The price of the experiment: AFB Bank doubled in size but could not make it pay

In our previous article we called AFB Bank a bank-experiment with an open ending and set out three indicators against which that experiment could be judged.

The audited financial statements for 2025 have now been published, and they show that two of those indicators have closed and the third has not — which is a result in its own right.

The detail follows in our analytical report.

The key findings

●        The forecast on dividends was confirmed to the letter: the shareholder distributed the profit in full — the third cycle in a row. The AZN 4,108 thousand paid out in 2025 is the entire net profit of 2024, matching to the manat per share. Because the bank’s profit is declining, full distribution eats into capital: over the year it contracted by AZN 1,914 thousand.

●        The operating result before provisions was AZN 414 thousand against assets of AZN 417.6 million. Some 85.3% of pre-tax profit was generated by the release of provisions rather than by the bank’s operations.

●        Net interest income fell even as the loan portfolio grew by 62%: interest expense rose by 67.1% against 14.5% for interest income.

●        The growth was paid for not with concessional state funding but with expensive retail money: term deposits of individuals grew 4.2-fold. Deposits of the ultimate controlling party, meanwhile, are placed at 4% — while the bank pays retail depositors 11.58%.

●        The capital adequacy buffer contracted by a third over the year, the liquidity coverage ratio fell by more than a third, and depositor concentration increased: 28 clients hold 62.7% of the deposit base.

●        The key test — the quality of loans issued in 2025 — is not closed by the statements: they contain no breakdown by year of origination.

What we were testing

In April 2026, First News Intelligence Unit published an analysis of AFB Bank — an institution that had spent five years losing assets, extracting capital through dividends and consuming its liquidity cushion, and that in 2025 abruptly became the fastest-growing bank in the country.

We chose to call it neither a crisis nor a renaissance. We called it an experiment, and we set out three indicators against which it was to be assessed: the 2025 audit and the quality of new lending, the shareholder’s dividend decision, and the trajectory of the complaints index.

Related reading:

AFB Bank: Diagnosis and Three Scenarios for the Future of a ‘Bank-Experiment’

The 2025 statements were audited by Baker Tilly Azerbaijan; the opinion was signed on 16 March 2026 and is unmodified — that is, unqualified.

One detail of that opinion will be needed straight away. The key audit matter — the one that demanded the greatest attention from the auditor — is identified as the allowance for expected credit losses on loans to customers. It is the movement in precisely that allowance, as will be seen below, that produced the greater part of the bank’s profit for the year.

Indicator one: dividends. One hundred per cent distributed — the third cycle in a row

The formulation in the April article was conditional: “If the shareholder again takes the profit in full, the strategy has not changed — only the scale of the operation has. If dividends are minimal, that is a sign of serious intent.”

The shareholder took the profit in full.

A clarification about the order of accounting is needed here, without which the figures read incorrectly. Dividends for a reporting year are declared and paid in the following year: what the bank earned in 2024 is distributed in 2025. The lag itself is ordinary practice and there is nothing remarkable about it.

What is remarkable is something else: AFB Bank’s distribution ratio equals one hundred per cent, and it has done so for three cycles running.

Table 1. Profit and its distribution, AZN million

Source: audited IFRS financial statements of AFB Bank for 2019–2025. The full distribution of the 2023 profit was independently recorded by Fitch Ratings in August 2025.

For 2024 the match is not approximate but exact to the last digit: the dividend per share was AZN 73.82, and earnings per share for 2024 were AZN 73.82.

From this follows the mechanism that explains where the bank’s capital goes.

AFB Bank’s profit declines every year. It is nevertheless distributed in full every year. It follows that a one-hundred-per-cent payout of the previous year’s profit inevitably exceeds what the bank earned in the current year — and the difference is deducted from equity.

In 2025 that difference came to AZN 1,914 thousand: AZN 4,108 thousand paid out against AZN 2,194 thousand earned. The bank’s IFRS equity fell from AZN 77,001 thousand to AZN 75,087 thousand — by exactly the same 1,914. The identity holds to the thousand manats.

A separate word on retained earnings, which stood at AZN 4,694 thousand at the end of 2025. There is no contradiction with full extraction here: the current year’s profit lands in that balance and remains there until the shareholder’s decision, taken in the following year. That is, these 4,694 include the 2025 profit of AZN 2,194 thousand, on which a decision will be taken in 2026.

What this implies for 2026 is a matter of simple substitution. Net profit for 2025 was AZN 2,194 thousand, half the previous year’s. If the rule holds, that is the sum that will be distributed.

The model we described in April as a liquidating dividend has not been cancelled by the lending sprint. It continues to operate on top of a balance sheet that has doubled — with the difference that there will soon be nothing left to distribute.

The central finding: the profit was made by releasing provisions

This is the central fact of the entire document, and it is not visible in the press-release figures.

Table 2. AFB Bank statement of profit or loss, AZN thousand

Source: audited IFRS financial statements of AFB Bank for 2025, Baker Tilly Azerbaijan.

The arithmetic: 2,395 ÷ 2,809 = 85.3%. Eighty-five per cent of pre-tax profit is not the result of the bank’s work with clients but an accounting release of provisions created earlier.

Without that line, what remains of the year is AZN 414 thousand — against total assets of AZN 417.6 million. Expressed against assets, that is 0.099%: one tenth of one per cent. For comparison, a year earlier the operating result before provisions was AZN 6,104 thousand against assets of AZN 292.3 million, or 2.09%. The bank’s capacity to earn before provisioning fell twenty-one-fold in a single year.

Operating income, meanwhile, barely moved — 20,968 against 20,321, a rise of 3.2%. General and administrative expenses over the same year rose by 36.7%. A bank that expanded its balance sheet by 43% did not increase its income, but did increase its costs.

The economics of growth: portfolio up 62%, interest income down 1.8%

The most awkward line in the statements reads as follows.

Net interest income for 2025 was AZN 17,987 thousand. For 2024 it was 18,316. A decline of 1.8%.

This against growth in the net loan portfolio from AZN 174.2 million to AZN 282.6 million, that is by 62.2%.

The decomposition explains everything: interest income rose by 14.5%, interest expense by 67.1%. The average effective rate on customer deposits in manats climbed from 8.76% to 11.58%.

The bank expanded its portfolio by more than half again and earned less in interest than a year earlier. This is what the phrase “the economics of growth are negative” means: not a loss, but growth that does not cover the cost of its own funding.

What paid for the growth

In April we had the Fitch Ratings assessment as of August 2025: wholesale borrowings at 36% of liabilities, predominantly long-term loans from state development institutions. That assessment described the structure as it stood before the main phase of the sprint. The annual statements show what actually financed the growth, and the picture is different.

Table 3. AFB Bank funding sources, AZN thousand

Source: audited IFRS financial statements of AFB Bank for 2025.

Term borrowings — the lines from the Mortgage Fund, the Business Development Fund and the Agrarian Agency — contracted over the year. State funding was not being scaled up; it was shrinking while the portfolio grew.

The increase of AZN 108.4 million was paid for by other means: term deposits of the population, which grew 4.2-fold; corporate demand balances; and interbank funds, of which the bank had had practically none a year earlier — AZN 419 thousand against 31,416. In addition, the securities portfolio was sold down: from 16,414 to AZN 503 thousand, with proceeds of 15,463 per the cash flow statement.

The difference between these sources lies not only in price but in character. A development fund line is a predictable five-year resource. A corporate demand balance and an interbank deposit leave on the day the client decides to leave.

A separate line

Deposits of the bank’s ultimate controlling party at the end of 2025 stood at AZN 18,843 thousand, placed at 4% per annum. The average rate the bank paid retail depositors on manat deposits in the same year was 11.58%.

Formally there is no breach here: a related party is entitled to place funds on any terms, and these terms are advantageous for the bank. In substance it means that the growth model rests on two different classes of money — the shareholder’s cheap money and the public’s expensive money — and that the latter grew faster in 2025.

Capital and liquidity: a third of the buffer spent in a year

Table 4. Capital adequacy and liquidity, AZN thousand and %

Source: audited IFRS financial statements of AFB Bank for 2025.

The mechanics are simple in the extreme. Total capital did not change at all over the year — 75,017 against 75,026, a difference of nine thousand manats. Risk-weighted assets grew by 47.5%. The denominator rises, the numerator stands still, and the ratio falls.

The regulatory minimums (10% for total adequacy and 5% for leverage) are met with room to spare, and there is no question of a threat to the ratios. But 11.5 percentage points of buffer were spent in a single year — roughly a third of what there was. If the pace of portfolio growth and the one-hundred-per-cent dividend policy are both maintained, that reserve is finite as a matter of arithmetic: capital is replenished neither by the profit the shareholder takes nor by contributions, of which there were none.

The liquidity coverage ratio fell from 219.70% to 137.3% — by more than a third. The requirement is still exceeded, but the direction over the year is unambiguous.

Concentration rose on both sides of the balance sheet

A year ago Fitch’s principal warning concerned asset concentration: the top 25 borrowers accounted for more than 70% of the corporate and SME portfolio. The 2025 statements do not disclose that measure on the same basis, but they offer one of their own.

On the asset side: 19 borrowers or groups of related borrowers with balances above AZN 1,000 thousand hold AZN 83,468 thousand. A note to the statements describes this as 28% of loans to customers; calculated against the net portfolio (83,468 ÷ 282,612) it gives 29.5%, and against the gross portfolio (83,468 ÷ 300,463) 27.8%. Measured against the corporate and entrepreneurial portfolios taken together (165,764), those same 19 borrowers account for 50.4%.

On the liability side: 28 clients with balances above AZN 1,000 thousand hold AZN 140,905 thousand — 62.7% of the entire deposit base. A year earlier this was 24 clients and 57.6%. Funding concentration rose at the same time as the funding itself: the inflow of retail term deposits did not dilute the base but was accompanied by its further tightening.

Insurance cover from the Azerbaijan Deposit Insurance Fund protects AZN 63,174 thousand, or 28.1% of deposits. Almost three quarters of the base sits outside the guarantee.

Related-party funds on deposit amount to AZN 20,955 thousand against 12,615 a year earlier, that is 9.3% of the base.

Portfolio quality: the ratio improved, the substance did not

Formally, the headline figure of the April article improved twofold. The share of loans in the third stage of impairment (Stage 3 under IFRS 9) came to 7.30% of the gross portfolio against 14.42% a year earlier.

The improvement, however, was delivered by the denominator.

Table 5. Portfolio and problem debt, AZN thousand

Source: audited IFRS financial statements of AFB Bank for 2025.

Debt overdue by more than 360 days rose in absolute terms — by AZN 2,184 thousand. Write-offs, meanwhile, amounted to 2,299 against 16,009 a year earlier: in 2024 the portfolio was being cleaned, in 2025 it essentially was not. The share fell because the gross portfolio grew by 57.6%, not because problem debt contracted.

At the same time — and this should be recorded just as plainly — provision coverage rose. Stage 3 is covered at 79.0% against 52.7% a year earlier; with purchased or originated credit-impaired assets in the denominator, 66.2% against 46.6%.

Where the risk sits

Table 6. Portfolio structure by segment as at 31 December 2025, AZN thousand

Source: audited IFRS financial statements of AFB Bank for 2025.

The corporate block holds AZN 12,504 thousand of debt overdue by more than 360 days — 11.05% of the segment. This is an inherited problem, and it has not gone anywhere.

The consumer portfolio, which grew almost threefold over the year, is secured by collateral of AZN 201 thousand — against a portfolio of AZN 64.8 million. This is neither a typo nor a dropped order of magnitude: three tenths of one per cent, or 31 gapik of security for every 100 manats of credit extended. The provision for the segment is 1.42%.

FNIU ASSUMPTION

We describe the consumer portfolio as effectively unsecured.

What it rests on: the collateral table in the notes to the statements shows AZN 201 thousand for the consumer segment against a portfolio of 64,792.

What could be argued: the absence of security is the norm in consumer lending, and it is compensated by the interest rate and the short tenor rather than by collateral. The low provision may reflect not an underestimate of risk but the youth of the portfolio: loans issued in the second half of 2025 have not yet passed their seasoning period.

What would settle it: the trajectory of the Stage 3 share in the consumer segment over 2026, and disclosure of quality by year of origination.

The third indicator does not close

The April test was formulated precisely: “The 2025 audit will show the real quality of AZN 109 million of new loans. If Stage 3 on new lending is already in the 10%-plus zone, the model is in question.”

Answering that question from the published statements is impossible.

The breakdown by year of origination (vintage analysis) is not disclosed in them. All AZN 204,083 thousand of new assets are recorded in the first stage. Transfers to the third stage over the year amounted to AZN 3,108 thousand gross, but attributing them to the 2025 cohort on the basis of the statements is not possible: loans issued earlier could equally have migrated to Stage 3 during the year.

We record this as a gap rather than substituting the overall Stage 3 share for an answer. The overall share improved. What happened to the specific 204 million issued in the year of the sprint does not follow from the document.

Operational context

The Central Bank’s complaints index, the third of the April indicators, is tracked by FNIU on a monthly basis, and according to our publications in June the bank remained in the upper part of the anti-rating. The statements add a dimension to this that the index does not contain.

Headcount over 2025 grew from 289 to 420 people — by 45.3%. The number of branches went from six to seven. Personnel expenses rose from AZN 7,753 thousand to AZN 12,170 thousand, or by 57%.

Infrastructure was being built up. But the loan portfolio grew by 62%, and the client base faster than the staff. The April diagnosis of “operational overheating” is confirmed by the statements rather than refuted: personnel costs are the largest component of that very 36.7% increase in administrative expenses which consumed the operating result.

Where the scenario has moved

In April we described three possible trajectories: managed transformation, stagnation at a new level, and overheating followed by a reversal.

The 2025 statements do not fit any of them completely — and that is perhaps the principal conclusion of the year.

From the first scenario there is growth as a fact: assets of AZN 417.6 million, a gross portfolio of AZN 300.5 million, a deposit base of AZN 224.9 million, and provision coverage of problem loans that rose rather than fell. The condition we attached to the first scenario — a restrained dividend policy — has not been met.

From the third scenario the essential element is absent: the collapse. The problem portfolio did not unravel, liquidity is within norms, the ratios are observed, and the auditor signed an opinion without qualification.

What has materialised is a configuration that was not in the April set: growth took place, but it did not pay for itself. The bank doubled its balance sheet while earning AZN 414 thousand operationally; financed this not with state funding but with retail money at 11.58%; handed the shareholder the entire profit of the previous year — AZN 1,914 thousand more than it earned in the current one; and spent a third of its capital buffer and reduced its liquidity coverage ratio by more than a third.

A year ago we wrote that before us was an institution that had changed its operating regime without changing its fundamental risks. The 2025 statements refine that formulation: the regime changed, the risks did not, and the cost of the new regime is now measurable.

Three indicators we will be following

The dividend decision on the 2025 profit. Retained earnings at the end of the year stood at AZN 4,694 thousand against an annual profit of 2,194. If the distribution is once again one hundred per cent, the question of where capital is to be replenished from will cease to be theoretical.

Debt overdue by more than 360 days in absolute terms, and the volume of write-offs in 2026. In 2025 it rose by 2,184 against write-offs of 2,299. If the absolute figure continues to grow in 2026 while the portfolio decelerates, the denominator effect will be exhausted and the share will turn upward.

The cost of funding. The effective rate on customer deposits in manats rose from 8.76% to 11.58% over the year. The relationship between it and portfolio yield will determine whether the interest margin recovers or the decline in net interest income becomes a trend.

***

First News Intelligence Unit (FNIU) is the analytical division of the 1news.az editorial office, specialising in research on the financial sector and the economy of Azerbaijan.

Read in other languages:

Цена эксперимента: AFB Bank вырос вдвое, но заработать на этом не получилось

Eksperimentin qiyməti: AFB Bank iki dəfə böyüdü, amma bu artımı qazanca çevirə bilmədi