Bank of Baku: growth on borrowed money, a thinning yield on loans and rising payouts to shareholders

Bank of Baku closed 2025 at record speed: assets grew by almost 21% against 7.7% for Azerbaijan’s banking sector as a whole — nearly three times faster than the market.

Profit, meanwhile, stayed exactly where it had been: 27.8 million manats against 27.1 a year earlier.

This is what happens when growth stops paying for itself. A bank earns on the difference between what it charges for its loans and what it pays for depositors’ money. Over the year that difference narrowed by almost two percentage points. And here is what matters most: the bank came to pay for money exactly what every other bank paid — while earning less on its lending in the very year the market earned more.

Why this happened the accounts do not explain, and it is the one genuinely unusual fact of the year.

The price of such speed can be reduced to a single figure. A year ago the bank needed some 34 manats of assets to earn one manat of profit. Now it needs almost 39.

What of this is worth knowing for the bank’s customer? Bank of Baku lives on the public’s money: 92% of customer funds came from individuals, and 82.6% of those deposits are insured by the state. The bank does hold a capital cushion: the requirement is 10.5%, the bank has 13.5%. What follows is less about the soundness of the institution than about the price it pays for speed.

There is also a detail without which the picture is incomplete. Over two years the shareholders did not put a single manat of new capital into the bank — the share capital did not change at all. Their own payout, by contrast, they increased: dividends rose by 14% while profit rose by 2.65%, and the distributed share climbed from 38.8% to 43.1%. As a result the addition to capital out of earnings came in below the previous year’s — 16.1 million manats against 16.9.

The price of growth, however, is such that paying it out of one’s own profit cannot last: the bank is growing appreciably faster than its own earnings permit. Sooner or later the shareholders will have to choose one of three courses — inject capital, cut the dividend or slow down.

In 2025, on the first two of those three, they moved in the opposite direction…

How this bank is built, and how it differs from the market

In January of this year First News Intelligence Unit examined the 2024 accounts of “Bank of Baku” OJSC and closed that piece with a question: how much longer can the expansion continue on such trends? The bank’s audited IFRS financial statements for 2025, signed by “KPMG Audit Azerbaijan” LLC on 17 April 2026, answer it — though not in terms of time.

The expansion did not merely continue, it accelerated. Assets grew by 20.9% and passed 1.18 billion manats, the loan portfolio net of allowances added 19.6%, customer funds 27.0%. Total interest income rose by 19,476 thousand manats over the year, and total interest expense by 17,907. What remained with the bank was 1,569 thousand: eight manats out of every hundred earned in addition.

Between those two figures lies the substance of the year. The accounts allow four sources to be named from which the growth was paid for: capital adequacy, recoveries on previously written-off loans, the shareholder’s subordinated debt and cash. By the end of the year three of the four had become smaller.

Before dissecting 2025 it is worth setting the bank beside the sector. The January piece had no such opportunity; the December 2025 statistics of the Central Bank of the Republic of Azerbaijan provide it.

Growth rates: the sector and the bank

Sources — Statistical Bulletin of the Central Bank of the Republic of Azerbaijan No. 12 for December 2025 and the audited IFRS financial statements of “Bank of Baku” OJSC. Bank figures: assets and capital under IFRS, loan portfolio net of allowances; for the sector — the loan portfolio of credit institutions and balance-sheet capital.

Three lines describe the whole construction: on assets and loans the bank outruns the market several times over, yet on capital it lags behind. It grows, in other words, not on what it earns but on what it borrows.

It differs from the market in whose money it grows on, too. Across the system individuals supply 42.7% of banks’ deposit base, and 63.2% of their deposits sit in term accounts.

At Bank of Baku individuals account for 92.0% of customer funds, and 89.3% of that money is in term deposits. A gap of twenty-six percentage points means something simple and important: the bank has almost none of that free cushion of current accounts on which the market’s funding rests. Every manat it buys.

Three structural differences from the market follow from the accounts.

The first: this is a bank of one product and one type of borrower. Of a gross loan portfolio of 928,530 thousand manats, 905,574 — or 97.5% — is lent to individuals; the auditor separately records that loans and advances to customers make up 76% of total assets. Business lending contracted by 20.8% over the year. The bank’s entire economics rests on a single spread — and in 2025 both sides of it moved against it, though, as will become clear, for entirely different reasons.

The second: the growth was paid for with leverage. The ratio of capital to assets fell from 19.70% to 17.63%, while the share capital did not move at all — 52,870 thousand manats both at the start of 2024 and at the end of 2025. Over two years the shareholders contributed not a single manat and distributed 43.13% of profit as dividends.

The third: cards are ceasing to be a credit product and becoming a payments product. The card portfolio contracted by 13.3% over the year while the portfolio as a whole grew by almost a fifth — which means everything that is not cards grew appreciably faster than the average. This explains more than it appears to at first sight.

Eight manats out of a hundred

Total interest income rose from 148,052 to 167,528 thousand manats, by 13.15%. Total interest expense over the same period rose from 54,664 to 72,571, by 32.76%. Expense grew two and a half times faster than income.

In absolute terms the picture is plainer still. The additional interest income of the year was 19,476 thousand manats, of which 17,907 went to pay for the money raised. Net interest income — the difference between what was received and what was paid — rose by 1,569 thousand: from 93,388 to 94,957, or 1.68%. The portfolio added a fifth; the bank’s principal operating flow stayed where it was.

Not one of these figures is a First News Intelligence Unit calculation: all of them stand as separate lines in the audited statement of profit or loss, and subtracting one from another is the whole of the method here.

Why: two sides of the spread and only one anomaly

Yield and cost of funding: FNIU calculated indicators

Indicators calculated by First News Intelligence Unit. Portfolio yield — interest income on loans to customers over average net loans for the period; cost of funding — total interest expense over average interest-bearing liabilities; cost of an individual source — interest expense on it over average balances. Source — the audited IFRS financial statements of “Bank of Baku” OJSC for 2025.

The calculated yield on the loan portfolio fell from 19.68% to 18.66% even though the numerator grew: interest income on loans to customers came to 153,477 thousand manats against 136,866. The denominator grew faster. The calculated cost of funding rose from 7.63% to 8.55%. The spread narrowed from 12.05 to 10.11 percentage points: 1.02 points were lost to cheaper lending and a further 0.92 points were consumed by dearer funding.

The two halves of that narrowing behave in completely different ways, and the difference is visible only against the market.

The policy rate of the Central Bank of the Republic of Azerbaijan was cut twice during 2025, in July and December, from 7.25% to 6.75%. This did not carry through to the deposit market: the average rate on manat term deposits of individuals across the banking system rose from 9.74% to 10.26%, roughly half a point over the year. The cost of Bank of Baku’s customer funds rose by 0.80 points, of which up to 0.20 is attributable to the shift of the book towards term money. That leaves about 0.60 against the market’s 0.50. On funding the bank moved in step with the market, and there is nothing here to reproach it for.

On lending the reverse is true. The average rate on manat loans to individuals across the system rose over 2025 from 16.32% to 17.07% — again about half a point. The yield on Bank of Baku’s portfolio over the same year fell by 1.02 points. The divergence is one and a half percentage points, and all of it runs against the bank.

This is the anomaly of the year, and it is found not where one would expect it. For money the bank paid what everyone paid. It came to earn less — in a year when the market earned more.

The accounts offer no direct explanation: the currency structure of the portfolio is not disclosed, interest on impaired loans is accrued on the net carrying amount, and within the portfolio one product is being replaced by another. Three versions, and they are not of equal weight.

FNIU ASSUMPTION · What ate the portfolio yield

What it rests on: The yield fell by 1.02 points. The first component is the growth of the non-earning part: the allowance on Stage 3 loans rose from 11,177 to 23,168 thousand manats, an increase of 11,991. At a lending rate of 18.66–19.68% this gives between 1.1 million manats of unrecognised interest income, if the increase was spread evenly over the year, and up to 2.4 million if none of it was earning from 1 January — that is, from 0.14 to 0.29 percentage points out of 1.02. The second component is the product shift: the share of the card portfolio in the gross portfolio fell from 15.0% to 10.8%, by 4.2 points, and cards are traditionally the dearest product in retail lending. For that shift to close the remaining 0.73–0.88 points in full, the yield gap between cards and the rest of the portfolio would have to be between 17 and 21 percentage points.

What could be argued: The bank does not disclose yields by product type, so the second component is measured not directly but through a threshold: the gap has been named at which it would account for the remainder entirely. The gross portfolio for 2024 has been reconstructed by calculation — from the Stage 3 share and from the allowance share, both routes giving about 768 million manats. Nor is a third version excluded, the currency structure of the portfolio, which is not disclosed at all.

What would settle it: A breakdown of yield by type of lending product, or a breakdown of the portfolio by currency.

One product held the profit

Profit before tax came to 35,696 thousand manats against 35,084 a year earlier, a rise of 1.74%. Net profit was 27,821 against 27,103, a rise of 2.65%; earnings per share 8.18 manats against 7.97. Total comprehensive income fell by 21.2%, but the comparison here is uneven: in 2024 it included a property revaluation of 8,210 thousand manats after tax, and in 2025 there was no revaluation.

More interesting is what this almost motionless profit is assembled from: within the statement of profit or loss two large opposing movements took place. The first is provisions: the expected credit loss line — the estimate of losses a bank recognises in advance on its loans and other debt assets, without waiting for actual default — added 2,629 thousand manats of releases to profit in 2024 and took away 4,900 in 2025, a deterioration of 7,529. The second is fees: the total fee and commission result improved by 7,166, from a deficit of 7,182 thousand manats to minus 16 thousands.

The difference between the two movements is 363 thousand manats against profit before tax of 35.7 million. The profit stayed where it was not because nothing was happening, but because what was happening cancelled itself out.

Card fees are payment for servicing payments, not for credit: the bank receives them from merchants and cardholders, and pays them to the payment systems, to the processing centre and on programmes such as cashback. And the fee improvement turned out to be entirely a card story — and that understates it. Card fee income rose from 16,045 to 19,031 thousand manats, expense fell from 18,773 to 14,465, and the balance swung from minus 2,728 to plus 4,566 — by 7,294 thousand, or 101.8% of the entire fee improvement of the year. The rest of the fee economy deteriorated over the same period: non-card income fell from 2,176 to 1,938 thousand, by 10.9%, and the non-card result worsened by 128 thousand. The card reversal covered not only the shortfall on provisions but the decline in everything that is not cards.

Without it, profit before tax would have been about 28.4 million manats — 19.0% below the previous year’s. The difference between a fall of 19% and a rise of 1.74% is the contribution of a single product to the result of the year. And it was obtained on a shrinking portfolio: the card portfolio itself fell from 115,236 to 99,943 thousand manats, income on it rose by 18.6% and expense contracted by 4,308 thousand. Of the 7,294 thousand swing, 59% came from falling expense and only 41% from rising income.

FNIU ASSUMPTION · Card expenses did not fall with volume

What it rests on: The card portfolio contracted by 13.27%, card fee expense by 22.95%. Had expense fallen strictly in proportion to the portfolio, the reduction would have been 2,491 thousand manats; the actual reduction was 4,308. The non-volume part, 1,817 thousand, is 42% of the whole reduction, and expense fell 1.73 times faster than the portfolio. Something therefore changed — a price, a programme or contractual terms — and the swing in the balance on which the year’s profit rests is not underpinned by business growth.

What could be argued: Card fee expense may be tied to card turnover rather than to the portfolio balance. Turnover is not disclosed, and the portfolio has been used as a proxy for it.

What would settle it: A breakdown of card fee expense by type, or an answer from the bank on the revision of terms with the payment system or the processing centre.

An allowance topped up out of something other than profit

The allowance for expected credit losses on loans to customers rose over the year from 18,432 to 32,702 thousand manats — by 14,270.

Movement in the allowance for expected credit losses on loans to customers

Of the increase of 14,270 thousand manats, 4,783 — or 33.5% — passed through profit or loss. Source — the audited IFRS financial statements of “Bank of Baku” OJSC for 2025.

Precision is needed here in two figures: the expected credit loss line in the statement of profit or loss is 4,900 thousand manats, the total charge across all of the bank’s debt instruments, and 4,783 of it relates to the allowance on loans to customers.

Recoveries on previously written-off loans — money returned on debt the bank had already removed from its balance sheet as irrecoverable — came to 12,999 thousand manats, or 91% of the entire increase in the allowance. Under the accounting rules such a receipt restores the allowance without passing through expense; together with write-offs and the unwinding of discount, a net 9,487 thousand, two thirds of the increase, bypassed the statement of profit or loss. Put in one sentence: the bank’s allowance was topped up chiefly not out of what was earned during the year but out of what was collected on debts it had itself once written off as hopeless.

Both mechanisms contracted over the year: recoveries fell by 37.8%, from 20,891 to 12,999 thousand manats, and write-offs dropped by 70.4%, from 16,623 to 4,926. At the same time the problem part of the portfolio doubled. Stage 3 loans — those on which a default event has already occurred, generally arrears of 90 days or more — rose from 19,810 to 40,117 thousand manats, and their share of the gross portfolio from 2.58% to 4.32%. Arrears of more than 360 days more than tripled, from 5,543 to 17,788 thousand. In the card portfolio, which contracted over the year, arrears of more than 360 days rose from 103 to 3,143 thousand manats.

The total allowance against the gross portfolio rose from 2.40% to 3.52%, yet Stage 3 coverage barely moved: 57.75% against 56.42%. The allowance grew alongside the problem rather than ahead of it. The bank moved its own model parameters in the same direction: the upper bound of the annual probability of default on retail lending rose from 16.59% to 27.90%.

Capital that nobody added to

The regulatory series reads unambiguously. Risk-weighted assets grew by 19.95% — from 1,033,356 to 1,239,517 thousand manats; total regulatory capital after deductions by 10.73%, from 151,103 to 167,313. Risk grew almost twice as fast as capital.

Capital adequacy — the ratio of regulatory capital to assets recalculated to reflect the riskiness of each of their kinds — fell from 14.62% to 13.50%, and Tier 1 capital adequacy from 11.37% to 10.55%. Our January piece cited 15.26% at the end of 2023: the indicator has declined for a third consecutive year.

The bank meets the requirements, and with a margin worth stating plainly. The Rules on the calculation of bank capital oblige banks outside the systemically important group to maintain total adequacy of not less than 10%, and from 1 March 2025 a countercyclical buffer of 0.5 points was added to that. Bank of Baku is not on the list of systemically important banks: its threshold is therefore 10.5%, the margin is three percentage points, and total capital of 167,313 thousand manats exceeds the minimum of 50 million more than threefold. This piece is about the price of growth, not about the soundness of the institution, and these figures are given so that the difference is plain.

The statement of changes in equity shows what the capital grew on — and what was absent from it. Under IFRS: 167,981 thousand manats at the start of 2024, 192,768 at the end of 2024, 208,589 at the end of 2025. Growth of 14.76%, then of 8.21% — half as fast in the very year that asset growth accelerated. The only movements that changed the amount of capital in 2025 were profit and dividends; there was no capital contribution in either year.

Dividends, meanwhile, rose by 14.0%: 12,000 thousand manats declared and paid against 10,526, or 3.53 manats per share against 3.10. The payout ratio, taking what was declared in a year against the profit of that same year, climbed from 38.84% to 43.13%.

The result is visible in a single line. Net profit rose by 2.65%, while the part of it that stayed in the bank and works as capital contracted: retained earnings were topped up by 16,120 thousand manats against 16,852 a year earlier — 4.3% less.

FNIU ASSUMPTION · Growth outpaces the bank’s own profit by thirteen points

What it rests on: IFRS capital of 192,768 thousand manats at the start of the year and 208,589 at the end, an average of 200,679. Return on equity of 13.86%, a retention ratio of 0.5687 after distributing 43.13% of profit. Their product is 7.88%: the rate of growth the bank is able to finance itself. Actual asset growth was 20.9%, the gap 13.0 percentage points, and it was closed with leverage. All the input figures are disclosed in the statement of changes in equity.

What could be argued: The sustainable growth rate is a model, not a disclosed figure: it assumes both the return on equity and the payout ratio remain unchanged.

What would settle it: Disclosure of the bank’s capital plans, or the resolution of the general meeting of shareholders on dividends for 2026.

January: the source that fell away

The January piece closed its section on capital with an expectation: subordinated debt of 16,762 thousand manats from “Azpetrol Neft Şirkəti” LLC, a shareholder with a 28.89% stake, was due for repayment in January 2026. It was repaid in full on 27 January 2026 — an event after the reporting date. The rate was 4.50% per annum, and it had been raised in January–March 2016.

Against that background, on 23 September 2025 the bank placed bonds for the first time: 20,000 thousand manats for 18 months, unsecured, with a 13% coupon at an effective rate of 14.46%. The effective rate is the full cost of the borrowing including issue costs and the payment schedule; it, and not the coupon, is what compares with the cost of other sources. The accounts establish no causal link between the placement and the repayment. Only the price is comparable: 4.50% and 14.46%.

FNIU ASSUMPTION · The January repayment is a funding event, not a capital event

What it rests on: Under clause 7.4 of the Rules on the calculation of bank capital, subordinated liabilities are included in Tier 2 capital with an annual reduction of 20 percent over the last five years to maturity. The debt was raised in January–March 2016 with repayment in January 2026, and so at the reporting date stood in the final year of that amortisation. The arithmetic confirms it: the gap between total adequacy of 13.50% and Tier 1 adequacy of 10.55% is 2.95 points, or about 36.5 million manats, whereas the current year’s profit alone, which enters Tier 2 under clause 7.2.1, is 27,821 thousand. Had the subordinated debt been included at full value, Tier 2 would have exceeded the observed gap. The repayment therefore has almost no bearing on capital adequacy: what it removes from the balance sheet is 16,762 thousand manats that cost 4.50% per annum.

What could be argued: The calculation assumes that the regulatory profit for the current year is close to the IFRS profit. The bank does not disclose the composition of Tier 2 capital.

What would settle it: A breakdown of Tier 2 capital.

The price of growth, reduced to a single ratio

Everything said so far converges in the return on assets. In 2024 the bank earned 2.93% on average assets, in 2025 — 2.57%. Put another way, a year earlier each manat of profit rested on about 34 manats of balance sheet, and now on almost 39. To obtain the same manat the bank needs fourteen percent more assets than a year ago.

The ratio remains above the market’s: the sector’s 2025 profit gives about 2.1% on average assets. Bank of Baku earns more on its assets than the system does — but the gap is closing, and closing at the bank’s expense.

This is the point at which to stop, because it is no longer an observation but a diagnosis.

A bank of one product is replacing the high-margin part of that product with a low-margin one, growing nearly three times faster than the market on a difference that is becoming smaller, and funding that growth with money for which it pays the market price, having no free cushion. Every element of the construction works as it should. The question is what their combined working costs — and the return on assets answers it.

What else is worth noting

Three figures move in the other direction, and they are set out here deliberately: without them the analysis would look tidier and would be worse.

The cumulative liquidity position over a 12-month horizon did not change in absolute terms — minus 90,422 thousand manats against minus 90,433 — and relative to financial assets it even improved, from minus 9.58% to minus 7.89%. That improvement, however, is arithmetic: an unchanged gap measured against a balance sheet that grew by a fifth. Within the gap itself the structure shifted towards the short end — the 3-to-12-month interval deepened almost twofold, from minus 53,151 to minus 100,862, and the share of contractual outflows falling due within 12 months rose from 61.6% to 68.8%.

Operating cash flow came out of the negative: plus 39,719 thousand manats against minus 5,957 a year earlier. Understanding what this swing of 45,676 thousand consists of matters here. Under IFRS the inflow of customer funds is, for a bank, an operating item, and the customer deposits line in the cash flow statement improved over the year by 89,784 thousand manats — twice the whole swing; the lending line, meanwhile, worsened by 49,227.

In other words, the positive operating flow of 2025 arose not because operations began to bring in more money but because the bank was raising funds faster than it was placing them. The figure is correct and correctly presented — it simply measures the pace of funding rather than the profitability of operations.

Operating efficiency improved genuinely, without qualification. Operating income rose by 10.4%, staff and administrative costs together by 4.8%; the cost-to-income ratio fell from 61.05% to 57.95%, by 3.10 percentage points. The bank manages its costs markedly better than it manages the price of money.

A correction to the January piece

The bank restated its comparative figures: under Note 2(d), 3,614 thousand manats were reclassified from interest income to card fee income, since that amount does not form part of the effective interest rate and does not relate to the price of the loan. The January FNIU article gave card fee income for 2024 as 12.4 million manats and the overall fee deficit as 10.8 million — both figures from the pre-restatement version; on the corrected basis the card deficit was 2,728 thousand. The January thesis about a widening gap in the card economy is exhausted: the direction has reversed. The restatement affects comparative figures only and has no bearing on the 2025 result.

Where the bank has ended up

The four sources named at the outset look, by the end of the year, like this.

Capital adequacy — 13.50% against 14.62%, a third consecutive decline, against a threshold of 10.5%. Recoveries on previously written-off loans — 12,999 thousand manats against 20,891, minus 37.8%, and they closed 91% of the increase in the allowance. The shareholder’s subordinated debt at 4.50% has been repaid with no replacement, while the bank’s newest debt instrument costs 14.46%. Cash — 131,952 thousand manats against 117,272: the only one of the four that grew, and still below the 155,841 of the start of 2024. Three of the four became smaller.

Above these particular conclusions stands one general one, and it concerns a break within a single year. Over 2024 the bank’s capital grew by 14.76%, over 2025 by 8.21%: capital growth nearly halved at precisely the moment asset growth accelerated to 20.9% against 7.7% for the sector. The share capital did not move by a single manat in either year. The bank did not build up leverage gradually — it changed regime, and did so in the year its portfolio yield went against the market.

What follows is arithmetic that requires no forecast. A bank growing faster than its own profit permits is increasing leverage by definition — there is no other way to close the gap, and it can be closed by only three routes: a capital contribution, a lower dividend payout or slower growth. Which of the three will be chosen the accounts do not say. They say that a choice will have to be made.

Two questions remain, and both are about money. First: what caused card fee expense to fall — the one segment that held the year’s profit rests on it, and 42% of that reduction is not explained by portfolio volume. Second: what will top up the allowance when recoveries on old write-offs contract further — in 2025 they closed 91% of its increase, and they themselves fell by almost 40%.

The January question ran: how much longer can the expansion continue on such trends? The 2025 accounts do not answer how much longer. They show the price.

Note:

First News Intelligence Unit will put these two questions to “Bank of Baku” OJSC together with a link to the present piece. The bank’s response will be published in full.

Methodological note

The principal source is the audited IFRS financial statements of “Bank of Baku” OJSC for the year ended 31 December 2025. The auditor is “KPMG Audit Azerbaijan” LLC, opinion dated 17 April 2026, unmodified and without qualification; the single key audit matter is expected credit losses on loans to customers. The 2024 and 2023 financial statements were audited by other auditors, who issued unmodified opinions on 1 April 2025 and 1 May 2024; “KPMG Audit Azerbaijan” LLC audited only the adjustments described in Note 2(d) and confirmed that they are correct. Sector indicators, deposit and lending rates and the interest rate corridor parameters are taken from Statistical Bulletin No. 12 of the Central Bank of the Republic of Azerbaijan for December 2025. Capital adequacy requirements are taken from the Rules on the calculation of bank capital and its adequacy, approved by the Management Board of the Central Bank.

The calculated indicators — portfolio yield, cost of funding, spread, coverage ratios, return on assets and on equity, the dividend payout ratio, the cost-to-income ratio and the sustainable growth rate — were computed by First News Intelligence Unit from the statement of profit or loss, the statement of financial position, the statement of changes in equity and the notes; the construction of each is stated in the text or in the table captions. The boxes headed “FNIU ASSUMPTION” contain the editors’ interpretations, not data from the accounts: each of them names the figures it rests on, the competing explanation by which it could be argued, and the disclosure that would settle it. The accounts do not disclose the composition of Tier 2 capital, the currency structure of the loan portfolio, yields by type of lending product, or the basis on which interest is accrued on Stage 3 loans; these places are marked in the text as undisclosed and have not been reconstructed by calculation. All figures are in thousands of manats unless stated otherwise.

Read in other languages:

Bank of Baku: рост в долг, тающая доходность кредитов и растущие выплаты акционерам

Bank of Baku: borc hesabına artım, əriyən kredit gəlirliliyi və səhmdarlara artan ödənişlər