SR Accord: What Really Sits Behind the Overdue Balances and the Collateral
SR Accord's overdue balances did not disappear in 2025. A meaningful share simply stayed outside impaired credit because management expects collection through collateral realization, legal process and debtor cooperation, so the real debate is not whether collateral exists but how quickly and at what haircut it can actually turn into cash.
The main article already argued that growth returned, but the real test moved to funding and underwriting. This follow-up isolates the point that is easiest to miss in a quick read of the credit-risk disclosure: the problematic balances did not really disappear. They moved between accounting buckets, based on management judgment about collection, collateral and realization timing.
That is the core debate. At the end of 2025, credit with a significant increase in credit risk together with impaired credit reached NIS 140.2 million, almost 7.7% of the credit book, almost exactly where it stood in 2024. But the mix flipped. Credit with a significant increase in risk jumped to NIS 88.8 million from NIS 8.1 million, while impaired credit fell to NIS 51.3 million from NIS 104.2 million. So the real question is not whether risk vanished. It is why so much of it stayed outside the impaired bucket.
The Risk Layer Did Not Disappear, It Was Shifted Out Of The Impaired Bucket
| Risk bucket | 2024 | 2025 | What really changed |
|---|---|---|---|
| Credit with significant increase in credit risk | NIS 8.1m | NIS 88.8m | More than an 11x jump |
| Impaired credit due to credit risk | NIS 104.2m | NIS 51.3m | Down about 51% |
| Both buckets together | NIS 112.3m | NIS 140.2m | Up about 24.9% |
| Share of both buckets in total book | 7.67% | 7.71% | Almost unchanged |
The risk layer did not shrink. It moved from the harshest bucket into the intermediate one, the bucket that still allows a recovery story. In 2024, impaired credit was almost the entire problematic layer. In 2025, 63% of that layer sat instead in the significant-increase-in-risk line. That is not technical noise. It is a management call.
The single most important number here is about NIS 28.9 million of balances overdue by more than 180 days that the company kept inside the significant-increase-in-risk bucket rather than classifying as impaired. Its explanation is explicit: based on information available to management, it expects to collect those balances. The support for that view is judgments and debt rulings already obtained in its favor, collateral realization processes, and debtor cooperation.
That is exactly why the debate is not about whether there is a problem, but about how severe the problem is. If there were no issue, these balances would not sit in the heightened-risk bucket at all. On the other hand, if management believed collection was already broken, they would sit in impaired credit. In practice, the company is asking readers to accept a middle reading: this is hard debt, deep in arrears, but still reasonably collectible in its view.
The provision rates support that reading. The provision rate on credit with a significant increase in risk jumped to 10.0% from 1.8% in 2024, so the company is not presenting that bucket as clean. Even so, the provision rate on impaired credit stood at 19.5%, only modestly below 21.0% in 2024. The line between the two buckets is therefore not whether risk exists. It is how much of the balance is still seen as recoverable, and over what time frame.
Collateral Is Part Of The Answer, But It Does Not End The Debate
Management's case should not be dismissed as cosmetic. The protection layer it presents is material. The presentation points to more than NIS 1.3 billion of tangible collateral, in addition to personal guarantees and post-dated checks. It also stresses that the solo-credit activity relies predominantly on first-ranking liens over tangible assets such as real estate, vehicles, engineering equipment and control stakes in public companies.
| What the presentation shows | 2025 | Why it matters |
|---|---|---|
| Total tangible collateral | More than NIS 1.3bn | This is the core support behind the recovery argument |
| Largest customer | 3.8% of the credit book | Single-name customer concentration is not extreme |
| Largest drawer | 2.3% of the credit book | Concentration is also limited on the payer side |
| Top 10 customers | 26.7% of the book | Dispersion helps management argue risk is not overly concentrated |
| Top 10 drawers | 23.0% of the book | The payment layer is also relatively spread out |
But this is exactly where the headline "there is collateral" becomes too comfortable. The company itself makes clear that expected credit losses in the problematic buckets are measured individually, net of the estimated realization value of tangible collateral and assigned receivables, and with an estimate not only of the amount to be collected but also of the time needed for full collection. In other words, collateral is not cash sitting in the bank. It is an estimated realization value, after time, legal process and friction.
Collateral is a recovery thesis, not a waiver of risk. That is why the NIS 28.9 million kept outside impaired credit still did not receive ordinary-credit treatment, but a 10% provision rate. What the company is really telling the market is not that this debt is clean, but that the ultimate loss should be lower than the raw days-past-due figure would suggest.
There is another layer that both strengthens and complicates that story. The presentation shows that 56% of the book sits in infrastructure, construction and real estate, and within that sector cluster roughly 75% of the credit is backed by tangible collateral. That is a meaningful protection layer, but one that also depends on asset quality, lien ranking, realization speed and the ability to sell or enforce the collateral on reasonable terms. So NIS 1.3 billion of collateral cannot be read as if it were NIS 1.3 billion of cash.
The same logic appears in the split by credit type. Credit backed by tangible collateral carried a 0.7% provision rate, versus 2.7% in discounting transactions. So the collateral genuinely lowers expected loss severity. But it does not eliminate the debate around deeply overdue balances, because at that stage the question is no longer just whether an asset stands behind the exposure, but what happens when that asset has to be realized for real.
What A Quick Read Misses In The Overdues
The aging table adds two more yellow flags. The first is that deeply overdue balances did not actually fall by much in absolute terms. The net balance in the over-180-days bucket stood at NIS 79.5 million, versus NIS 82.3 million in 2024. That is too small an improvement to claim the issue is resolved, especially when the total book grew by about 24%.
The second is that the "not overdue" line is not entirely clean either. It includes about NIS 23.6 million of customers that are not overdue only because they are under debt arrangements or agreed payment deferrals. In 2024, the comparable number was about NIS 24.1 million. So part of what looks current at first glance is current only after restructuring.
Here too the headline can mislead. The weighted default rate in the over-180-days bucket fell to 12.74% from 21.0% in 2024, and the provision on that bucket fell to NIS 11.6 million from NIS 21.9 million. A quick reader can take that as a sharp improvement in book quality. The more forensic reading is more cautious: part of that improvement comes from moving risk out of impaired credit, not from hard balances disappearing.
The same pattern shows up at total-book level. Total credit-loss allowance was almost unchanged at NIS 35.0 million versus NIS 35.1 million a year earlier, while the credit portfolio grew by more than 24%. That is also where the overall allowance ratio of 1.9% versus 2.4% comes from. It is a better headline number, but it is built on mix, classification and recovery assumptions, not only on a clean improvement in asset quality.
The filing itself adds another reason for caution: from the end of 2024, the general allowance has been measured under a new economic model, and that change in estimate reduced the general allowance by about NIS 3 million relative to the previous model. That does not cancel the improvement seen in 2025. It does mean that the lower allowance ratio cannot by itself prove that credit risk has been cleaned up.
Conclusion
This continuation is not arguing that management is wrong to lean on collateral. It is clarifying what exactly management is asking the market to believe. The claim is not that deep arrears disappeared. The claim is that a larger share of deep arrears is still collectible, and therefore should remain outside the impaired bucket.
That is a case the market can understand, but it should not accept automatically. For it to hold, the next reports need to show not just another collateral story, but actual collections, a real decline in balances overdue by more than 180 days, and further shrinkage in the risk layer without simply moving the same debt around again. Until then, collateral is not the end of the discussion. It is only the framework inside which the real debate is happening.
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.