Harel 2025: Is the New CSM Actually Better Than the Legacy Book?
Harel’s CSM is now more heavily tied to growth products and less to run-off books, but near-term earnings will still be released largely from the legacy portfolio. In 2025 the stock of future profit looked better, yet the pace at which the old book burns off still matters as much as the pace at which the new one is built.
The main article argued that Harel’s 2025 profit no longer rested only on a strong capital-markets year, and that the CSM balance was still moving up. This follow-up isolates the narrower question that really matters now: what exactly is replacing the legacy book that is being released. For a life and health insurer, this is not an accounting side note. CSM, the contractual service margin, is the reservoir of future profit that has not yet been recognized. If the new reservoir is being built from shorter-duration products that are still being actively sold, earnings quality improves. If it is swelling mostly because of accretion, assumption changes, or older books still carrying the load, the headline is stronger than the underlying economics.
The short answer is yes, but not because the NIS 17.2 billion headline says so on its own. The strongest evidence is mix shift, not just balance growth. At year-end 2025, about NIS 12.0 billion of net CSM already came from life and health growth products, while only about NIS 5.2 billion sat in run-off books. New-business CSM reached NIS 1.707 billion, above the release from growth products and slightly above total annual CSM release. This is no longer a picture of an old book simply unwinding while the company lives off embedded margins. It is a picture of a legacy future-profit stock being gradually replaced by a new one.
But that is exactly where discipline matters. Anyone who looks only at the total balance can miss the fact that near-term earnings are still heavily tied to the old book. The expected release schedule shows about NIS 5.7 billion of CSM to be recognized in 2026-2030, of which about NIS 4.4 billion comes from run-off books and only NIS 1.3 billion from growth products. The inventory already looks newer than the earnings stream.
| Test | What 2025 says | Why it matters |
|---|---|---|
| Net CSM balance | NIS 17.177 billion, of which NIS 11.976 billion comes from growth products | Roughly 70% of the future-profit stock already sits in a newer book |
| New-business CSM versus release | NIS 1.707 billion versus NIS 1.622 billion of release | Harel is not just consuming future profit, it is rebuilding it |
| New-business CSM versus growth-product release | 137% ratio | The new book is growing faster than it is being monetized |
| Expected 2026-2030 release | NIS 4.4 billion run-off versus NIS 1.3 billion growth products | Near-term earnings are still legacy-heavy |
| Actuarial-study effect on CSM | Net decline of NIS 11 million | The CSM improvement was not manufactured by broad assumption relief |
Where The New Book Really Looks Better
The first signal is the mix. Out of NIS 17.177 billion of net CSM, about NIS 11.976 billion is already tied to growth products that the group is still actively selling in life and health. That means roughly 70% of the future-profit inventory already sits in products that are still being marketed, rather than in portfolios whose main role is to release through earnings over time. That is not a cosmetic distinction. It means the stock of future profit is moving toward shorter-duration products that Harel is still replenishing through new sales.
The second signal is build rate. In 2025 Harel released about NIS 1.622 billion of gross CSM, but generated NIS 1.707 billion of CSM from new business. At the growth-product level the ratio was even stronger, 137%. That is a sharp message: in the engines Harel actually wants to scale, the new book is being created faster than it is being consumed. Without that, the entire “future profit reservoir” story would remain a technical accounting label. With it, it begins to look like a real business engine.
The third signal is in today’s underwriting profit, not only tomorrow’s stock. In life insurance, underwriting profit reached NIS 919 million in 2025, and about NIS 651 million of that came from life-risk products, the growth side, versus only about NIS 268 million from the older savings books, yield-guaranteed and profit-participating. Health tells a similar story. Underwriting profit reached NIS 978 million, while growth products, medical expenses, critical illness, personal accidents, and short-term health, contributed about NIS 804 million. The run-off long-term-care books contributed only about NIS 174 million net, after the negative drag from group LTC. In other words, even in the current year’s earnings, not only in the future stock, the center of gravity has already shifted.
Where The Legacy Book Still Dominates
The good news on inventory quality does not erase the fact that the next few years still lean on old economics. Harel’s expected CSM release profile shows that roughly one third of the balance is expected to be recognized over the next five years, but inside that window most of the amount still comes from run-off books. In 2026-2030, the split is about NIS 4.4 billion from run-off and only NIS 1.3 billion from growth products. Even in 2031-2035, the legacy bucket is still larger, NIS 2.5 billion versus NIS 1.0 billion. The balance only gets cleaner later.
That distinction matters because it separates inventory quality from near-term earnings quality. It is fair to say the new book is better. It is not yet fair to say the near-term earnings stream has already detached from the legacy book. Anyone trying to read 2026 as if it were already a pure “new CSM” year is getting ahead of the company’s own release timetable.
The same point comes through in the breakdown of insurance-service profit. In 2025 total underwriting profit came to NIS 2.314 billion. Of that, NIS 1.622 billion came from CSM release, another NIS 343 million from RA release, and only NIS 417 million from non-life net. So even after the underwriting improvement, the core life and health earnings base is still materially driven by the release of profit reservoirs built earlier. That is not a flaw, it is how the accounting model works. But it does mean the quality test is not whether CSM exists. It is whether the new book is replacing the old one at the right speed.
Why The Balance Increase Alone Is Not Enough
This is the most important gap between headline and understanding. Net CSM rose from NIS 16.359 billion to NIS 17.177 billion, but that increase did not come only from new business. The 2025 roll-forward included NIS 543 million of interest accretion and another NIS 319 million of changes related to future services. Against that, CSM release totaled NIS 1.622 billion and reinsurance changes reduced the balance by NIS 129 million. In other words, new-business CSM nearly covered release plus the reinsurance drag, but not much more than that. The move above zero was also supported by more technical items.
That does not make the story weak. It just clarifies what should be measured next. The strongest signal is not that the stock grew, but that its composition improved. If the balance keeps rising in coming years mainly because of accretion and financial effects, that would be a weaker story. If it keeps rising because growth-product new business keeps building faster than release, that is a genuine economic upgrade.
The actuarial studies reinforce that reading. Their total effect on CSM in 2025 was only a NIS 11 million net decline. Within that, life saw an NIS 82 million CSM decline, while health posted a NIS 71 million increase. At a deeper level, the lapse study in life cut CSM by about NIS 120 million, while the occupational-disability morbidity study added about NIS 40 million to CSM and another NIS 50 million to profit because it reduced incurred-claims liabilities. In health, studies on medical expenses, critical illness, and accident products added about NIS 289 million to CSM, while private long-term-care studies cut about NIS 218 million. That matters a lot. The new book does not look better because management broadly eased assumptions. It looks better because genuine value is being built in the growth products, while legacy LTC and older savings books still generate noise and require monitoring.
So, Is The New CSM Actually Better?
The answer is yes, but it is a gradual improvement, not an overnight reset. Harel’s future-profit stock is already more tied to medical expenses, critical illness, and life-risk products, and less to old savings and LTC run-off books. The 2025 underwriting mix also shows that the newer engines are now carrying a larger share of current results. That is exactly the upgrade the main article pointed to, but did not unpack fully.
Still, the next step has to be read carefully. The market will continue to receive a large part of profit through the release of legacy books, especially in the 2026-2030 window. So the next real test is not only whether total CSM keeps rising, but whether the ratio of new-business CSM to release stays positive, and whether the weight of growth products inside the release stream itself starts moving up. If that happens, Harel’s earnings quality will genuinely start to look different. If it does not, 2025 will remain a year in which the inventory looked cleaner faster than the earnings stream did.
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.