Turanbank: a paper-thin net profit, “state drip-feed,” and a wound the bank has failed to heal for 10 years

Turanbank’s net profit fell 89 percent in 2025. But this is not a collapse — it is a rerun: the same script — record income against nearly vanished profit — the bank had already run through just two years earlier.

Everything comes down to one line of the report: the reserves the bank sets aside against the possibility that loans it has issued will not be repaid.

The reported profit of this bank, therefore, does not reflect how well it operates. It shows only one thing — where in the reserve cycle the bank currently stands: in a year of new provisioning, reserves devour profit; in a year of release, they conjure it almost out of thin air.

And all of this — at a bank with no controlling owner, one that has paid no dividends for ten years yet receives fresh capital injections again and again. Over sixteen years — not a single auditor’s qualification, and twice in three years profit swallowed almost whole by reserves. Who recapitalizes such a bank, and why, is the central question of this story.

* * *

On 26 May 2026, the international audit firm Baker Tilly signed off on the 2025 accounts, confirming a spotless 16-year record of clean reports.

And the very first figure most readers will open the report for looks like a verdict: net profit for 2025 came to just 852 thousand manats, against nearly 8 million (7,966 thousand) a year earlier — a fall of almost 89 percent in twelve months.

Reading this as a collapse is easy, and nearly everyone will do exactly that. But no collapse, in the usual sense of the word, took place here — and therein lies the whole point. A hundred million manats of interest income (a record 100.4 million against 87.8 a year earlier) turned into a mere 852 thousand of profit not at all because the bank had forgotten how to earn. On the contrary: on its core business — the difference between the interest on loans it issues and the interest it itself pays on deposits and borrowings — the bank earned more in 2025 than ever before. This core income, before deducting reserves, rose to 45.1 million manats: 12 percent more than in 2024, and almost double the 2022 figure.

Profit was destroyed by a single line of the report — the charge for reserves against possible loan losses.

When a bank has grounds to believe that some of the money it has lent may not come back, it is required to write off the expected loss as an expense in advance; this is loan impairment, or in accounting language, expected credit losses (ECL). In 2025 the bank booked twice as much of these reserves as before — 25.4 million manats against 12.8 — and that was enough to ensure that almost nothing of the hundred-million income reached the bottom line.

The temptation to see an accounting trick here — a way to understate profit tax, say — should be dismissed at once. Reserves under the IFRS 9 standard are not discretionary: the bank does not choose whether to book them; they are triggered by a real deterioration in the quality of the portfolio.

And that deterioration is plain to see: the volume of problem corporate loans grew over the year from 3.3 to 14.2 million manats. This is not magic and not optimization, but the mechanical consequence of a sagging loan book.

And here something emerges that cannot be discerned within a single reporting year: this had happened before — two years earlier. In 2023 the bank posted a profit of 811 thousand manats against rising core income and a collapse of the bottom line created entirely by those same reserves; in 2025 — a profit of 852 thousand, again against rising income and again because of reserves.

Two structurally identical years within a three-year span.

From this comes the central conclusion of this piece, and we state it plainly, without hedging: Turanbank’s reported profit does not measure how well the bank operates. It shows only where in the reserve cycle it stands. In a year when reserves are booked, they devour profit; in a year when they are released, they create it. Profit in the ordinary sense, in essence, the bank does not have — there is a reserve cycle, and everything else is built around it.

To see this, one report is not enough. For this study we collected and cross-checked Turanbank OJSC’s audited accounts for sixteen consecutive years, from 2010 to 2025: sixteen annual opinions, and in every one the auditors confirmed the accounts without qualification. To this series we added the CBAR complaint index, which FNIU tracks month by month through its own reviews, and open sources. The very depth of the horizon becomes an argument here: a bank that, read year by year, appears now as victim, now as victor, looks entirely different across a sixteen-year distance — like a mechanism, and one that repeats.

Let us name it plainly: this is neither a “reliable” nor a “suspect” bank, but a convalescent bank (one still recovering from an illness), living from one reserve cycle to the next.

The backdrop is worth spelling out too. A surface reading of the latest report leads easily to reassuring conclusions — of an “ideal portfolio equilibrium,” of “powerful capital and a reliable buffer,” of “a bank that held the line.” The sixteen-year series shows something else: not equilibrium but a pendulum; not a closed buffer but an unhealed wound.

Sixteen years at a glance

The whole story surfaces at once, the moment you lay three lines of the accounts side by side — the bottom-line profit, the reserves against loan losses (ECL), and the accumulated result, that is, the sum of all the bank’s profits and losses over its years of operation.

Thousand AZN. Source: audited accounts of Turanbank OJSC, 2010–2025. A plus in the reserves column means that in that year the bank was not booking new reserves but releasing previously created ones back into profit.

Here is how it reads. Up to 2014 — a steady climb, peaking at a profit of 5.6 million manats.

The 2015 devaluation of the manat breaks the trajectory: the bank goes into loss for the first time, and the accumulated result turns negative — the very hole in capital that will then take years to close.

The bottom — 2016: a loss of 13.3 million, and almost all of it created by a single reserves line (13.3 million as well). That year the bank did not so much lose on operations as it was forced to recognize, all at once, expected losses on previously issued loans.

The blow was compounded by the switch to the new IFRS 9 accounting standard in 2018 — which requires expected losses to be recognized earlier and more fully — and the bank’s capital drops to 25.1 million manats, the lowest in its history.

Then — four consecutive years of profit, from 2019 to 2022, ranging from 0.4 to 6.8 million manats. What matters here is not the size but the nature of this profit: the bank did not, in essence, earn it — it “returned” it. The driver was not work with clients but the flip side of reserves: in earlier years the bank had set them aside with a margin, and now released part of that margin back into profit (in the table, these are the years with a plus in the reserves column). To this was added roughly 25 million manats of fresh money that shareholders contributed to capital through new share issues in 2019–2021. Profit exists on paper — but the hole in capital is still not closed.

From 2023 the cycle turns the other way: reserves once again begin to be booked rather than released, and every new charge strikes at profit. The turn happens precisely in 2023: a charge of 9.4 million compresses profit to 811 thousand. 2024 turns out to be the only reprieve — rising core income just manages to cover a charge of 12.8 million and carry profit to nearly 8 million. And 2025, under a doubled charge of 25.4 million, breaks again, dropping to 852 thousand.

The accumulated loss — that same hole in capital, born in 2015 — has been shrinking gradually since 2019, but across an entire decade it still does not close: as of 31 December 2025, a minus of 8.2 million manats of uncovered loss still “hangs” in the bank’s capital. The dynamics of healing are telling too: in the bad years it almost stops. In 2023 the hole shrank by exactly 811 thousand — precisely the amount of the year’s profit and not a manat more. The wound closes only when no new wave of reserves is running.

Diagnosis: two reserve cycles in a decade

If sixteen years are reduced to a single motion, the following emerges. The bank’s core income — what it earns on the spread between interest rates — grows almost continuously: selling loans is something the bank knows how to do. But the bottom-line profit, year after year, is determined not by this income but by one reserves line, and in both directions at that — both when they are released and when they are booked.

Over the decade the bank went through two full such cycles: a wave of provisioning in 2015–2018 (the first reserve fire), then years of releasing reserves and paper profit in 2019–2022, and a new wave of provisioning from 2023 — the second fire.

The distinctive feature of the second wave is that core income did not fall during it, but grew. Before deducting reserves, it traveled from 24.9 million manats in 2022 to 32.3 in 2023, 40.1 in 2024, and 45.1 in 2025; the last year’s interest income became a record, and the average rate at which the bank issues manat loans rose from 12.1 to 15.5 percent. In the year profit vanished, the bank was earning on loans more actively than ever. Profit was killed not by a drop in income, but by reserves.

What exactly required these reserves in 2025 is worth pinning down precisely, because there were two processes, not one.

The first — in loans to small and medium-sized business: the bank wrote off bad debts of 20.2 million manats out of 28.1 million of all write-offs for the year.

The second, independent — in corporate loans: large loans of 18.7 million manats the bank reclassified as problem loans. Under international accounting rules, loans are divided into three “stages” by quality: the first — normal, the second — those whose risk has risen noticeably, the third — problem loans, whose repayment is already in question and against which maximum reserves must be held. The volume of such problem corporate loans grew over the year from 3.3 to 14.2 million manats.

Two different processes in two different segments — and both ran up against a single expense line.

A legitimate question arises, one the reader asks almost automatically: at whose expense does the bank cover these losses — and does all this not resemble a pyramid? The data give the answer, and it is no. In a pyramid, payments to earlier participants come from the money of new ones, with no real earnings underneath. At Turanbank it is different. It covers loan losses, first, out of its own interest margin — and that is real money, 45.1 million of pre-tax operating profit for the year — and, second, out of capital that shareholders replenish.

The money of development institutions, on which the bank grows, has nothing to do with it: these are borrowed funds the bank is obliged to return with interest, not somebody’s deposits used to pay off others.

More precisely, this is not a pyramid but a treadmill: cheap targeted money accelerates the growth of the portfolio, growth turns into bad loans a few years later, bad loans into reserves, reserves devour profit and capital, and capital is topped up by shareholders. And the circle repeats.

The linkage that gives the picture its edge we record strictly as an observed sequence, without declaring it the sole cause: over 2019–2025 the bank’s loan portfolio (net reserves) nearly doubled — from 315 to 675 million manats, and the bill for this rapid growth comes due in reserves in 2023–2025.

A parallel suggests itself with what we already described in our piece on the “hidden price of growth” at Bank of Baku: rapid growth has a price that shows up not in the year the loans are issued, but several years later. The portfolio’s quality bears this out: of 711.5 million manats of the gross portfolio (before reserves), 26.7 million are counted as problem (Stage 3) loans — about 3.8 percent against roughly 2.7 percent a year earlier.

The same series contains one more observation, which we present without attributing a cause.

Over sixteen years the bank twice, and for exactly one year each time, changed its principal auditor — bringing in KPMG in 2016 and Nexia EA in 2023, whereas the rest of the time, from 2017 on, the accounts have been certified solely by Baker Tilly (before that the auditors were KPMG, Deloitte, and PwC).

Both one-off departures fell on years of a sharp deterioration in results: on the trough of the first wave and on the turn of the second. The Nexia EA opinion for 2023, incidentally, has no “key audit matters” section, which Baker Tilly’s did in 2019–2022 and which touched precisely on loan impairment. The coincidence we record; the conclusion we leave to the reader.

Alongside — a change of the chairman of the management board between the signing of the 2023 and 2024 accounts: Fazail Musayev was replaced by Orkhan Garayev.

The dividend inversion and a bank with no ultimate owner

In this plot Turanbank inverts a storyline the sector knows well. In our piece on AFB Bank we described a “dividend vacuum cleaner” — a situation in which a shareholder pumps profit out of the bank through dividends that sometimes exceed the earnings themselves. At Turanbank the picture is mirror-image: across an entire decade, from 2016 to 2025, the bank paid not a single dividend.

This is neither generosity nor stinginess, but a structural impossibility: as long as the hole in capital — that same accumulated loss — is not closed, there is simply nothing to distribute among shareholders. Where at some banks in the sector the owner spends a decade pumping profit out, here he does the opposite — spends a decade topping up and taking nothing.

The bank’s capital was replenished more than once through new share issues: the charter capital grew from 55 million manats in 2018 to 100 million in 2024, with the last tranche of 20 million falling entirely in 2024.

Who puts up this money? The accounts state it directly: the bank has no ultimate owner — it has 29 shareholders, each holding less than 10 percent, and none has the authority to run the bank in its own interest.

A bank with no obvious master, one that for ten years returns nothing to the contributors of its capital yet receives fresh injections again and again — this is a configuration whose motive the accounts themselves do not reveal. We will not invent one.

But we are obliged to raise the question at full height: who keeps this bank afloat for ten years, and why, receiving nothing in return?

The business model: an institutional drip-feed

Take the familiar sign off Turanbank — “a bank that gathers deposits and issues loans” — and beneath it another construction comes to light. It finances its growing loan portfolio not so much with depositors’ money as with targeted borrowings from state and international development institutions, while its own deposit base has stalled.

Client deposits in 2021–2025 stand almost still — 419, 433, 464, 466, and finally 476 million manats; the growth of the last year was just 2 percent.

The growing portfolio is financed by another balance-sheet item — borrowings, which after a trough of 97.7 million manats in 2019 grew to 284 million in 2025, adding 65 percent in the last year alone. Within this item a single lender dominates — the Mortgage and Credit Guarantee Fund of the Republic of Azerbaijan: 127.2 million manats, about 45 percent of all borrowings, at rates from 1 to 8 percent. Alongside — the Azerbaijan Business Development Fund, the Agrarian Credit and Development Agency, and international development institutions: BlueOrchard, EMF, the Islamic Corporation for the Development of the Private Sector (ICD), the ECO Trade and Development Bank, the Black Sea Trade and Development Bank, INCOFIN, as well as credit lines from Raiffeisen and BCP.

In essence this is the same model we examined with Bank Avrasiya — a bank that grows on other people’s targeted money on top of a thin deposit base.

2025 exposed the fragility of this construction from an unexpected side. The structure of borrowed money changed sharply: interbank borrowings — money the bank borrows from other banks — nearly vanished, falling from 76.5 to 12.8 million manats, by 83 percent. And this is not a one-off episode but the finale of a three-year trend (123.7 million in 2022, then 83.1, 76.5, and 12.8). The gap that opened up was closed by those same development-institution borrowings with their 65-percent growth, while deposits added only 2 percent.

To call this “a transition to a new model” would be to assert precisely the opposite: the bank did not emerge from its dependence on financial institutions — it deepened it.

This model has one more small but telling touch — the technical reclassification of 2016, when almost 48 million manats, listed as “client deposits,” were moved into the category of funds of banks and financial institutions. A purely accounting operation that did not affect the bottom line — but it shows what share of the “client” base was in fact always institutional.

Capital — a chronic bottleneck

The bank’s total capital grew from a low of 25.1 million manats in 2018 to 94.2 million in 2025, and on paper this looks like a confident recovery. But three details of the last year add up to a different picture.

The first — a decline in capital adequacy. Capital adequacy is, simply put, the ratio of a bank’s own funds to its risk-weighted assets; the regulator requires that it not drop below a certain threshold, and it is, in essence, a safety margin against losses.

At Turanbank this indicator fell over the year from 14.73 to 13.12 percent, and its highest-quality part — Tier 1 capital, that is, the shareholders’ own money and accumulated profit — from 10.81 to 10.30 percent. And all of this against a raised bar: in 2025 the Central Bank lifted the minimum for total adequacy from 10 to 10.5 percent, and for Tier 1 capital — from 5 to 5.5 percent. The safety margin is shrinking precisely as the required minimum rises to meet it.

The second detail — subordinated debt. These are long-term loans that, in the event of the bank’s bankruptcy, are repaid last of all, after all other creditors and depositors; in exchange for this “subordination,” the regulator allows them to be partly counted toward capital. The bank has 23.9 million manats of such debt on its balance sheet, but only 14.5 million is now counted toward capital, against 22.2 a year earlier — as maturity approaches, the eligible share melts away. The bank raised a large loan of 25 million in 2024; in 2025 there were no new ones, and this “shrinkage” mechanically presses on capital, while the year brought no profit with which to offset it through its own accumulation.

The third detail — that same accumulated loss, minus 8.2 million manats. This is the unhealed wound of 2015, sitting directly in the structure of capital and reducing it. And here — a direct counterpoint to the reassuring conclusions about a “reliable buffer”: the bank does indeed have a buffer, but for nine years it goes toward darning a hole opened a decade ago, and in 2025 the darning stopped.

Loans to related parties are worth a separate mention. The portfolio contains a loan issued to a shareholder holding 5 percent or more, of 6.8 million manats at 8.27 percent, against which a reserve of 1.8 million has already been booked — meaning the bank considers it possible not to get about a quarter of the sum back.

In itself this is no crime, but precisely this combination — a loan to “one’s own” shareholder, one requiring a sizeable reserve at that — falls among the risks regulators watch first: loans to related parties are easier to issue on non-market terms and harder to recover. We record the fact; there are no grounds to draw from it a conclusion about intent.

The complaint index: the alarm is not where it is sought

There is one more cross-section usually read in haste — the CBAR complaint index, a monthly measure of how often clients complain to the regulator about a bank relative to its market share. And here Turanbank springs a surprise that works directly in favor of the piece’s central thesis.

In the first half of 2025 the bank was not “in the middle of the list,” as it might seem, but among the leaders of the anti-rating: third place in January, second in March, and first in May with a peak value of 3.24. But then the index turned downward — and in October and November 2025 not a single countable complaint about the bank reached the CBAR. By the spring of 2026 Turanbank had descended to the very bottom of the anti-rating: sixteenth place in April and fifteenth in May, with minimal values of 0.32 and 0.36. On average over 2025 the index held at 1.93.

Set the two lines side by side. The bank’s financial condition deteriorated all through 2025: reserves grew toward a peak of 25.4 million. And client dissatisfaction, at the very same time, was falling — all the way to zero. The quality at the teller’s window and the quality of the balance sheet diverged in opposite directions.

The conclusion from this is simple and important: Turanbank’s problem is not visible to the ordinary client. It is not in the queues, the refusals, or the app crashes — it is stitched into the loan portfolio, into those very corporate and SME loans that went into reserves. Like the reported profit, this bank’s complaint index is deceptive — and deceptive in the same direction: calmer outside than in.

Three scenarios for the future

To none of the three paths do we assign a probability — we merely describe the conditions under which each might materialize. The main question is one: has the second wave of reserves passed its peak — or was 2025 only its middle?

No. 1 — A managed clean-up

If the reserve blow of 2025 turns out to be a one-off — if it absorbed the consequences of the rapid growth of 2021–2024 and provisioning declines in 2026 — then rising income will pull profit up, and the accumulated loss will finally close. This path can be recognized by three signs: reserves in the 2026 report turn downward, the accumulated result turns positive, funding sources stabilize.

No. 2 — A third blow

If, on the other hand, portfolio quality continues to deteriorate in the wake of the recent rapid growth, provisioning in 2026 does not subside, and capital melts against an already-raised bar, then the hole in capital deepens again — a new wave on the 2015 pattern. The signs of this path: reserves stay high, the volume of problem loans grows, capital adequacy falls further.

No. 3 — A niche conduit for other people’s money

A third outcome is also possible, in which the bank remains a channel through which development institutions finance targeted programs, never returning to its own sustainable profitability; profit will fluctuate around zero in time with the reserve cycle; the hole in capital will hang for years. Only the scale will change, not the model itself. The signs: income grows, but once every few years reserves devour it, and dependence on institutional funding does not go anywhere.

Three indicators we will be watching

The main one — the sign and size of reserves in the 2026 audit: whether provisioning turns down from 25.4 million or the second wave continues.

The second — the accumulated loss of 8.2 million manats: whether it finally closes or goes deep again.

The third — capital adequacy at 13.12 percent and the fate of the subordinated debt: whether the safety margin holds or continues to melt against a bar raised to 10.5 percent.

As a second circuit we will watch whether interbank borrowings return after their collapse from 76.5 to 12.8 million, whether funding becomes even more concentrated on a single state fund, and what the CBAR complaint index shows — an indicator FNIU tracks month by month. At Turanbank it has now, after its lead in the anti-rating in the first half of 2025, moved to the very bottom of the list: the bank is among those that clients complain about least relative to their size.

A separate indicator — whether Baker Tilly retains the audit mandate for 2026 after the one-off episode with Nexia in 2023.

The finale

Let us return to the question the reader has been asking from the very start: is this a normal bank or not, is it stable or unstable?

The direct answer is this — Turanbank is solvent and in sixteen years has received not a single auditor’s qualification; in that sense it is clean and stable. The failure, after all, sits not in violations but in reserves, and to set aside reserves is not a violation but compliance with the rules. But “stable” here does not mean “healthy.”

This is a bank that for ten years patches the same wound of 2015 without ever closing it, and which has just taken a second — fresh and open, of eight million manats.

Its profit says little about how it operates; its capital holds, but nine years go toward darning the old hole, and in 2025 the darning stopped; it grows on other people’s targeted money on top of a thin deposit base; and it is recapitalized by an owner whose motive is unclear (he has received no dividends for 10 years now).

That is why the final question is not “will the bank survive” — it has survived twice. The question is whether the second wave of reserves that brought down profit in 2025 has passed its peak — or whether we are seeing only its middle. The wound of 2015 is still open at eight million manats, and it is precisely the 2026 audit that will show which of the three scenarios laid out was the right one.

The data, as always, will speak first.

First News Intelligence Unit. This analytical piece was prepared on the basis of the audited financial statements of Turanbank OJSC for 2010–2025, the CBAR complaint index, and open sources.

Read in other languages:

Turanbank: мизерная чистая прибыль, «госкапельница» и рана, которую банк не может залечить 10 лет

Turanbank: cüzi xalis mənfəət, "dövlət damcısı" və bankın 10 ildir sağalda bilmədiyi yara