Direct Finance: Can the Car Book Recover Margin Without Buying Growth at a Higher Cost?

What This Follow-up Is Testing

The main article set the wider frame: mortgages are now carrying more of the profit load, but the car book is still where Direct Finance has to prove that growth is not being bought at too high a cost. This follow-up isolates that exact issue. The question is not whether the company can keep originating auto loans. The question is whether it can restore the economics of that book while three pressure points are hitting at the same time: more expensive distribution, less generous fair-value economics, and weaker recoveries.

That matters now because the car segment is no longer the main profit engine, but it is still the larger revenue engine. In 2025 the car segment generated ILS 1.223 billion of revenue, versus ILS 312.3 million in mortgages. So even after mortgages overtook on net profit, the car book still determines whether Direct Finance has two healthy engines or one strong engine and one reset segment.

• Growth did not disappear, margin did. Vehicle-secured balances rose to about ILS 6.09 billion from ILS 5.3 billion, and car-loan originations increased by 28%. Even so, car-segment net profit fell to ILS 57.7 million from ILS 94.4 million.
• Commissions shifted from a distribution cost to a growth cost. Commission income in the car segment rose 19%, but the company also says commission expense increased by about ILS 25 million, partly because of new settlements with Trade Mobile and Eldan and partly because average commissions moved higher as competition intensified.
• Fair value still helps, but less per unit. Fair-value changes rose to ILS 277.4 million, yet the price effect was negative because discount rates increased while customer rates declined.
• Collateral did not prevent the reset. 96% of the loans originated in 2025 were backed by collateral, yet the credit-loss rate on car loans rose to 4.08% from 3.32%, and recovery fell by about 7%.
• There is one early positive signal. In the fourth quarter, car-segment expenses rose only 8% while originations jumped 59%, and segment net profit increased to ILS 14.4 million from ILS 6.5 million.

Where Margin Broke

The cleanest way to read 2025 is through the paradox. The car segment did not suffer a collapse in activity. Revenue from financing rose to ILS 1.203 billion from ILS 1.177 billion, and net revenue rose to ILS 867.5 million from ILS 842.4 million. But after credit losses and operating costs, profit before tax dropped to ILS 97.6 million from ILS 151.3 million. The problem was not that the company stopped producing loans. The problem was that the layers above origination all weakened at once.

One useful detail is that the noise around the core business actually improved. The company shut down car sales activity, so revenue from car sales and advertising fell by 81%, but gross profit on that line moved from a loss of ILS 1.5 million to a profit of ILS 7.2 million. In other words, one messy side activity stopped weighing on the segment. Even then, profit before tax still fell by 35%. That tells you the deterioration sits inside the lending economics, not outside them.

That chart says a lot. Interest and indexation income actually fell by 5%, while fair-value gains and commissions rose sharply. The segment therefore relied more on lines that are highly sensitive to transfer economics, customer pricing, and distribution arrangements, and less on the steady interest income of a seasoned book. That does not automatically make the model weaker, but it does make profitability more exposed to market conditions and to the price the company has to pay to keep volume moving.

The waterfall sharpens the point. Even after an ILS 8.7 million improvement from shutting down the car-sales line, and even after ILS 25.1 million of additional net financing income, two layers wiped most of that out. Credit losses added ILS 32.7 million of pressure, and total expenses added another ILS 54.8 million. That is not a quarterly wobble. It is a margin reset.

Commission Inflation Became Part of the Story

This is probably the most important finding in the follow-up. When a lender grows fast, the real question is not only how many loans it originates, but how much it has to pay to bring those loans in. In 2025 selling, marketing, and G&A expenses in the car segment rose by ILS 54 million. The company breaks that increase into three buckets: about ILS 25 million of higher commission expense, about ILS 8 million of additional advertising expense tied to a new branding launch, and a more moderate increase in the rest of the cost base.

Not all of that expense growth has the same quality. Branding spend can be more temporary. Commission inflation is a different issue. Here the company explicitly says the increase came not only from higher volume, but also from new settlement arrangements with Trade Mobile and Eldan and from a rise in average commission as competition intensified. That is no longer a side-effect of growth. It starts to define the quality of growth.

The related-party and commercial footnotes make the picture more specific. On January 30, 2025, the company signed an amendment with Trade Mobile under which, until the joint venture receives its license, Direct Finance pays a fixed commission for each loan originated for those customers. On March 17, 2025, it signed a similar amendment with Eldan. In other words, distribution channels that may eventually move into joint-venture economics remained, during 2025, in an interim model where the company pays a fixed fee per originated loan. That is exactly how a lender can preserve volume while giving up economics on each unit.

It would still be too strong to say that all of this cost inflation automatically becomes permanent. The Eldan agreement expired on December 31, 2025, after its conditions precedent were not met, so the filing does not prove that the full structure rolls forward unchanged into 2026. But the filing does prove something important: in 2025 Direct Finance had to pay more to defend origination volume. That is the heart of the reset.

Fair Value and Transfers No Longer Cover Everything

The company has long operated with a model that includes loan sales. In the car segment it says that, as part of its business model, it typically assigns about half of the car loans it originates, and that assigned loans carry materially lower credit losses. On paper that should help protect segment economics: grow, transfer part of the book, preserve capital and liquidity, and recognize fair-value or sale economics along the way.

In 2025 that mechanism still helped, but it became less generous. Fair-value changes in the car segment rose to ILS 277.4 million from ILS 244.6 million, mainly because the volume of loans designated at fair value or transferred increased. But the company also says the price effect was negative by 25%, because the discount rate used for fair-value estimation rose with indexed government bond yields, while customer loan rates declined.

That is easy to miss on a first pass. A higher fair-value line does not mean the economics per loan improved. It can simply reflect more volume flowing through the channel. That is what happened here. More volume ran into less favorable market terms. So larger transfer volumes alone do not solve the issue. If each new pool is sold or measured against a higher discount curve and a lower customer rate, profit per unit still deteriorates even while activity looks strong.

This is why transfers should no longer be viewed as an automatic cushion. They remain an important funding and capital-management tool, but they cannot by themselves offset weaker customer pricing and fiercer competition.

Recoveries Weakened Even Though the Book Stayed Secured

One easy mistake in a story like this is to describe the problem as simple credit deterioration. Credit is part of it, but not all of it. 96% of the loans originated in 2025 were backed by collateral. So this is not a case of a major shift into unsecured lending. Yet the credit-loss rate on car loans rose to 4.08% from 3.32%, while supplementary loans actually improved to 4.38% from 5.94%.

That matters because it isolates where the deterioration occurred. Earlier underwriting tightening already helped the supplementary portfolio. The pressure moved into the core auto-loan book. The company explicitly says the main reason was a decline in recovery on loans that entered default, alongside growth in the average book. That is not the same as saying underwriting collapsed. It means the company extracted less value from problem loans once they had already gone bad.

That table is useful for one reason: the collateral did not disappear. Even in 2025 the collateral value attached to defaulted car loans remained above the net carrying value. Yet the company still says recovery fell by about 7%. So the issue is not the absence of collateral on paper. The issue is that the company got less back from that collateral in realized terms than it used to. That is a real yellow flag in vehicle lending, because it weakens the intuitive comfort investors tend to attach to a secured book.

2026: The Efficiency Plan Can Buy Time, Not Solve the Whole Gap

The good news is that management did not ignore the problem. In July 2025 the company launched a focused efficiency plan in the car segment, including a roughly 13% reduction in car-segment headcount and additional cost cuts. The expected annual savings amount to about ILS 40 million before tax, with a negligible contribution in late 2025 and a full effect starting in 2026. The investor presentation repeats that target.

This is a cost repair plan, not a yield repair plan. If the full savings are delivered, they can offset a large part of the ILS 54.8 million increase in total car-segment expenses between 2024 and 2025. That is meaningful. There is even an early hint in the fourth quarter, when expenses rose just 8% while originations rose 59%. But the plan does not change customer rates, it does not lower discount rates by itself, and it does not repair the recovery deterioration.

So the right 2026 question is not whether the efficiency plan "works." It probably will work at the cost layer. The real question is whether it is matched by enough improvement in the three deeper variables. If not, it only buys time. If yes, it can become the trigger for a real margin recovery.

Conclusion

2025 was a margin-reset year in the car segment, not a growth-failure year. Direct Finance again showed that it can produce volume, keep the book largely secured, and continue to use transfers as part of the model. But it also showed that this volume came with more expensive commissions, less generous fair-value economics, and weaker recoveries on problem loans. That is a different car-book economics profile from the one investors had grown used to.

The thesis from here is simpler than it looks. The market does not need more proof that the auto book can grow. It needs proof that each new car loan can once again generate a reasonable return after commissions, after fair-value assumptions, and after defaults. If efficiency savings flow through, if recoveries stabilize, and if transfer economics stop deteriorating, the car segment can return to being a decent profit engine rather than only a volume engine. If one of those three variables stays under pressure, 2025 will look less like a one-off dip and more like the reset point to a structurally lower margin.