Max IT Follow-Up: Did Credit Quality Really Improve, or Did the Mix Just Get Safer
Max IT's loss metrics improved in 2025, but the clearest conclusion is still a faster shift toward vehicle-backed lending. As long as the broader problematic-credit indicators do not improve alongside the loss numbers, it is hard to say underwriting itself has already stepped up.
What This Follow-Up Is Really Testing
The main article argued that Max IT’s 2025 credit growth still had to pass a funding and profitability test. This follow-up isolates the other half of the debate: when loss metrics improved in 2025, was that evidence of better underwriting quality, or mainly evidence that the book moved faster into safer products.
The short answer is that the improvement is real, but the strongest proof is for a safer mix, not for a clean step-up in borrower quality. The reason is straightforward. Most of 2025 growth came from vehicle-backed lending, the product with collateral. At the same time, loss metrics improved meaningfully, while the broader problematic-credit indicators did not fully confirm the same story.
That distinction matters because in consumer lending two very different stories can produce the same headline. If underwriting truly improved, the book should eventually show cleaner behavior even after stripping out the mix shift. If instead the portfolio simply became safer because vehicle-backed credit gained weight, loss severity and reserve intensity can fall even without a broad improvement in borrower behavior.
Three points stand out immediately:
- The move toward vehicle-backed credit is not a side detail. Vehicle-backed lending reached 34% of the total credit portfolio at the end of 2025, up from 29% a year earlier and only 22% at the end of 2023.
- The sharpest improvement in loss metrics appeared in direct credit, not in card receivables. That pattern fits a safer mix better than it fits a book-wide underwriting leap.
- The earlier-warning metrics did not fully clear the picture. The private problematic-credit ratio rose to 3.73% from 3.45%, and the ratio of private exposures outside execution rating, meaning borrowers to whom the company no longer extends new credit, rose to 6.97% from 6.46%.
Where The Mix Did The Heavy Lifting
To see why it is hard to attribute the full improvement to better underwriting, the right place to start is the retail product mix. Across the three main private-credit products, solo loans, vehicle-backed credit, and revolving credit, the combined balance rose to NIS 11.372 billion from NIS 9.852 billion. But the growth was far from evenly distributed.
Solo loans rose by only NIS 324 million, to NIS 4.860 billion. Revolving credit rose by just NIS 49 million, to NIS 2.060 billion. Vehicle-backed credit, by contrast, jumped by NIS 1.147 billion, to NIS 4.452 billion. Put differently, about 76% of growth across those three products came from the collateralized vehicle product.
That is the key point. When the main growth engine is a product secured by a vehicle, loss metrics can improve without requiring a dramatic change in borrower behavior. The company itself explains that expected credit loss for consumer credit is measured through a PD/LGD framework segmented by internal customer rating and by asset type. A product-mix shift can therefore change the reserve and expected-loss profile directly, even before any major underwriting breakthrough.
That does not mean Max IT is blindly growing risk. There is clearly a disciplined risk machine here: model-based underwriting, separate models for new and existing customers, independent validation, and quarterly monitoring of risk limits. But that is still different from proving that underlying borrower quality improved across the book.
What Really Did Improve
The cleaner read on 2025 should not be dismissed as an accounting illusion. The loss metrics themselves did improve.
In credit to private customers, the credit-loss-expense rate fell to 1.29% from 1.87%, while the net charge-off rate fell to 1.41% from 1.86%. Card-receivable metrics also improved, but much more modestly: the loss-expense rate edged down to 0.35% from 0.37%, and the net charge-off rate to 0.37% from 0.42%.
| Metric | 2024 | 2025 | The right read |
|---|---|---|---|
| Credit-loss expense rate, private credit | 1.87% | 1.29% | A sharp improvement in realized loss metrics |
| Net charge-off rate, private credit | 1.86% | 1.41% | Lower realized loss severity |
| Credit-loss expense rate, card receivables | 0.37% | 0.35% | Improvement exists, but is much smaller |
| Net charge-off rate, card receivables | 0.42% | 0.37% | Only a moderate improvement |
That asymmetry matters. If underwriting had improved broadly across all consumer channels, it would be more natural to see a similar pattern in card receivables as well. Instead, the sharp move is concentrated exactly where the product mix changed the most. That does not rule out better underwriting. It does mean the first explanation should be mix.
Management’s language also points more to stability than to a decisive victory lap. The company says there has been no material deterioration in consumer-credit risk. But it still does not break performance down here by vintage, default behavior within each product, or underwriting-score migration. So there is still no clean basis for saying that solo loans themselves, for example, became materially safer regardless of the move into vehicle-backed lending.
What Still Does Not Fit A Clean Quality-Improvement Story
This is the part that prevents the conclusion from becoming too comfortable. While loss metrics improved, the broader quality indicators did not all move in the same direction.
The private problematic-credit ratio rose to 3.73% from 3.45%. The ratio of private exposures outside execution rating rose to 6.97% from 6.46%. In shekel terms, private problematic accruing exposures increased to NIS 441 million from NIS 363 million, and private exposures outside execution rating increased to NIS 1.157 billion from NIS 1.005 billion.
By contrast, private non-accrual balances barely changed at all, NIS 159 million versus NIS 158 million. That is why the improvement in the non-accrual ratio to 0.99% from 1.05% tells a more nuanced story than the headline suggests. The hard tail did not worsen, but it also did not clean itself up in any dramatic way. This does not look like a book that improved uniformly across the full risk chain. It looks more like a book where the heavier loss layer improved while broader stress buckets still showed pressure.
The reserve layer reinforces the same reading. The private allowance ratio fell to 1.82% from 2.04%, and at the total-exposure level allowance declined to NIS 362 million from NIS 386 million while gross balances rose to NIS 62.116 billion from NIS 59.190 billion. That is consistent with a book becoming safer. It is not, by itself, proof that borrower behavior improved across the board.
This is exactly where the company’s methodology matters. When the reserve is driven by both asset type and expected loss given default, a shift toward collateralized products can compress reserve intensity even without a major improvement in early-stage risk indicators. So the lower allowance cushion, together with the rise in broader problematic-credit measures, supports the read that the book became safer, but not necessarily cleaner at every layer.
What Would Actually Prove Better Underwriting
Max IT’s next test will not be only whether loss metrics stay low, but where the confirmation comes from. Three signals matter most:
- Further improvement without another big jump in vehicle mix. If vehicle-backed credit stops gaining share so quickly and loss metrics still improve, that would be much stronger proof of better underwriting.
- Relief in the earlier-warning buckets. The problematic-credit ratio and the outside-execution-rating ratio need to start falling, not only the tail-end metrics such as charge-offs.
- A reserve cushion that stays low for the right reason. If allowance remains lean while the broader problem indicators also stabilize or improve, it becomes easier to argue that the improvement is not just mix.
Until then, the conservative read remains the better one: 2025 showed that Max IT can grow through a safer product, and maybe through more precise underwriting as well, but it did not yet provide a clean proof that credit quality improved broadly across the book.
Follow-Up Conclusion
What 2025 really proved is a safer book, not necessarily a better borrower. That is not a small achievement. The move toward vehicle-backed lending is a real strategic decision, and a large part of the improvement in loss metrics probably comes directly from that choice.
But anyone trying to prove a deeper improvement in underwriting should stay careful. Loss metrics look better, charge-offs are down, and non-accrual balances did not deteriorate. On the other hand, broader problematic-credit measures rose, the allowance cushion fell, and most of the growth came from the safer product. So for now, it is easier to argue that Max IT improved book quality than to argue that it already proved a broad-based improvement in the risk quality of the borrowers themselves.
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.