Bank Leumi in the first quarter: earnings held, but 2026 is now a credit and capital test

Bank Leumi opened 2026 without an earnings break, but also without a full answer to the question left by 2025: what replaces net interest margin when it keeps falling. Net profit fell only 2.4% to NIS 2.346 billion, and it actually rose after excluding the special bank tax, but NIM fell to 1.96% from 2.35% a year earlier and 2.01% in the fourth quarter. The bank held the result through credit growth, higher fees, non-interest income and lower expenses, which means the earnings engine is still working, but with less help from rates. The more important shift is that the 2026 test has moved from profit alone to two layers: credit quality, especially construction and real estate, and capital room for both growth and distributions. Collective provisioning rose to NIS 264 million even though total credit loss expense was only NIS 166 million, and the credit-risk sensitivity shows that a 1% macro deterioration could add about NIS 1.046 billion to the collective allowance. At the same time, a 55% payout and a new buyback plan still signal confidence, but they depend on CET1 not falling below 11.15% and on the bank's ability to absorb risk-weighted asset growth. The current read is positive but less comfortable: the bank showed that profitability does not break at a lower margin, but the coming quarters must prove that the cost does not arrive through provisions, capital or a slower payout.

What The Bank Is Selling Investors Now

The bank is a banking earnings machine with two engines that need to work together: interest spread on a large credit and deposit base, and an income and efficiency layer that lets profit hold up when spread erodes. In the first quarter, the credit base grew quickly: net credit to the public rose to NIS 547.8 billion from NIS 520.0 billion at the end of 2025, up roughly 5.4% in three months. That is a strong engine, but it also increases risk-weighted assets, consumes capital and widens the area where credit quality can surprise negatively.

The 2025 annual analysis framed this year as a stabilization and proof year: net interest margin was no longer carrying most of the profit alone, real estate had become the main credit friction point, and the capital structure had to support both growth and distributions. The first quarter did not close those questions, but it did change their weight. The margin concern is no longer theoretical, because the decline continued after the fourth quarter. On the other hand, the bank showed that its operating and financial offset layer is stronger than the margin figure alone suggests.

Margin Fell, Fees And Efficiency Prevented A Break

The number that explains the quarter is not the small decline in net profit, but the fact that the bank delivered almost similar profit despite sharp margin erosion. Net interest income fell 2.7% to NIS 3.909 billion, while NIM fell by 39 basis points versus the parallel quarter. Sequentially, net interest income recovered modestly from NIS 3.824 billion to NIS 3.909 billion, but the margin itself continued to fall from 2.01% to 1.96%.

The offset came from other places. Non-interest income rose 13.7% to NIS 1.556 billion, and its share of total income rose to 28.5% from 25.4% a year earlier. Fees rose 10.6% to NIS 1.129 billion, with growth in securities fees, financing activity, distribution products and foreign-exchange conversions. At the same time, operating expenses fell 8.0% to NIS 1.592 billion, and salary expenses fell to NIS 879 million from NIS 960 million.

That is why pre-tax profit actually rose 3.0% to NIS 3.707 billion. The decline in net profit was mostly tax: the effective tax rate jumped to 42.6% from 35.9% a year earlier because of the special bank payment. After excluding that special tax, net profit was NIS 2.604 billion versus NIS 2.553 billion in the parallel quarter, and return on equity was 15.1% instead of 13.6%. This also explains the target reset: the 2026 net profit target fell to NIS 9 billion to NIS 11 billion, while the 2027 target rose to NIS 10.5 billion to NIS 12.5 billion. Management is asking the market to separate temporary tax from earnings power, but the coming quarters must show that this power comes from fees, efficiency and credit quality, not only from accounting timing or supportive capital markets.

Collective Provisioning Puts Real Estate At The Center

Total credit loss expense rose to NIS 166 million from NIS 55 million a year earlier. That is still a very low rate relative to the portfolio, 0.12%, so looking only at the total line can miss the point. Beneath it, collective provisioning was NIS 264 million, partly offset by specific recoveries or releases of NIS 98 million. In other words, profit benefited from the specific-portfolio layer, but the macro and collective-estimate layer became more expensive.

The stronger signal is in allowance sensitivity, not only in the quarterly charge. A uniform 1% deterioration in macro parameters would increase the collective allowance by about NIS 1.046 billion, while a 1% improvement would reduce it by about NIS 385 million. That does not mean such a loss will arrive, but it does show that the risk is not symmetrical: when macro conditions deteriorate, the allowance can move faster than the quarterly expense suggests.

In real estate, the picture is still contained, but less quiet. The focused real estate credit analysis noted in early March that the friction was appearing first in project finance, deferred-payment structures and sales pace, not only in household mortgages. In the first quarter, that friction did not become a broad credit event, but it also did not disappear. Credit risk in construction and real estate in Israel rose to NIS 226.7 billion, problematic credit in those industries rose to NIS 2.024 billion from NIS 1.898 billion at the end of 2025, and the problematic-credit ratio rose slightly to 0.89% from 0.87%.

The sector details matter more than the total. In construction, problematic credit rose to NIS 1.176 billion from NIS 1.073 billion, while nonaccrual debt fell to NIS 162 million from NIS 182 million. In real estate activity, problematic credit rose slightly to NIS 848 million, and nonaccrual debt rose to NIS 271 million from NIS 202 million. If friction remains mostly in the accruing problematic layer, it can still be treated as early and manageable pressure. A broader move into nonaccrual debt would change the quality of risk.

Regulation supports that caution. Bank of Israel limits on contractor-subsidized bullet and balloon housing loans, and a 150% risk weight for projects where more than 25% of apartments were sold with significant payment deferral, are aimed at the place where the local market tried to preserve sales through easier payment terms. The bank complies with these limits, and implementation is not material to its capital ratios, but the limits clarify the test: apartment sales with 20/80-style terms are not the same as normal project credit advancing at a natural pace.

Payout Continues, But Capital Now Sets The Pace

The bank continues to behave like a cash-return equity, not only as a growing bank. The board approved a distribution of 55% of first-quarter profit: a cash dividend of NIS 968 million, equal to 41.25% of net profit, and a NIS 322 million buyback. It also approved a new buyback plan of up to NIS 1.6 billion through April 2027, after the 2025 plan was completed in May 2026 with the purchase of 21.5 million shares for NIS 1.45 billion.

The payout looks comfortable if viewed only through profit, but the real operating limit is capital. CET1 was 11.74% at the end of March, down from 12.05% at the end of 2025 and above an internal target of 10.85%. The total capital ratio was 14.07%, almost unchanged from the end of 2025, but only 0.57 percentage points above the 13.50% requirement. Risk-weighted assets rose to NIS 580.6 billion from NIS 561.1 billion, and every additional NIS 1 billion of risk-weighted assets lowers CET1 by about 0.02 percentage points and the total capital ratio by about 0.03 percentage points.

The capital-structure analysis marked total capital as the tighter layer, and the first quarter confirms it. Tier 2 capital grew to NIS 13.5 billion from NIS 11.4 billion at the end of 2025, so the total capital ratio did not erode despite the rise in risk-weighted assets. That is an important move, but it also means the payout relies on active capital-structure management, not only on surplus profit. The buyback plan has a clear stop condition if the last published CET1 ratio falls below 11.15%, so the distance between 11.74% and that stop point is now a number the market must track.

The first quarter strengthens the conclusion that the bank remains on the strong side of the system, but also sharpens the weak point of 2026. Net profit of NIS 2.346 billion and ROE of 13.6% in an environment of special tax, margin decline and higher collective provisioning are good numbers. After excluding the special tax, ROE was 15.1%, so it is hard to argue that the bank is breaking. The more precise argument is that profitability currently holds because of a broader mix of sources, not because of a comfortable interest-rate spread.

The next phase will be decided by three tests: whether NIM stabilizes around current levels or keeps falling even as credit grows, whether collective provisioning and real estate friction remain an early caution layer or start turning into specific losses, and whether CET1 remains far enough above the buyback stop point to allow distributions without slowing portfolio growth. If all three tests work, the first quarter will look like a reasonable start to a transition year. If one moves against the bank, the market will focus less on whether profit is high and more on the price the bank is paying to preserve it.