the line between a deeply delinquent balance and an impaired balance is not automatic. Impaired balances totaled NIS 89.4 million, but balances already past 180 days totaled NIS 99.5 million. That means year-end 2025 still contained a layer of more than NIS 10 million that was already deeply overdue but did not map one-for-one into the accounting definition of impairment.
The note says that explicitly: of the balances past 180 days, about NIS 21.5 million were not classified as impaired because, in the company's view, they remained collectible. The explanation relies on ongoing offsets, debt arrangements that are being paid as agreed, debtor cooperation, and legal collectability. Those are legitimate collection arguments. They also mean that the gap between deep delinquency and impairment rests materially on management judgment and recovery assumptions, not on a hard mechanical trigger.
The structure of allowance rates sharpens that point even further. In the bucket with no significant increase in risk, the total allowance rate is 0.72%. In the bucket with a significant increase in credit risk, it jumps to 17.28%. In the impaired bucket it stands at 17.91%. In other words, the main accounting cliff appears when an exposure leaves the clean book. Beyond that, the difference between high-risk and impaired classification is already quite narrow.
| Risk bucket | Gross balance in 2025 | Allowance in 2025 | Allowance rate |
|---|---|---|---|
| No significant increase in risk | NIS 931.8 million | NIS 6.7 million | 0.72% |
| Significant increase in risk | NIS 23.1 million | NIS 4.0 million | 17.28% |
| Impaired balances | NIS 89.4 million | NIS 16.0 million | 17.91% |
That table matters because it shows that the discussion is not only whether the problem exists, but at what point the company chooses to harden classification, and how much of the burden still sits inside a gray zone of possible recovery.
How Loss Recognition Is Actually Happening
This needs to be read fairly: this is not a story of no recognition. The company did clean the book in 2025. It wrote off NIS 16.0 million of bad debts, added NIS 1.2 million to the general allowance, added NIS 19.0 million of specific provisions, and reversed NIS 0.9 million for doubtful debts that were collected. The board report also says credit-loss expense rose to NIS 18.3 million mainly because of higher specific provisions on impaired balances and the recognition of a time-value provision.
But that recognition is still not running at the same pace as deep delinquency. The specific allowance against balances past 180 days was NIS 15.0 million, up from NIS 12.3 million in 2024. That is about a 21% increase. Over the same period, the balances past 180 days themselves rose by about 42%. Put differently, the deep-delinquency layer expanded faster than the specific accounting cover applied to it.
The practical meaning of that chart is not necessarily that the allowance is inadequate. It does mean that the rise in delinquency has not been translated one-for-one into accounting cover. For now the gap is being closed through three things: write-offs already taken, collateral that reduces expected loss, and management's view that part of the balance should still be recovered. That is a legitimate model. It is also a model that now needs proof in the next reports.
Another sign is the coverage rate on the impaired bucket as a whole. Specific allowance represented about 17.6% of impaired balances at year-end 2025, down from about 21.7% a year earlier. Again, that is not automatic proof of trouble. It is a clear indication that the company is leaning more heavily on the collections and collateral story than on faster recognition of loss.
Why Duration and Dispersion Do Not Solve the Issue
This is where the company's own framing matters. In the board report and in the March 2026 presentation, the message is short duration and wide drawer dispersion. In the board-report duration section, 75% of the book is due within 120 days, and in the presentation the message is slightly sharper, around 77% within 120 days. Concentration also looks manageable: the presentation shows the top ten drawers at about 21.2% of the book in aggregate, with the single largest drawer at 3.59%.
That framing is not wrong, but it is incomplete. The reason sits in the footnote to the board-report duration table: that table is presented excluding overdue balances of NIS 116.8 million. In other words, the short duration profile describes the scheduled and still-performing part of the book. It does not describe the stressed collections layer that has already moved off schedule.
| Reassuring framing at first glance | What it really says | What it does not solve |
|---|---|---|
| 75% to 77% of the book comes back within 120 days | The functioning part of the book is still relatively short | The delinquent layer that already left the schedule is excluded from the duration view |
| The top ten drawers are about 21.2% of the book | There is no extreme single-drawer concentration | Dispersion does not prevent a collections problem once balances are already stuck |
| 56% of the 181-plus slice in the presentation is collateral-backed | Collateral has a real role in the risk framing | Collateral buys time, but it does not remove the realization and collections test |
That distinction matters because it separates underwriting quality on the fresh book from workout quality on the troubled book. A lender can absolutely run a short, diversified, better-collateralized portfolio and still carry a deep layer of balances that take much longer to resolve.
Collateral Is a Cushion, Not a Shortcut
Collateral is the main reason the picture is not worse. The company says that without the collateral posted against impaired customer balances, it would have recognized an additional NIS 12.9 million credit loss. That is a very material sum, close to half the year-end total allowance.
That also explains why the allowance rate on collateral-backed credit is lower. In collateral-backed credit, mainly real estate and engineering equipment, gross carrying value stood at NIS 497.5 million with an allowance of only NIS 2.6 million, an expected-loss rate of 0.51%. In the direct post-dated check-discounting channel, gross carrying value stood at NIS 481.5 million with an allowance of NIS 8.4 million, an expected-loss rate of 1.74%.
But this is exactly where the easy conclusion needs to be resisted. Collateral does not erase risk, it changes its form. Instead of asking only whether the debtor will pay on time, the real questions become what the collateral is worth, how long realization takes, how much legal and operational friction sits in the path, and what remains after that process. So collateral explains part of the gap between deep delinquency and impairment, but it also pushes the proof burden into the realization and collections stage.
Bottom Line
The main article asked whether Shoham Bizness's book growth was still clean. This follow-up gives a sharper answer: inside the layer already past 180 days, loss recognition is rising, but not at the same pace as delinquency itself.
That can still end in two very different ways. If the roughly NIS 21.5 million already past 180 days but still not classified as impaired is collected, repaid, or resolved through collateral and arrangements, 2025 may look in hindsight like a cleanup year in which the company remained conservative enough. But if those balances keep rolling forward, the company will face the same question again with less room: whether to take more provisions, more write-offs, or to admit that part of the collateral story merely bought time.
So the metric to watch now is not only book growth or book size. The critical metric is the exit rate from the 180-plus tail, through real collections, collateral realization, or harder classification. That is where it will become clear whether 2025 was a controlled cleanup year, or a year in which loss recognition simply stayed behind the deterioration.
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.