Beinleumi: How Much Capital Is Really Left To Return as the Spread Cools and Regulation Tightens

The main article already made the broad point: Beinleumi enters 2026 as a strong bank, but no longer as a bank enjoying the same unusually supportive spread conditions of 2024. This continuation isolates a narrower question: how much of that capital is truly available for return once the accounting surplus is separated from the capital layers that really constrain the bank, and once funding and liquidity are brought back into the picture.

That matters now because the first read looks tempting. Net profit for 2025 came in at NIS 2.26 billion, return on equity was 16.2%, and return on equity at the bank's target-capital level reached 19.1%. The March 2026 presentation then added a far more aggressive distribution tone: NIS 522 million already approved for March 2026, a possible framework for roughly another NIS 1 billion in additional dividend steps, and a plan to examine buybacks equal to about 25% of quarterly profit over the next two years. The problem is that the surplus seen from a distance is much larger than the surplus that actually remains after all the constraints are applied.

Not Every Capital Surplus Sits At The Same Layer

The March 10, 2026 immediate report on the approved repurchase plan gives the cleanest starting point. According to that filing, the bank had distributable retained earnings under the Companies Law of NIS 13.746 billion at the end of 2025, and NIS 11.355 billion that was actually permitted for distribution after the control-permit restriction. That is a very large number, but it sits at the legal and accounting layer. It does not mean that this amount is truly available to shareholders without affecting capital ratios, leverage or liquidity.

That gap between layers is the core of the story. At the end of 2025, common equity tier 1 stood at 11.10%, versus a regulatory requirement of 9.23% and a board target of 9.50% in a normal-business scenario. In amount terms, that left about NIS 2.49 billion above the regulatory minimum and about NIS 2.13 billion above the board target. But that is only half the picture. Total capital stood at 13.19% against a 12.50% minimum, which means the buffer at the total-capital layer was only about NIS 927.5 million. This is the non-obvious point: anyone looking only at CET1 will think the bank has a wide-open distribution buffer. Anyone looking at total capital as well will see that the real tap is much narrower.

The table below shows a static exercise on the year-end 2025 balance sheet. It does not try to predict future profit or future changes in risk-weighted assets. It simply shows how much room remains if year-end 2025 is held constant and only the already disclosed distributions and sensitivities are layered on top:

The meaning is straightforward. The March 2026 dividend fits inside the cushion. Even the extra NIS 128 million maximum buyback layer tied to fourth-quarter 2025 profit still works on the same static base. But from that point onward, the margin for error at the total-capital layer becomes small. The bank's own sensitivity shows that a further two-notch Israel downgrade by S&P would reduce CET1 by 0.47 percentage points and total capital by 0.56 percentage points. On the year-end 2025 base, combining the March dividend with that sensitivity would already push the total-capital ratio below minimum. That is not the base case, but it is exactly why it is wrong to translate NIS 11.355 billion of permitted retained earnings into real distribution capacity.

This is also where it becomes critical to separate two very different buyback stories. The March 10, 2026 immediate report deals with a small plan of 32,815 shares and an estimated cost of about NIS 10 million, designed to support the bank's equity-compensation program and hold treasury shares against future dilution from options. The board explicitly states that this plan is not expected to affect capital structure, capital adequacy, leverage or liquidity. That is technical and immaterial at group scale. The presentation, by contrast, discusses a possible future program of buybacks equal to about 25% of quarterly net profit over the next two years. These are not the same program and not the same order of magnitude. The market should not confuse a roughly NIS 10 million anti-dilution repurchase with a materially larger capital-return buyback program.

The Spread Is Cooling Exactly Where The Funding Mix Changed

If the question is what is squeezing that buffer, the answer starts with the earnings engine. In 2025, net interest income rose slightly to NIS 4.822 billion, but recurring financing profit actually fell 1.3% to NIS 4.868 billion. Net interest income to interest-bearing assets fell to 2.12% from 2.29%, the margin from lending activity fell to NIS 2.009 billion from NIS 2.073 billion, and the margin from deposit-taking activity fell to NIS 2.501 billion from NIS 2.614 billion. So even without a credit-quality shock and without a big provisioning problem, the engine that had been producing peak-era profitability is already moving back toward normal.

The annual report is explicit on why. The bank says the decline in deposit-taking margin was driven, among other factors, by a change in public-deposit mix: more activity from institutional bodies and large businesses, less from retail and small businesses, alongside lower US dollar rates. In other words, deposits kept growing, but less from the cheap and sticky pockets, and more from the wholesale, price-sensitive ones.

The most striking number here is the jump in institutions, from an average balance of NIS 64.729 billion in 2024 to NIS 90.271 billion in 2025. Large businesses also rose from NIS 19.828 billion to NIS 23.331 billion. By contrast, households fell from NIS 73.087 billion to NIS 69.977 billion, and small and micro businesses fell from NIS 26.659 billion to NIS 24.951 billion. This is not just movement between rows in a table. It explains why the bank is still adding balance-sheet scale, but no longer enjoying the same deposit economics as in the last two years.

And the pressure does not stop at margin. By the end of 2025, the three largest depositor groups already accounted for 9.7% of public deposits, up from 6.5% at the end of 2024. That does not turn the story into a liquidity problem on its own, but it does mean that continued capital return now depends not only on profit generation, but on profit generation built on a funding base that is becoming more wholesale.

Liquidity Is Still Strong, But The Direction No Longer Supports Automatic Generosity

It would be too simplistic to read the fall in LCR and NSFR as a sharp deterioration in absolute liquidity. The bank actually increased average high-quality liquid assets to NIS 101.452 billion in the fourth quarter, from NIS 87.721 billion a year earlier. Core liquid assets reached NIS 119.2 billion by year-end 2025, and the bank also notes that it issued about NIS 4.1 billion of bonds and subordinated instruments during 2025 with maturities of one to three years. This is not a story of absolute liquidity weakness.

The problem is the relative direction. Net cash outflows for LCR rose to NIS 78.907 billion from NIS 53.042 billion, up 48.8%, much faster than the 15.7% increase in HQLA. In NSFR, available stable funding rose only 2.6% to NIS 136.362 billion, while required stable funding rose 12.8% to NIS 107.410 billion. Put differently, the bank is holding a larger cushion, but uses and regulatory outflow assumptions are growing faster than that cushion.

The bank's own explanation for the weaker liquidity ratios is fully consistent with the margin story: more short wholesale deposits, more long-duration credit, and some decline in retail and small-business deposits. It also says that 87% of contractual deposit cash flows, including current-account credit balances, are short-dated at less than three months, even if in practice these deposits tend to renew. That is exactly the difference between accounting stability and regulatory stability. The balances stay, but the regulatory coefficients become less favorable.

That is why the capital-return question is not just about how much profit remains after the dividend. It is also about how much flexibility management wants to preserve against a funding system that is becoming more dependent on institutional money, market issuance and liquidity ratios that remain high, but are moving steadily lower.

Regulation Is Not Closing The Tap, But It Is Shrinking The Room For Error

Precision matters here. Not everything labeled Basel or regulation is a headwind to the buffer. In the risk report, the bank says the updated operational-risk instruction, effective January 1, 2026 as part of Basel IV adoption, is expected to increase the capital-adequacy ratio by about 0.14 percentage points. That is a modest positive, not a tightening.

But the same risk report also makes clear that in credit risk the bank has only completed a quantitative impact study for Basel IV, and that no binding updated Instruction 203 has yet been published. So there is no basis for treating future relief, or future tightening, as if it were already available for distribution. Capital that has not actually been freed is not shareholder capital yet.

Outside Basel as well, the board report points to a tougher competition and regulatory environment: reforms aimed at increasing competition in retail banking, a draft supervisory framework for small and new banks, reconsideration of compensation models in securities activity, and an explicit warning that such changes may affect banking product prices and market interest rates. So even if current capital is enough for short-term distributions, the quality of the profit that refills that capital may be lower than the market has grown used to.

That is exactly why the sovereign-rating sensitivity matters more here than at a bank sitting on a much wider buffer. According to the bank's own estimate, a further two-notch Israel downgrade by S&P would reduce CET1 by 0.47 percentage points and total capital by 0.56 percentage points. That does not erase the buffer, but it does change the order of magnitude. Once the total-capital surplus is less than NIS 1 billion on a year-end 2025 base, even a few hundred million shekels of macro sensitivity moves from background noise to a real distribution constraint.

Conclusion

The answer to the title question is not one number. At the legal retained-earnings layer, Beinleumi has a lot of room. At the CET1 layer, the cushion is still comfortable. But at the total-capital layer, in funding mix and in liquidity direction, the room for error is already much smaller. That is why the capital story for 2026 and 2027 is no longer "how much can be distributed", but "how much must be re-earned in order to keep distributing".

The March 2026 dividend and the technical roughly NIS 10 million buyback are not the problem. Even the additional NIS 128 million layer tied to fourth-quarter 2025 profit still does not break the picture. But the broader framework, more dividend steps and buybacks equal to roughly 25% of quarterly profit over two years, already requires four things to happen at once: the deposit margin must stop eroding, risk-weighted assets must stop growing faster than capital, the deposit mix must stabilize, and no external capital shock should arrive through Israel's rating or through more conservative regulatory implementation. Without that, Beinleumi's excess capital remains real, but far less endless.