17% ROTE, an Upgraded Guide, and a Stock That Has Already Paid for Both: Downgrading ING to Hold

Results vs. Consensus

2Q2026 Scorecard

ING is a euro-reporting IFRS-EU bank, and the consensus that matters is the euro-denominated poll the company compiles from contributing brokers ahead of each print. Fifteen brokers submitted for this round, which was collected between 9 and 15 July, two weeks before the print. That is the basis the Amsterdam line trades against and the basis used throughout this note. Dollar-converted vendor estimates circulate for the New York depositary receipt and are not comparable to a euro income statement, so they are excluded. ING reports before the European open: the release landed on the morning of 30 July and the analyst call followed at 09:00 CET the same morning.

Two mechanical notes before the interpretation. The poll is a set of line-item medians, so the components do not add to the subtotals; a line-by-line bridge to the €204M profit-before-tax gap reconciles to €203M, with the residual being non-additivity rather than performance. And "all other income" is not a line on the face of the income statement: it is the poll's own construction, defined as total income less commercial net interest income less fee income, which gives €6,284M less €4,174M less €1,278M, or €832M against a €700M median and a €770M maximum.

Nine of the eighteen lines above did not merely beat the median, they cleared the top of the broker range: total income, gross result, profit before tax, net result, earnings per share, return on tangible equity, all other income, risk costs and risk-weighted assets. On any conventional reading this is a very large beat. The scorecard's most interesting cell, though, is the one marked in line.

Revenue

Total income of €6,284M rose 10.2% year-on-year and 7.9% sequentially, and cleared the top of the broker range. The composition is where the reading changes. Commercial net interest income, the company's measure covering lending and liability products and stripping out Financial Markets and Treasury funding, reached €4,174M, up 10.7% year-on-year and 2.8% sequentially. Against the poll it was neither a beat nor a miss but an exact match on the median. Fee income of €1,278M grew 13.9% and beat by 3.6%, sitting just under the €1,284M range top. All other income of €832M beat by 18.9% and cleared its range top by €62M.

So the €160M income beat is 83% attributable to the line the bank does not control and explicitly guides down for the second half. That is not a criticism of the quarter, which was genuinely strong on volumes: net core lending grew €15.2bn, an 8.1% annualised pace, and net core deposits €15.9bn. It is a statement about what the Street was surprised by. Volume and margin, the two things that produce the recurring line, surprised nobody.

Margins

The cost/income ratio of 49.1% is down from 53.2% a year ago and 55.3% in the first quarter, and 50bp better than the poll. The sequential improvement flatters: regulatory costs fall from €324M in the first quarter to €78M in the second because ING recognises annual charges in full in the first quarter. The year-on-year comparison is the honest one, and on that basis the 410bp improvement is income-led. Income rose €582M, or 10.2%, while operating expenses rose €52M, or 1.7%. Inside that expense line, incidental items fell from €116M to €46M, so everything else rose €122M, or 4.2%. Income growth did the work and lower restructuring charges added to it; the underlying cost base was a partial offset, not a contributor.

The underlying cost line is the one place the quarter was worse than expected. Operating expenses excluding regulatory costs and incidental items were €2,961M, up 4.2% year-on-year and 3.4% sequentially, against €2,941M in the poll. Last quarter the same measure grew 1.1%. Management attributed the step to annual salary increases, marketing costs that had been unusually low in the first quarter, and the consolidation of Goldman Sachs TFI in Poland, and pointed to year-to-date cost growth of 2.7% as evidence the full-year guide is intact. Both statements are true. Neither addresses the arithmetic in the guide, which is dealt with below.

The commercial net interest margin was 2.26%, flat sequentially and 3bp better year-on-year, and it is the composition inside it that matters. The liability margin rose 3bp to 107bp on the replication tailwind, partly offset by campaign-related deposit costs that had been unusually low in the first quarter. The lending margin fell 2bp to 1.24%. Management was explicit that the lending-margin decline was mix and not price competition.

Earnings

Earnings per share of €0.68 grew 21.4% year-on-year on a net result up 16.2%, and cleared the €0.67 poll maximum. The 5.2 percentage point gap between earnings and profit growth is the buyback: average shares outstanding fell to 2,868.7M from 3,012.5M, a 4.8% reduction, and period-end shares stand at 2,859.7M against 2,980.8M a year ago.

Below the line, the effective tax rate of 30.9% is the one clearly unfavourable variance in the scorecard: taxation of €902M against a €798M median, and €37M above the poll's maximum. Two things drive it. The prior-year comparison is flattered by a tax refund that pulled 2Q2025 to 26.7%, and Poland's substantially higher bank corporate income tax rate applies from 1 January 2026. The 1H2026 rate of 30.0% sits at the top of the 29–31% range guided in the first quarter, so this is guidance being met rather than a negative surprise, but it is a permanent drag that the pre-tax beat has to work through.

Year-Over-Year Comparison

The income statement below is as filed. Note the two net-interest-income lines. Commercial net interest income is the company's measure covering lending and liability products; the IFRS line adds the funding costs and interest of Financial Markets and Treasury positions whose offsetting revenue is booked in other income. The two grew at materially different rates again this quarter, and reading only the IFRS line overstates the recurring improvement by roughly six percentage points.

The single most important number in that table is the gap between the two interest lines. Reported net interest income grew 16.7%; commercial net interest income grew 10.7%. The six-point gap is other net interest income improving from negative €236M to negative €47M, and it is not income creation. Those are Financial Markets and Treasury funding costs whose associated revenue sits in other income, which fell 19.4% over the same period. Marking ING on the IFRS interest line double-counts a geography shift. This is the same trap as last quarter, when the gap was five points, and it is getting wider.

The second observation is the risk-weighted asset line. Customer lending grew €17.6bn in the quarter alone and the loan book is up substantially year-on-year, yet risk-weighted assets rose only 1.8% over twelve months and fell €2.4bn over three months. That is the capital-velocity story management has been telling since the 2024 capital markets day, and it worked. The question the call did not settle is how much of it repeats.

Quarter-Over-Quarter Comparison

The sequential comparison is where the volatile-item swing and the regulatory-cost seasonality both live, and both flatter the second quarter. The table is shown as reported, with the adjustment noted underneath.

Three of the sequential moves are mechanical rather than operational. Regulatory costs drop €246M because ING books annual bank taxes and deposit-guarantee contributions in full in the first quarter. Volatile items swing €143M in the bank's own reconciliation. And the €299M jump in other income is largely the same swing viewed from the other side: hedge ineffectiveness moved from negative €81M to positive €91M as market volatility subsided.

What is left after those adjustments is still good. Commercial net interest income up 2.8%, fee income up 3.4%, and the liability margin up another 3bp. That is a genuine sequential operating improvement of the same order as the first quarter's. It is not a 29% improvement in profit before tax.

First-Half View

The six-month figures matter more than usual this quarter because the full-year guidance is now anchored on them, and because the half-year strips out the first quarter's regulatory-cost distortion.

Two half-year numbers are worth carrying into the guidance section. Total income of €12,107M against a full-year floor of €24.5bn means the second half needs only €12,393M, or roughly €6,197M per quarter, which is below the €6,284M just printed. And the first-half return on tangible equity of 15.3% already exceeds the full-year floor of "above 15%", which means the guide implies a weaker second half than the first. Both of those are floors set to be cleared, not stretch targets.

Segment Performance

ING reports three segments: Retail Banking, which is roughly 70% of group income and is disclosed across four geographic units; Wholesale Banking, disclosed across four product lines; and a Corporate Line that carries foreign-currency hedging, financial stakes and unallocated items. The Corporate Line swung to a loss this quarter and is the reason group profit before tax is smaller than the sum of the two operating segments.

Segment return on equity is calculated by the company on equity set at 13.0% of that segment's risk-weighted assets, so it is a capital-normalised measure and not directly comparable to the group's 17.0% return on tangible equity. Corporate Line profit before tax swung from positive €55M a year ago to negative €118M, a €173M drag, which is why group profit before tax of €2,919M is €118M below the €3,037M sum of the two operating segments.

Retail Banking

Retail delivered the quarter. Income of €4,369M grew 14.0% year-on-year against the group's 10.2%, profit before tax grew 31.3%, and the capital-normalised return on equity reached 27.0% against 22.4% a year ago. Commercial net interest income rose 11.0% and fee income 15.5%, both faster than the group. The cost/income ratio of 45.9% is more than three points better than the group's, and 3.7 points better than Wholesale's on the same basis.

The volume story is the strongest part. Net core lending grew €12.1bn, including €7.1bn of residential mortgages concentrated in the Netherlands, Germany, Italy and Australia, with business lending and consumer lending adding the rest. Net core deposits grew €16.7bn, more than the group total because Wholesale deposits declined.

"In Retail Banking, lending has grown by €12.1 billion, or 9% on an annualised basis. We have helped more people finance their homes, leading to a €7.1 billion growth in mortgages, especially in the Netherlands, Germany, Italy and Australia."
— Steven van Rijswijk, CEO

Unit figures are as filed and carry the filing's own rounding, so the four units sum to €2,232M of profit before tax against a €2,233M segment total, and to €16.8bn of deposit growth against €16.7bn. The totals are the reported figures.

Retail Netherlands

A 38.2% return on equity and a 34.1% cost/income ratio in the home market, on income up 14.1% and profit before tax up 32.6%. Two items inside that are worth isolating. Risk costs were literally zero basis points because a partial release of the interest-only mortgage overlay almost exactly offset new provisions, and other income of €160M includes a €22M gain on the sale of an equity stake. Operating expenses actually declined year-on-year despite one-off marketing costs tied to the FIFA World Cup, because lower external staffing offset collective labour agreement salary increases, and €10M of restructuring costs were offset by a €10M legal-provision release.

Assessment: The Dutch franchise is running at a level that is hard to improve on and easy to give back. A zero-basis-point provision charge is not a run-rate. Stripping the €22M equity-stake gain and charging risk costs at the group's 15bp on the unit's €188.9bn book, rather than at zero, would take roughly €90M off the €947M. Profit before tax would still have grown about 20% year-on-year, so the quality is real; the precise level is not repeatable.

Retail Germany

The most interesting unit this quarter. Income grew 14.6% and deposits grew €7.8bn, nearly half the group's entire net core deposit inflow, against a €1.1bn outflow in the year-ago quarter. Management was direct that this was campaign-driven and specifically a below-the-line campaign targeted at existing customers rather than a teaser-rate land grab. Fee income declined sequentially as investment trading activity normalised after an unusually volatile first quarter. Expenses carried €19M of restructuring related to a workforce reduction at Interhyp.

"In Germany, we did a below-the-line campaign. So that's a campaign to existing customers, whereby we then do fresh money campaigns to which also, therefore, increase the deposits over there."
— Steven van Rijswijk, CEO

Assessment: Germany is the clearest evidence for the bull case and the clearest cost of it. Gathering €7.8bn in a quarter demonstrates the franchise pulls deposits when it chooses to, which underwrites the volume half of the net interest income guide. It also explains why the liability margin only rose 3bp when the replication tailwind alone should have delivered more: campaign costs are the offset, and management declined to quantify the marginal pass-through assumption behind the margin guide.

Retail Belgium and Retail Other

Belgium's 47.3% profit growth is the largest in the group and is substantially a regulatory-cost artefact: the Belgian bank tax and deposit-guarantee contribution are booked in the first quarter, so second-quarter regulatory costs were negative €5M against €182M in the first quarter. Underlying, commercial net interest income was flat sequentially and fee income grew strongly on structured-note issuance and investment entry fees. Deposits fell slightly on tax outflows.

Retail Other, which covers Spain, Italy, Australia, Poland, Romania and Türkiye, grew income 13.2% and lending €3.3bn with €2.3bn of that in mortgages. It carries the group's most consequential portfolio change this quarter: ING acquired the remaining 55% of Goldman Sachs TFI in Poland, which brought a €25M positive fair-value revaluation of the existing stake, a Polish dividend in investment income, a reclassification of fee income out of other income, and roughly €30M of incremental cost this year. Expenses also carried €22M of legal provisions.

Assessment: Retail Other has the weakest cost/income ratio and the weakest return on equity of the four units, and it is now the unit where inorganic activity concentrates. The TFI consolidation is small but it is the second acquisition-driven distortion in two quarters, and with the Singular Bank stake in Spain closing in the first quarter of 2027 there will be a third. Analysts should expect the "consolidation of TFI" phrase to appear in the year-on-year cost bridge for four more quarters.

Wholesale Banking

Wholesale grew income 10.9% year-on-year with expenses down 6.0%, producing 34.9% gross-result growth and a 49.6% cost/income ratio against 58.5% a year ago. The expense decline is not underlying: restructuring costs were €90M in 2Q2025 and nil this quarter, and management gave underlying cost growth of 2.4% year-on-year. Income over average risk-weighted assets reached 513bp, up from 466bp in the first quarter and 454bp a year ago, and above 500bp for the first time in the five periods this release discloses.

Product lines are as filed and sum to €1,878M against the reported €1,879M total; the total is the income-statement figure.

Risk-weighted assets fell €5.3bn inside the segment despite €3.0bn of net core lending growth, which is the capital-velocity engine working. But the two fastest-growing lines are also the two least predictable: Financial Markets grew 13% on "more favourable market conditions" and Treasury & Other grew 61.7% sequentially on €18M of one-off gains plus the reversal of the first quarter's hedge ineffectiveness.

Assessment: Wholesale's return on equity of 12.5%, computed on equity set at 13.0% of its own risk-weighted assets, is less than half Retail's 27.0% on the identical basis and remains the group's structural drag. The improvement is real and the direction is right, but two of the four product lines carry the volatile income, and the risk-cost line moved the wrong way this quarter: 27bp against 12bp in the first quarter and 19bp a year ago, on "increases to provisions for a limited number of Stage 3 files and the impact of a weaker economic outlook." Wholesale is now simultaneously the best capital story and the worst credit story in the group.

Key Performance Indicators

Two of these deserve more weight than they usually get. Assets under management of €322bn, up 27% year-on-year, is the engine behind the fee-income upgrade and it is compounding faster than any other balance in the group. And the Stage 3 coverage ratio has now fallen in two consecutive quarters, from 34.4% at the end of 2025 to 33.5% and now 33.3%, while the Stage 3 book itself grew from €13,402M to €13,652M. Coverage falling while the impaired book grows is a slow-moving disclosure, and it is not one the call discussed.

Key Topics & Management Commentary

Overall Management Tone: Management was more confident than at any point in the four quarters we have reviewed, and unusually willing to be pinned to numbers: two years of guidance were raised on the same slide, a 2027 and 2028 liability-margin figure was volunteered, and a segment capital-allocation target was declared met eighteen months early. The one topic where the answers turned framework-level rather than quantitative was capital: the composition of a €2.8bn model-driven risk-weighted-asset release was declined, and the marginal deposit pass-through assumption behind the margin guide was declined. Confidence and disclosure moved in opposite directions on the same subject, which is the quarter's most useful tell.

1. The Income Beat Came From the One Line Management Guides Down

All other income of €832M beat the poll by €132M and cleared its range top by €62M. It is 83% of the total income beat. On the bank's own reconciliation, €88M of it is volatile items: hedge ineffectiveness of positive €91M reversing the first quarter's negative €81M, €15M of capital gains, €8M of other, offset by −€25M of Financial Markets valuation adjustments. The remainder is a €25M revaluation of the pre-existing Goldman Sachs TFI stake, a €22M equity-stake sale gain in the Netherlands, €18M of Wholesale one-off gains, and a €26M Van Lanschot Kempen dividend.

Management did not present it as a beat. The chief financial officer walked the line down rather than up, and was explicit that the underlying trend excluding the hedging reversal is negative.

"Year-on-year, when excluding for positive results from hedging ineffectiveness, all other income decreased. This is largely due to lower results from foreign currency exchange hedging in treasury, where the benefit from interest rate differentials between our main currencies has gradually come down over the past 12 months."
— Ida Lerner, CFO

The full-year guide for the line was left unchanged at €2.5–2.7bn. First-half all other income was €1,360M, so the second half is guided to €1,140–1,340M, or €570–670M per quarter against the €832M just printed. Management is telling the market this line falls roughly 20% to 31% from here.

Assessment: Every euro of the income beat that cleared the range top is in a line the bank has told you will be materially lower next quarter. On the bank's own five-quarter table this line has moved income by at least €71M in every one of the last five quarters, in both directions: −€110M, +€77M, +€75M, −€71M and now +€88M. Anyone extrapolating the 10.2% income growth rate is extrapolating hedge accounting.

2. Commercial Net Interest Income Landed Exactly on the Poll Median

The recurring engine printed €4,174M against a company-compiled median of €4,174M. The broker range ran from €4,115M to €4,227M, so the result sat at the fiftieth percentile of fifteen submissions to the million. Sequentially the line grew €114M: €97M from liability net interest income on higher deposit volumes and a 3bp better margin, and €16M from lending net interest income where 8% annualised volume growth was partly given back by a 2bp margin decline.

That decomposition is the quarter's operating truth. Volume growth was outstanding, margin was steady, and the combination produced precisely the number the Street had modelled. The full-year guide then went up, which is the more meaningful signal than the quarter itself.

"Looking ahead, on the back of a very strong first half of the year, we expect a higher level of commercial NII than previously guided for the full year. We now expect commercial NII for the full year to be between EUR 16.8 billion and EUR 17 billion."
— Ida Lerner, CFO

Assessment: An exact match on the most-forecast line in the model tells you the market understood this business going into the print. It also means the 5.1% move in Amsterdam was paid for the guidance, not the quarter. The new range of €16.8–17.0bn against €8,233M delivered in the first half requires the two remaining quarters to average €4,284–4,384M, a 2.6% to 5.0% step up from the second quarter, with the liability margin guided only into the upper-middle of a 100–110bp band. Volume has to carry most of that, and volume is the part management says will eventually normalise toward 5%.

3. The Liability Margin Stepped to 107bp, and 2027 Is Now Guided Above 110bp

This is the pillar the April initiation was built on, and it delivered. The liability margin rose another 3bp after a 5bp step in the first quarter, and the full-year guide moved from the "mid-part" of 100–110bp to the upper-middle of the same range. The forward disclosure is the genuinely new information.

"As previously indicated and what we also continue to say today is that we expect the liability margin to be above 110 basis points in '27 and '28. So slightly higher than what we expect it to be coming out of 2026."
— Ida Lerner, CFO

The mechanism is unchanged: roughly 55% of the retail eurozone replicating portfolio has an average remaining maturity between one and fifteen years, so a large share of the liability margin reprices on contracts rather than on the policy rate. The deposit-side offset is now visible in the numbers. Actual average pass-through in the quarter was around 42%, equivalent to roughly 86bp of total deposit costs, and around 109bp on savings and term deposits alone. The bank quantifies the sensitivity: every 10bp of pass-through on total savings and term deposits is worth roughly €0.4bn of commercial net interest income.

Assessment: A contracted margin tailwind guided above 110bp two years out, disclosed alongside a €0.4bn-per-10bp pass-through sensitivity, is a strong and unusually specific disclosure. It is also now largely in the price. The margin path being guided is the same path the April note argued was being withheld from guidance; management has now put it in guidance, and the shares have moved 22% since.

4. The Upgrade Raises 2026 Numbers and Ratifies 2027 Ones

Six outlook lines moved across two years. Set them against the pre-print poll and they do not all mean the same thing.

On 2026, both the income floor and the return floor now sit above where the fifteen brokers were, so consensus has to move. On 2027, the income floor sits €235M below the poll median, the fee range brackets it, and the return floor equals it exactly. ING has raised its published 2027 floor to the level the Street was already carrying. That is a real event, because a floor that consensus already exceeds converts an estimate into a commitment and removes downside. It is not, however, an increase in 2027 numbers.

"And as a reflection of strong and disciplined execution of our strategy, we are upgrading our ROTE outlook by 1 percentage point, now expecting an ROTE of more than 15% in 2026 and more than 16% in 2027."
— Steven van Rijswijk, CEO

Assessment: The market treated a two-year upgrade as two years of estimate revisions. One year of it is. The 2027 move is de-risking, which is worth something to a multiple but not to an earnings number, and the shares rallied 5.1% on the day.

5. Fee Income Reached the €5bn Mark a Year Early

Fee income of €1,278M grew 13.9% year-on-year with Retail up 15.5% and Wholesale up 11.4%, and the full-year target has been pulled forward.

"For the full year, we expect to generate EUR 5 billion in fee income, which is up EUR 400 million year-on-year and implies that we will reach our EUR 5 billion target 1 year ahead of plan."
— Ida Lerner, CFO

The driver is assets under management, up 27% year-on-year to €322bn, with 110,000 investment-account customers added in the quarter to roughly 5.3 million and €21bn of net inflows over twelve months. Part of the growth is inorganic: the full consolidation of Goldman Sachs TFI in Poland contributes to both the assets-under-management figure and Retail Other fee income, alongside a structural reclassification from other income into fees from 2026.

Assessment: The most durable good news in the print, and the pillar that has strengthened most since April. Note the arithmetic, though: first-half fees of €2,514M against a ~€5.0bn full-year target implies second-half fees of roughly €2,486M, or about €1,243M per quarter, which is below the €1,278M just printed. As with income, the upgraded target is a floor, and it is a floor the consensus median of €4,986M already matched.

6. Underlying Costs Accelerated to 4.2%, and the Second-Half Guide Requires More

This was the open question from the first quarter and it has moved, though not in the direction the bull case wanted. Operating expenses excluding regulatory costs and incidental items were €2,961M, up 4.2% year-on-year against 1.1% last quarter and 3.4% sequentially. Management's framing was that the year-to-date figure is what matters.

"On a year-to-date basis, our cost growth is tracking at 2.7%, which is well in line with our previously communicated full year outlook."
— Ida Lerner, CFO

The full-year outlook of €12.6–12.8bn was reiterated, on a basis that excludes incidental items booked after the first quarter. First-half operating expenses were €6,305M, of which €46M were second-quarter incidental items, so the first half on the guided basis is €6,259M and the second half is €6,341–6,541M. That is €3,171–3,271M per quarter against €3,040M in the second quarter on the same basis, a 4.3% to 7.6% sequential step. And because ING books its annual regulatory charges in the first quarter, almost none of that step can come from regulatory costs.

Management also flagged that the treatment of incidental items has changed at the margin.

"Incidental items in the second quarter and those that may be booked in the subsequent quarters will be incremental to the full year outlook."
— Ida Lerner, CFO

Assessment: The cost guide is the only line in the outlook table that was not raised, and it is the line whose arithmetic is hardest to reconcile with the run-rate. Either the second half carries a genuine step in underlying spend that has not been explained, or the guide is conservative by roughly €200–400M and the return on tangible equity floor is correspondingly soft. Both readings are plausible; the point is that no one on the call asked which it is. Not one of the nine questioners raised the expense line in a quarter where underlying cost growth went from 1.1% to 4.2%.

7. Risk Costs of 15bp Are the Group Average of Two Opposite Moves

The group charge of €279M, or 15bp of average customer lending, is well below the 20bp through-the-cycle average, 4bp below the median of fifteen broker estimates and 2bp below the lowest of them. Underneath it, the two operating segments moved in opposite directions and both moves were non-recurring in character.

Retail fell to 10bp because of non-performing loan sales in several countries and a partial release of the Dutch interest-only mortgage overlay following a comprehensive review. Wholesale rose to 27bp on "increases to provisions for a limited number of Stage 3 files and the impact of a weaker economic outlook," plus charges related to significant risk transfers and credit and political risk insurance which the bank notes are reflected in risk costs going forward. Total Stage 1 and Stage 2 charges for the group were €8M; essentially the entire €279M is Stage 3.

Assessment: Last quarter's concern was that Retail credit was drifting while the group number stayed benign. This quarter the drift moved segments rather than stopping. A 15bp print built on asset sales and an overlay release in one segment and a weaker economic outlook in the other is not a 15bp run-rate, and the balance-sheet ratios agree: Stage 2 outstandings rose €2.8bn to 8.2% of the book, Stage 3 rose to 1.6%, and coverage fell again to 33.3%.

8. The Middle East Overlay Was Released and Immediately Replaced

The first quarter's €94M management overlay for the war in the Middle East was the subject of an explicit forward commitment on the April call: that from the second quarter the process would revert to standard macro-consensus-driven provisioning and the overlay would diminish. The release happened. The reversion did not.

"The €94 million management overlay recorded in 1Q2026 to address the possible impacts of the war in the Middle East was released in full and replaced by both the impact from updated macroeconomic forecasts and an additional sector-based overlay. This reflects continued uncertainty regarding a credible path to a structural resolution of the conflict and the resulting second-order effects on vulnerable sectors."
— 2Q2026 results release

No size was given for the new sector-based overlay, no sectors were named, and no analyst asked. Separately, a partial release of the Dutch interest-only mortgage overlay flowed through Retail, so the quarter contains at least three discretionary overlay movements with only one of them quantified.

Assessment: Grading the commitment honestly: half met. The named overlay is gone and the disclosure is candid about what replaced it, which is more than many banks offer. But an unquantified sector overlay substituting for a quantified geopolitical one leaves the provision line exactly as discretionary as it was, in a quarter where that line supplied 43% of the beat.

9. CET1 Rose on Model Updates Management Will Not Forecast

The CET1 ratio rose 8bp to 13.1% while 100% of the quarter's €1,947M net profit was reserved outside CET1 capital. Available CET1 capital was therefore essentially flat, at €44,692M against €44,729M. The entire ratio improvement is the denominator: risk-weighted assets fell €2.4bn to €341.9bn. Inside that decline, market risk-weighted assets fell €2.2bn, operational was flat, and credit fell €0.6bn excluding a €0.5bn currency effect, even as customer lending grew €17.6bn.

The credit line only works because of two items, one of which management sized only under direct questioning.

"As previously mentioned by Steven, the SRT that we did related to our wholesale banking portfolio in Germany gave approximately EUR 1 billion of relief. In addition to that, we have model updates, which is generating EUR 2.8 billion of release. Apart from that, we don't give any granular details, but you can also see overall that there is a positive development on risk-weighted assets overall."
— Ida Lerner, CFO

Pressed on whether the model updates were part of a rolling regulatory programme or company-specific optimisation, the answer was that ING will not guide the line and that it can move either way: "that could also go in a positive direction, but it could also be in a negative direction depending on this." Meanwhile the significant risk transfer programme is running behind its own schedule.

"So in terms of the capital, we've now done 4 basis points of SRTs. We said for the year, we would do 15 to 20 basis points in capital improvement, which will largely come from wholesale Banking. So there's still quite a bit to go."
— Steven van Rijswijk, CEO

Assessment: This is the bear point from April, confirmed and sharpened. Capital no longer builds visibly from retained profit, so the ratio depends on risk-weighted-asset relief; the largest single source of relief this quarter was an internal model update the bank will neither decompose nor forecast and which it warns can reverse; and the one source of relief that is committed and scheduled is a quarter delivered at the half-year mark. There is nothing improper here. There is also nothing to extrapolate.

10. Capital Allocation Crossed the 2027 Retail Target Eighteen Months Early

At the June 2024 capital markets day ING set out to shift group capital allocation from 50/50 between Retail and Wholesale to 55/45 in favour of Retail by the end of 2027. That target is met.

"Then we said it would be 55%, 45% in '27. And now we are 56%, 44% for retail mid-'26, so we're 18 months ahead."
— Steven van Rijswijk, CEO

The mechanism was Wholesale risk-weighted assets falling €4.6bn year-on-year while its lending book and revenues grew, through secondary loan sales, insurance, portfolio optimisation and significant risk transfers. Retail risk-weighted assets rose 6.8% year-on-year to €181.1bn while Wholesale fell 3.1% to €143.9bn.

Assessment: A genuine strategic achievement, and the clean explanation for why group return on tangible equity has moved 270bp year-on-year without any single line doing anything spectacular: capital moved from a 12.5% return business to a 27.0% return business. It also raises the obvious question the call did not put: what is the target now? Meeting a 2027 goal in mid-2026 leaves the mix optimisation without a published destination.

11. Building the Third Retail Pillar, and the Subscription Layer Underneath It

Two income-diversification initiatives were given real airtime. The first is private banking as a third retail pillar alongside private individuals and business banking, launched in Italy, with Spain next and supported by a roughly 40% stake in the Spanish wealth manager Singular Bank, expected to close in the first quarter of 2027.

"And that's why we bought the 40% with an option we said already in the press release to buy the total at a later point in time."
— Steven van Rijswijk, CEO

The second is a global subscription model, replacing product-by-product pricing with bundled tiers spanning daily banking, protection and lifestyle services, launched in four markets.

"So we used a subscription -- 4 subscription packages in a number of our markets earlier this year. And to date, 17 million customers have been migrated and by default, customers migrated to an equivalent package."
— Steven van Rijswijk, CEO

Migration is at scale; monetisation is not. Management said upselling requires time, that initial pricing incentives are in place, and that fee benefits should begin to come through later in the year, with "a number of thousands of people" having moved to higher tiers so far.

Assessment: Both are the right shape for a bank whose income is still roughly two-thirds interest-linked, and both are years from mattering to earnings. Seventeen million customers migrated onto a subscription architecture with a few thousand upsells is a distribution asset, not yet a revenue line. The Singular structure, a minority stake with an option over the balance, is the cautious version of the deal and appropriately so, but it also means the Spanish wealth ambition will not consolidate revenue before 2027.

Guidance & Outlook

The outlook excludes the potential impact of the intended exit from Russia and any further incidental items, both of which were also carved out of the prior guide. Incidental cost items booked from the second quarter onward are explicitly incremental to the operating expense range rather than absorbed within it, which is a change in treatment relative to the €30M of first-quarter incidentals that the guide already contained.

Implied second-half ramp. Working from first-half actuals, the guide requires the following of the remaining two quarters:

Operating expenses on the guide basis exclude the €46M of second-quarter incidental items from the €6,305M first-half reported figure and from the €3,086M second-quarter figure.

Street at: the pre-print poll median had 2026 total income at €24,385M, so the >€24.5bn floor is roughly 0.5% above it; 2026 return on tangible equity at 14.5% against a >15% floor; 2027 total income at €26,235M against a >€26bn floor, and 2027 return on tangible equity at 16.0% against a >16% floor. Net: the 2026 lines require upward revision, the 2027 lines do not.

Guidance style: conservative, and consistently so. Every element of the 2026 guide implies a second half at or below the second quarter's run-rate except commercial net interest income, where volume growth has to carry a 2.6% to 5.0% sequential step with the liability margin guided only into the upper-middle of its band, and operating expenses, where the required step is upward and unexplained. A guide whose return-on-tangible-equity floor implies a weaker second half than the first, in a bank that has beaten and raised twice running, is a floor rather than a forecast. That is a compliment to management and a warning to anyone valuing the shares off the guidance as if it were the base case.

Analyst Q&A Highlights

Nine questioners and a narrow distribution of subjects. The liability margin was raised three separate times, capital and risk-weighted assets three times, and commercial momentum or deposit gathering three times. Nobody asked about operating expenses, the second-half cost step, the new sector-based credit overlay, the Dutch mortgage-floor release scheduled for December, or the Russia exit.

Whether the Liability Margin Keeps Stepping at This Pace

The first question of the call went straight to the pillar the whole equity story rests on: is 3bp a quarter the new run-rate, or does deposit-campaign competition cap it from here. The answer was a clear number for the current quarter, a raised full-year band, and an explicit refusal to characterise campaign activity going forward beyond calling the quarter a normalisation from an unusually quiet first quarter.

Q: "Maybe you can give a bit more color on the liability margin going forward now with the deposit campaigns. Should we expect a more modest increase in liability margin going forward? Or is that 3 basis points a good momentum given the tailwinds you have in the replicating portfolio?"
— Benjamin Goy, Deutsche Bank

A: "As you noted, the liability margin increased by 3 basis points in the quarter and is now at 107 basis points. This reflects a disciplined deposit pricing and also, of course, a continued benefit from the replication portfolio and the tailwind that we already started to see in the second half of last year and continues to see now."
— Ida Lerner, CFO

Assessment: Answered on the level, deflected on the trajectory. The useful part came later in the same response, where the full-year band was placed in the upper-middle of 100–110bp and 2027 and 2028 were put above 110bp before normalising back toward historical levels. That last clause is the one to keep: the tailwind is described as a multi-year bulge, not a permanent re-rating of the deposit franchise.

Whether the Deposit Inflow Is Franchise or Purchased

Sixteen billion euros of net core deposit growth in a quarter invites the obvious question of what it cost and who it came from. Management's answer was unusually specific on mechanics and unusually candid on attrition, disclosing a two-thirds retention assumption on teaser-rate money, while declining to be drawn on the pricing behind it.

Q: "I'd just like to ask about the nature of some of the deposit growth you've seen in the quarter, which has been really quite strong, particularly Germany. Is it largely sort of new-to-bank customers?"
— Shrey Srivastava, Citi

A: "And typically, we say when we do a campaign, 2/3 of the fresh money will stay and 1/3 will flow out after the campaign ends. If you look at existing customers, that was the below-the-line campaign that we did this time around in Germany, those fresh money campaigns are a tool to increase the share of wallet and then we give attractive retention rates and short payback periods."
— Steven van Rijswijk, CEO

Assessment: A genuinely informative disclosure, and one that cuts both ways. A one-third run-off assumption on campaign money means the €15.9bn of net core deposit growth is not €15.9bn of durable funding, and the German inflow was the largest single component. It also explains the gap between the replication tailwind and the modest 3bp margin step: the campaign cost is the toll paid to move the volume that carries the net interest income guide.

Whether the Return Ambition Now Extends Beyond 2027

With the 2027 target raised and the capital-allocation goal met eighteen months early, the natural question is what the medium-term destination is and when it will be published. This was the one exchange where an explicit forward commitment was requested and declined without a substitute.

Q: "But Steven, you are already ahead of a few targets of the previous Capital Markets Day, the capital allocation, the profitability. So -- and European banks in general are approaching ROTEs closer to 20%. So can we start dreaming about high teens ROTEs, especially as we look into 2028? And when can we hear about your midterm ambitions next?"
— Giulia Miotto, Morgan Stanley

A: "When you talk about the ROTE outlook, that's why I started to smile. Yes, look, of course, we updated it. I think what we're doing is very good. You see that the machine is humming. And that's why we are able to update the outlook for '26 and '27. And like I also said in the presentation, we keep on working also in the years thereafter to further increase our ROTE. More to come about that at a later point in time."
— Steven van Rijswijk, CEO

Assessment: A friendly non-answer, and the most consequential one on the call. The stock is being valued on a return trajectory that extends past the guided horizon, and there is currently no published target beyond 2027 and no date for one. For a bank that has just met a three-year capital-allocation goal in eighteen months, the absence of a refreshed medium-term framework is the gap between what the shares are discounting and what management has committed to.

Whether 8% Balance Growth Is Sustainable or a Pull-Forward

Lending and deposits have both been compounding at more than 8% annualised against a stated medium-term expectation of 4–5%, and a recurring line of questioning pressed whether the guide is simply too conservative. The answer separated the segments and put a time frame on normalisation.

Q: "So the first one is on the sustainability of this very strong commercial momentum. You are growing lending and deposits more than 8% for quite some time actually. So can we expect your 4% to 5% range to be conservative? And do you think you can sustainably grow more than 4% to 5%?"
— Benoit Petrarque, Kepler Cheuvreux

A: "And at some point, in a longer-term time, we believe that lending and deposit growth will hover around the 5%. But in the shorter term, we believe these will remain at elevated levels."
— Steven van Rijswijk, CEO

Assessment: Straight and useful. Growth stays elevated near term and converges to about 5% eventually, with the mix skewing to Retail because Wholesale is described as more cyclical. That is the right assumption set for a model, and it matters more than it sounds: with the liability margin guided to normalise after 2028 and lending margins drifting down on mix, volume is what has to carry commercial net interest income in the out-years, and management has just told you volume halves in growth rate.

Whether the Lending Margin Decline Signals Price Competition

The 2bp sequential fall in the lending margin was the one negative in the interest-margin disclosure, and the question was whether it reflects a deliberate mix choice or competitive pressure. The answer was mix, in both segments, with an explicit statement that it was not a decision.

Q: "there's a lot of focus on the liability margin, but the lending margin deterioration in the second quarter. I just wanted to ask if this was a conscious business decision, i.e., to go into lower margin, higher ROE business."
— Namita Samtani, Barclays

A: "I think on the lending margin, that was not a conscious business decision. What you are seeing is that we continue to grow mortgages at a rapid pace, which is lower risk, lower RWA and also lower margin activity compared to other parts of the loan book. And within Wholesale Banking, there was also a shift to higher investment-grade loans."
— Steven van Rijswijk, CEO

Assessment: Reassuring on price and quietly important on structure. Growth is concentrating in low-risk-density assets, which is why risk-weighted assets barely moved against €17.6bn of lending growth and why return on equity keeps improving. The trade is a permanently lower lending margin for permanently lower capital intensity, and management guided the margin to hover around this level for the rest of the year.

Why the Risk-Weighted Asset Decline Cannot Be Modelled

The most probing exchange of the call. A €2.4bn sequential fall in risk-weighted assets alongside strong volume growth is the single largest driver of the capital ratio, and the request was for a decomposition of model updates and the Thai stake reduction, and for whether this is a rolling regulatory programme or company-specific optimisation.

Q: "All on the RWA side, which is really strong at EUR 2.4 billion Q-on-Q despite good volumes. I just wondered if you could break out the benefits from model updates and TMB within the Q-on-Q delta. And in particular, what drove those favorable model updates in terms of product or business line?"
— Farquhar Murray, Autonomous Research

A: "We continuously update our model portfolio and also in dialogue with ECB, and that could also go in a positive direction, but it could also be in a negative direction depending on this. So we're not giving any guidance in terms of future potential on the model side."
— Ida Lerner, CFO

Assessment: The question extracted a number, €2.8bn of model-driven release, that would otherwise not have been disclosed, and then hit a wall on composition and repeatability. Both halves are informative. The largest driver of this quarter's capital improvement is a line the bank will not decompose, will not forecast, and warns can reverse. That is an honest answer and it is also a reason to give the capital ratio less credit than the headline 8bp improvement invites.

Whether Reserving 100% of Profit Signals a Higher Payout

ING now reserves the full quarterly net result outside CET1 capital while its stated dividend policy is a 50% payout. The question was whether the reserving change is a regulatory requirement or a signal of intent, and the answer restored the distinction cleanly.

Q: "So you are accruing 100% of earnings. Your dividend policy is 50%. But because you pay those extra distribution, are you then required the ECB to accrue 100%, but doesn't mean you would pay 100%, you adjusted full year or that means you're actually intending to distribute 100%."
— An analyst from Bank of America, name indistinct on the call recording

A: "As you might remember, in the first quarter, we changed our reserving policy also to be in line with EBA guidelines. So as of the first quarter 2026, we reserve both our regular 50% dividend payout policy and potential additional distribution outside the core equity Tier 1. There is no change to our dividend policy."
— Ida Lerner, CFO

Assessment: The right answer to an important question. Full reservation is compliance, not a promise, and the distribution framework is unchanged: 50% of net profit as ordinary dividend, then profitable growth, then structural excess capital above a 13% CET1 ratio returned to shareholders. The practical consequence is that the reported CET1 ratio no longer signals distribution capacity the way it did before 2026, and the next actual decision point is the third-quarter results in October.

Whether Growing Reliance on Risk Transfers Changes the Credit Profile

The final question was the sharpest framing of the capital-velocity strategy on the call: if the bank increasingly places risk with third parties rather than holding it, does the through-the-cycle cost of risk assumption still hold, and does the retained earnings profile in a downturn change.

Q: "SRT is becoming more structural for you. You've done very little in the past. Now you started doing more and more SRT. So you're effectively renting out risk that you used to hold. Does that change the through-the-cycle cost of risk we should assume or the earnings you keep in a downturn?"
— Alberto Cordara, Intesa Sanpaolo

A: "it is not our intention to change materially our risk appetite or underwriting standards on the back of the externalization of the risk through SRT. So we don't expect any material impact resulting from SRT other than the capital optimization on our cost of risk."
— Andrea Cesaroni, Chief Risk Officer

Assessment: The answer addresses underwriting standards, which was not quite the question. Whether risk appetite changes is separate from whether the economics change, and they demonstrably do: this quarter's Wholesale risk-cost line already includes charges related to significant risk transfers and credit and political risk insurance, which the release notes are reflected in risk costs going forward. Placing risk has a running cost that lands in the provision line while the capital benefit lands in the ratio. That is a good trade at 27bp of Wholesale risk costs; it is a worse one in a downturn, which is exactly what was asked and not answered.

What They're NOT Saying

1. Nobody asked about costs, and the guide requires a step nobody explained. Underlying operating expenses accelerated from 1.1% to 4.2% year-on-year growth, and the reiterated full-year range implies €3,171–3,271M per quarter in the second half against €3,040M in the second, with almost no regulatory costs available to absorb it. Management characterised year-to-date cost growth of 2.7% as "well in line" and no one tested the phasing, for the second consecutive quarter.
2. The new sector-based credit overlay was not sized and its sectors were not named. The €94M Middle East overlay was released in full and replaced by updated macro forecasts plus an additional sector overlay. In a quarter where the provision line supplied 43% of the profit-before-tax beat, the discretionary component of that line is unquantified.
3. The Dutch mortgage floor is not mentioned once. The April call identified the 1 December 2026 expiry of the Dutch mortgage risk-weight floor as roughly 15bp of CET1 relief on about €4bn of risk-weighted assets, to be assessed in the October distribution review. It appears nowhere in this release or this call, and no analyst raised it, three months closer to the date.
4. There is no published ambition beyond 2027. The 2024 capital-markets-day capital-allocation target has been met eighteen months early and both guided years have been raised. Asked directly about 2028 and about when medium-term ambitions would be refreshed, management offered "More to come about that at a later point in time" with no date and no capital-markets-day announcement.
5. The composition of €2.8bn of model-driven risk-weighted-asset relief is withheld. Its size was disclosed only under direct questioning, its product and business-line drivers were declined, and forward guidance on the line was explicitly refused with a warning it can move negatively.
6. The marginal deposit pass-through assumption behind the margin guide is withheld. Asked whether the roughly 100% marginal pass-through assumption still applies to coming policy moves, the answer was that ING does "not provide insight in terms of our estimates around pass-through rates." The bank does disclose actual second-quarter pass-through of around 42%, and a sensitivity of roughly €0.4bn of commercial net interest income per 10bp on savings and term deposits, so the sensitivity is knowable while the assumption is not.
7. The Singular Bank stake carries no price and no financial impact. A roughly 40% stake in a Spanish wealth manager, with an option over the balance, closing in the first quarter of 2027, and no consideration, no expected contribution and no return metrics disclosed.
8. The Russia exit remains carved out of guidance with no timing update. Both the 2026 and 2027 outlooks explicitly exclude the potential impact of the intended exit, unchanged from the prior quarter, and it was not discussed on the call.

Market Reaction

ING's primary listing is Euronext Amsterdam and the New York line is a depositary receipt where one receipt equals one ordinary share. The two legs moved differently on the day and both are shown, because the Amsterdam move is the equity move and the New York move carries the currency on top of it.

• Pre-print setup (Amsterdam): INGA closed at €28.75 on 29 July, having run from €28.95 a week earlier through an intraday peak of €29.54, then given back 1.8% over the two sessions immediately before the print. The New York line closed at $32.51, up 16.1% year to date against 6.9% for the S&P 500, up 38.9% over twelve months, and up 3.6% over the trailing thirty days. Entering the print the receipt sat inside a 52-week closing range of $22.71 to $33.49, roughly 3% below its own high.
• Reaction session, Amsterdam (30 July): opened at €29.25, a 1.7% gap, traded to an intraday high of €30.36 and closed at €30.215, up 5.10%. The session low of €28.935 was above the prior close, so the shares never traded down after the open. Volume of 9.6M shares was 1.9 times the average of the prior four sessions.
• Reaction session, New York (30 July): opened at $34.50, a 6.1% gap, and closed at $34.94, up 7.47%, on 2.9M receipts against a 3.4M thirty-day average. The 2.4 point gap to Amsterdam is currency and the clock: the euro appreciated roughly 2.3% against the dollar across the two closes, and Amsterdam finished at 17:30 CET while New York ran to 16:00 ET.
• Index comparison: the AEX rose 1.06% and the S&P 500 rose 1.7% on the same session. ING outperformed its home index by 4.0 points.

The shape of the move is more informative than its size. This was not a gap-and-fade or a gap-and-hold. Amsterdam opened up 1.7%, roughly what a €84M net-result beat on its own is worth, and then added a further 3.4 points through the session as the market worked through the guidance slide and the call. The bulk of the day's return was paid after the numbers were already public, which is consistent with the guidance being the event and the quarter being the occasion.

That reading matters for what was actually bought. A market reacting to the printed quarter would have had to underwrite a €132M beat on the mark-to-market line and a provision charge 2bp below the lowest broker estimate, neither of which is a durable earnings input. A market reacting to the guidance was buying a 1 point uplift to the return-on-tangible-equity floor in each of two years, a commercial net interest income range raised by roughly €300M at the midpoint, and and a reiterated expectation that the liability margin runs above 110bp in 2027 and 2028. The second is the better trade and it is evidently the one that was made.

The setup also amplified it. The shares had drifted down 1.8% over the two sessions into the print while the AEX rose 1.1%, so positioning going in was not stretched despite a 38.9% twelve-month run. There was room to move and a reason to.

Street Perspective

The debates below reflect the arguments circulating around the shares after the print, framed generically. Coverage was already predominantly positive going into the quarter, and the print did nothing to change that direction; the disagreement is about price rather than about the business.

Debate: Is a 17% Return on Tangible Equity the New Level or the Peak?

Bull view: The bull case on the Street is that 17.0% is not a fluke because the drivers are structural: capital has been permanently shifted toward a Retail franchise earning 27% on allocated equity, the liability margin is contracted upward through 2028, and fee income is compounding at 14% with assets under management up 27%. On this reading the four-quarter rolling figure of 14.5% is the number that has to converge upward toward the quarterly print, not the other way round.

Bear view: The bear camp points out that the quarterly print contains €88M of positive volatile items against a full-year guide that takes that line down 20% to 31%, a provision charge 5bp below the through-the-cycle average, and a cost line the company itself says will rise. Normalise those three, the bears argue, and the quarter's return drops back toward the 14.5% four-quarter rolling figure.

Our take: The bears have the quarter and the bulls have the trend, and both overshoot. Stripping the €42M net volatile-item benefit, charging risk costs at the 20bp through-the-cycle rate rather than 15bp, and adding the second-half cost step management has already guided to takes roughly €286M off pre-tax profit and about 170bp off the quarterly return. That lands near 15.3%, which is precisely where the first half already sits and above both the guided floor and the 14.5% rolling figure the bears anchor on. The sustainable level today is mid-fifteens with a credible path toward 17% by 2028, materially better than the April view and roughly what the shares now discount.

Debate: Does the 2027 Upgrade Actually Raise Anything?

Bull view: A one point increase to the 2027 return floor and a €1bn increase to the 2027 income floor, delivered eighteen months ahead of the period, is a strong statement of confidence from a management team that has now beaten and raised in consecutive quarters. Floors from this team have historically been cleared.

Bear view: The pre-print poll already had 2027 income at €26.2bn and the 2027 return at exactly 16.0%. A floor set at €26bn and above 16% does not move a single estimate; it merely stops estimates falling. The market paid a two-year re-rating for one year of upward revision.

Our take: The bear framing is arithmetically correct and the bull framing is right about what it is worth. Converting a consensus estimate into a company commitment reduces the variance around it, and lower variance justifies a higher multiple even with an unchanged central case. What it does not justify is treating the 2027 upgrade as if it were an earnings revision, and the 5.1% single-day move implies at least some of the market did.

Debate: Is Capital Return Getting Easier or Harder?

Bull view: The optimists note that CET1 rose to 13.1% while the bank reserved 100% of its profit outside capital, that 65bp of capital was generated in a single quarter, and that the identified pipeline is substantial: 11 to 16bp of significant risk transfers still to execute this year, a Dutch mortgage-floor release in December, and a semi-annual distribution review in October. On this view the buyback capacity is larger than the current €1bn programme suggests.

Bear view: The sceptics observe that CET1 capital was flat in absolute terms and the whole ratio improvement came from a €2.4bn fall in risk-weighted assets, of which €2.8bn was an unrepeatable model update the bank will not decompose or forecast. Strip that out and risk-weighted assets rose. Meanwhile only 4bp of the 15 to 20bp risk-transfer target has been delivered with the year half gone, and the target ratio of around 13% leaves no structural excess at 13.1%.

Our take: Harder, and the disclosure is honest about why. The distinction that matters is between capital generation, which was genuinely strong at 65bp, and the reported ratio, which improved only because the denominator fell on drivers management explicitly refuses to guide. Both things can be true. But an investor underwriting a step-change in buyback capacity is underwriting model updates and risk transfers, not retained earnings, and October is when that gets tested rather than argued.

Debate: Is 1.78x Tangible Book the Right Price?

Bull view: A bank guiding above 16% return on tangible equity with a contracted margin tailwind, 4% of market value returned in dividends and roughly 2% more in buybacks, should trade at a premium to a European sector that mostly does not earn its cost of capital. On 10.6 times the 2027 consensus earnings line and 9.3 times 2028, the shares are not expensive against that return profile, and European peers approaching 20% returns set the destination.

Bear view: The multiple has done all the work. The shares have gone from 1.44x tangible book at the end of April to 1.78x, and from under 9 times to 10.6 times the 2027 line, in three months, while tangible book per share actually fell to €16.96 from €17.14 because distributions exceeded retained profit. The earnings upgrade is real but smaller than the re-rating that has already been taken.

Our take: This is the crux and the bears have it, narrowly. A residual-income frame using a 10.5% cost of equity and 3% terminal growth puts justified price-to-tangible-book at 1.73x on the guided 16% 2027 return and 1.87x on the 17% consensus 2028 return, a range of €29.50 to €31.50 against a €30.215 close. The shares are inside their own fair-value band, near the middle. That is a Hold, not a mistake.

Model Framework & Valuation

The changes below reflect the raised guidance, the first-half run-rate, and the composition of the beat rather than its headline size.

The valuation arithmetic. Tangible book value per share at 30 June is shareholders' equity of €50,246M less intangible assets of €1,740M over 2,859.7M shares, or €16.96. That is down from €17.14 at 31 March, because the €2,116M final 2025 dividend and €996M of treasury-share movement together exceeded the €1,947M of profit. At the €30.215 close the shares trade at 1.78x tangible book and 1.72x reported book, against 1.44x tangible book at the €24.75 close on the day of the first-quarter print.

A residual-income frame values a bank at justified price-to-tangible-book of (return on tangible equity less growth) divided by (cost of equity less growth). At a 10.5% cost of equity and 3.0% terminal growth: a 16% return justifies 1.73x, a 17% return justifies 1.87x, and a 15% return justifies 1.60x. Against €16.96 of tangible book those are €29.40, €31.66 and €27.14 respectively. The 2027 guided floor and the 2028 consensus return bracket a fair value of €29.50 to €31.50, a midpoint of €30.50, which implies +0.9% against the €30.215 close. On the New York line, at the 1.1564 euro-dollar rate implied by the two closes across the reaction session, the same band is $34.10 to $36.40 with a midpoint of $35.27, or +0.9% against the $34.94 close.

Tangible book per share is held at the 30 June level throughout for comparability; it compounds at roughly the retained share of a mid-teens return, so each scenario is conservative on a two-year view by a similar amount. The distribution is what matters: 12.3% of upside to a bull case that requires ING to reach a return level no guidance currently contemplates, against 17.7% of downside to a bear case built entirely from normalising three items the company has already flagged. That skew is the rating.

Yield support. The interim dividend of €0.40 is up from €0.35 and represents about one third of first-half resilient net profit against a 50% policy, so the final dividend carries the balance. On the pre-print poll's €1.23 full-year dividend the yield is 4.1% at the closing price, with roughly a further 2.3% from the consensus €2.0bn of 2026 buyback, for a total distribution yield near 6.4%. That is genuine support and it is roughly 120bp lower than it was in April, purely because the share price rose.

Thesis Scorecard

The scorecard grades the standing thesis established at initiation on 1 May, unchanged in its pillars, against what this quarter's print and call revealed.

Overall: the thesis is confirmed on substance and its two-sided character has sharpened. All four bull pillars delivered, and the one that slipped, operating leverage, slipped on cost rather than on income. Both capital-related bear points moved against the position. What changed most is not the business but the price: the operating case improved by roughly one point of sustainable return, and over the same three months the shares re-rated from 1.44x tangible book to 1.78x.

Action: hold. The April thesis argued that the upside would come from the withheld replication uplift and from risk-weighted-asset relief rather than from a re-rating. Both arrived, and the re-rating came too and consumed the gap. At 1.78x tangible book, inside a €29.50 to €31.50 fair-value band, with 12.3% of identified upside against 17.7% of identified downside and a distribution yield 120bp lower than at initiation, the risk and the reward are balanced. We would add again below €27.50, which is roughly the 2026 guided-floor value, and would revisit the rating upward if the October distribution review converts the risk-transfer pipeline into a materially larger buyback, or if third-quarter costs land inside the guide without the implied second-half step.