Yaakov Finance: Why the Reserve Fell Even as the Book Grew
Yaakov Finance grew its gross credit book to NIS 3.18 billion, yet total reserves fell to NIS 9.7 million and the general reserve coefficient dropped to 0.27%. That helped 2025 earnings, but it also shifted more weight onto the model, the macro assumptions, and the need to prove that the improvement is durable.
The main article focused on the gap between book growth and dependence on bank lines. This follow-up isolates the accounting question inside that story: how did the gross credit book grow 36.6% to NIS 3.18 billion while total reserves fell 19.6% to NIS 9.7 million.
That is the core issue because for a non-bank lender, expected credit losses, ECL, are not only a prudence layer. They also shape how quickly volume growth turns into reported profit. When the general reserve coefficient drops to 0.27% from 0.46%, that is a 19 basis-point move on a book of more than NIS 3.1 billion. This is not a footnote. It is an earnings event in its own right.
There is a real positive side here. Balances that required specific provisioning fell to NIS 2.6 million from NIS 3.6 million, most of the growth came from Stage 1 loans, and the company says its historical loss rates improved. But the same year also contains a clear reminder that earnings quality is leaning on a model. The auditor flagged the general ECL reserve as a key audit matter because of judgement, macro assumptions, and subjective estimation.
What Actually Fell
The key number is not only the year-end reserve balance. The income statement also got help. In the company presentation, the combined general-and-specific reserve line moved from an expense of NIS 223 thousand in 2024 to income of NIS 2.092 million in 2025. That is a NIS 2.315 million swing, roughly 12% of the increase in net finance income after reserves.
| Metric | 2024 | 2025 | Change | Why it matters |
|---|---|---|---|---|
| Gross credit book | NIS 2,328.6 million | NIS 3,180.1 million | 36.6% | The earning base expanded quickly |
| Total reserve balance | NIS 12.1 million | NIS 9.7 million | 19.6%- | The balance-sheet cushion got smaller |
| General reserve balance | NIS 11.0 million | NIS 8.7 million | 21.4%- | This is the main driver |
| General reserve coefficient | 0.46% | 0.27% | Down 19 bps | Every basis point is now worth real money |
| Reserve line in the P&L | NIS 0.223 million expense | NIS 2.092 million income | NIS 2.315 million improvement | Earnings got help from the model |
The chart shows why this point cannot stay buried in a note. The book itself grew quickly, yet each shekel of exposure carried a lighter reserve burden by year-end. From an earnings-quality perspective, that combination is both stronger and more fragile.
Where the Decline Came From
Not Only Mix, But Lower Rates Inside Most Buckets
The decline did not happen simply because the book got larger. It happened because reserve rates inside most buckets fell. The most important bucket is Stage 1 loans. It already accounts for 83.1% of the non-specific book, up from 74.4% a year earlier. That balance jumped 52.8% to NIS 2.64 billion, yet the reserve attached to it fell to NIS 7.0 million from NIS 8.8 million. That alone explains NIS 1.84 million of the drop in total reserves.
The pattern was similar elsewhere. In Stage 2 loans, the reserve rate fell to 1.06% from 1.53%. In Stage 1 third-party endorsed checks, it fell to 0.27% from 0.31%. In Stage 2 endorsed checks, it fell to 0.83% from 1.13%. The company did not only get a larger book. It also got a model willing to place less cushion on most units of exposure.
The second chart sharpens the point. The book did not grow through the visibly troubled buckets. It grew mainly through Stage 1 loans, while balances under specific provisioning actually declined. That does not prove the risk is gone, but it does explain how the company could combine rapid growth with a smaller reserve stock.
The Model Also Became More Constructive, But Not to the Extreme
The company describes a three-step process. First, it checks whether any customers require specific provisioning because of a default event. Second, it identifies a significant increase in credit risk. Only the remaining balances then receive general provisioning. That part is built by sector and by service type, using the company’s own default history since inception, adjusted with forward-looking information.
The more interesting point is that the filing offers two complementary explanations, not one. On one hand, the company says that by year-end 2025 Israel’s macro backdrop had improved, security and economic uncertainty had dropped, bond yields had declined, and the Bank of Israel’s inflation, unemployment, and GDP forecasts had improved. Together with better historical reserve behaviour, that led to a lower reserve coefficient than in the prior year.
On the other hand, the same methodology note says the minimum and maximum reserve-rate range is set with the help of an independent external valuer, and that the company chose to use an average rate within that range. Its explanation is that uncertainty in the economy still exists, even though historical reserve rates improved thanks to more selective underwriting, more customers, and lower balances with large customers. That distinction matters. The coefficient fell, but not because management declared the risk gone. It fell because the midpoint of the model moved lower.
What This Means for Earnings Quality
This is not a proven accounting error. The lower reserve may be justified, and the filing gives it support. But it is also not the same as hard operating income that has already been collected in cash. It is income created because the model concluded that a larger book now needs a thinner cushion.
This is exactly where the key audit matter becomes relevant. The auditor did not flag the line because of an obvious hole. It was flagged because the calculation depends on significant estimates, the reasonableness of assumptions, and macro parameters. For the reader, the implication is straightforward: part of the 2025 earnings improvement rests on judgement and on an assumption set, not only on realised collections.
There is also a clean quantitative way to see how sensitive this question is. If the company had kept the 0.46% general reserve coefficient on the non-specific year-end 2025 book, the general reserve balance would have been about NIS 14.6 million, roughly NIS 5.9 million above the figure actually recorded. That is why a 19 basis-point move changes the picture much more than a first read suggests.
This does not mean the 2024 rate should mechanically come back. It means the rise in earnings quality still has to pass a test. If the book keeps growing and specific provisioning stays low, the company will have a good case that the lower coefficient reflects better underwriting and a better backdrop. If the coefficient starts rising again, part of the 2025 improvement will look more generous than durable.
What To Watch Next
The first signal will be whether Stage 2 balances and specifically provisioned balances stay small while the book keeps expanding. The second will be whether the general reserve coefficient can remain low without another sharp reset. The third is the income statement itself: in 2025 the reserve line moved from expense to income. It will be much harder to get the same tailwind again in 2026.
In the end, this follow-up does not contradict the main article. It reinforces it. Yaakov Finance did not only grow faster in 2025. Through the model, and with help from a friendlier macro backdrop, it also decided that each shekel in the book needs less protective weight. If that assumption holds, earnings quality really did improve. If it does not, this is exactly where a fast-growing book will start to look less clean.
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.