Analytics
Analytics · data as of 1 Sep 2026
Since 2022, bank loans across the three countries have grown 1.7–2.9-fold in dollar terms; the share of problem loans has risen only in KazakhstanSince 2022, bank loans across the three countries have grown 1.8–3.2-fold in national currencies; the share of problem loans has risen only in Kazakhstan
| Volumegrowth, $growth, nat. | Qualityshare of problem loans | Structurehousehold share | Concentrationstate-owned banks | |
|---|---|---|---|---|
| Uzbekistan | ×1.7×1.8$55.0bn648.9tn UZS | 3.7%↓ from 4.9% | 37.5%↑ from 24.5% | 65%↓ from 84% |
| Kazakhstan | ×2.1×2.0$92.0bn42.8tn KZT | 4.0%↑ from 3.6% | 61.8%→ from 62.1% | –no data |
| Kyrgyzstan | ×2.9×3.2$7.1bn618 bn KGS | 11.5%*↓ from 12.8% | –no data | –no data |
From September 2022 to the latest data: 1 Sep 2026, Kyrgyzstan – 1 Aug 2026. Problem loans are defined differently across countries, so levels are not compared between them.
Below: each factor across the three countries, then a detailed look at the problem loans of Uzbekistan's banks and microfinance organisations.
Region · four factors across three countries
Volume
Since September 2022, bank loans in Kyrgyzstan have grown nearly threefold in dollar terms, Kazakhstan – twofold, Uzbekistan – by 69%Since September 2022 in national currencies, bank loans in Kyrgyzstan have grown 3.2-fold, Kazakhstan – twofold, Uzbekistan – by 83%
Bank loan portfolios in dollars, index: first point = 100. Uzbekistan and Kyrgyzstan from September 2022, Kazakhstan from October 2022Bank loan portfolios in national currencies, index: first point = 100. Uzbekistan and Kyrgyzstan from September 2022, Kazakhstan from October 2022
- Uzbekistan
- Kazakhstan
- Kyrgyzstan
- preliminary data
| Start | Latest data | Growth | |
|---|---|---|---|
| Uzbekistan | $32.5bn355.5tn UZS | $55.0bn648.9tn UZS 01.09.2026 | +69%+83% |
| Kazakhstan | $44.3bn21.1tn KZT | $92.0bn42.8tn KZT 01.09.2026 | +108%+103% |
| Kyrgyzstan | $2.4bn194 bn KGS | $7.1bn618 bn KGS 01.08.2026* | +193%+219% |
Portfolios are defined differently. Uzbekistan: bank loans from the central bank's NPL table. Kazakhstan: loans to the economy of second-tier banks. Kyrgyzstan: commercial bank loans under the NBKR classification. The growth rate is therefore what is compared. Dollar amounts are an indicative conversion at exchange rates on each snapshot date; tenge and som are converted to dollars via the official rates of the Central Bank of Uzbekistan. An asterisk marks preliminary NBKR data.
Sources: Central Bank of Uzbekistan; ARDFM, monthly banking sector releases; NBKR, banking statistics bulletin, table IV.3.5.
Quality
Over four years, the share of problem loans has risen only in Kazakhstan – from 3.6% to 4.0%
Share of problem loans under national definitions, %; each country has its own scale, levels are not compared between countries
NPL per the central bank’s definition
4.9% → 3.7%
NPL90+ per ARDFM: overdue more than 90 days
3.6% → 4.0%
Loans classified by the NBKR as substandard, doubtful or loss
12.8% → 11.5%*
Indicator definitions by country
| Indicator definitions by country | Uzbekistan | Kazakhstan | Kyrgyzstan |
|---|---|---|---|
| What counts as a problem loan | NPL per the central bank’s definition | NPL90+: overdue more than 90 days | The three worst categories of the NBKR classification |
| Regulator | Central Bank of Uzbekistan | ARDFM | NBKR |
| Source currency | sum | tenge | som |
| Can share levels be compared | no Definitions differ; only the direction and pace of change are compared between countries. | ||
* Kyrgyzstan – data as of 1 Aug 2026; the NBKR reports the last three months as preliminary. For Kazakhstan, separate monthly ARDFM releases have been collected; where an adjacent month is missing, points are not joined by a line.
Sources: Central Bank of Uzbekistan; ARDFM; NBKR, banking statistics bulletin, table IV.3.5.
Structure
The share of household loans in Uzbekistan rose from 24.5% to 37.5%, while in Kazakhstan it is about 61.8%
Household share in bank loans. Light circle: September 2022 (October for Kazakhstan), dark: latest data
Uzbekistan – loans to individuals in the banking system portfolio per the central bank's loan-type table; Kazakhstan – household loans within loans to the economy per ARDFM releases. The structure of all loans is shown, regardless of their quality.
Sources: Central Bank of Uzbekistan, loan portfolio balances by loan type; ARDFM.
Concentration
State-owned banks account for 65% of Uzbekistan's loans and 69% of its problem loans; in 2022 — 84% and 86%
Uzbekistan: share of state-owned banks in the loan portfolio and in banks' NPL, %
- in banks' NPL
- in the loan portfolio
The group's composition is set by the central bank and changes, including when state stakes are sold. For Kazakhstan and Kyrgyzstan, bank-level data are not yet in the dataset.
Source: Central Bank of the Republic of Uzbekistan, commercial bank NPL table, “Banks with state share” line.
Uzbekistan in detail
Problem loans of Uzbekistan's banks and microfinance organisations
How dollars are calculated
Dollar amounts are an indicative conversion by Avenex at the central bank's official rate on each snapshot date (as of 1 Sep 2026 – 11,801.23 sum per dollar); the central bank quotes them in sums. Changes over the period include an exchange-rate effect: over four years the dollar rate rose 8%, over the last year it fell 5%. In dollars, a change may differ from the change in sums in magnitude and even in direction.
01 – Four years · interactive chart
Total bank NPL grew 27% over four years, the loan portfolio – 69%Total bank NPL grew 37% over four years, the loan portfolio – 83%
Uzbekistan’s banks: NPL and portfolio growth, 01.09.2022 = 100, in dollars at the central bank rate on the snapshot dateUzbekistan’s banks: NPL and portfolio growth, 01.09.2022 = 100, in sums
Grey bars – months with no central bank file in the archive; lines are not extended across them. In “Growth” mode, the first available date of the period counts as 100. On the log scale, zero values are not shown.
Dollar amounts are an indicative conversion at the central bank's rate on each snapshot date; in dollars a change may differ from the change in sums in magnitude and in direction.
Source: Central Bank of the Republic of Uzbekistan, “Information on non-performing loans of commercial banks”, monthly data 1 Sep 2022 – 1 Sep 2026; historical – from the public web archive.
Data: Central Bank of Uzbekistan, “Information on non-performing loans of commercial banks”, 34 over 49 months (1 Sep 2022 – 1 Sep 2026); historical files taken from the public web archive and verified byte-by-byte.
Exchange rate: Central Bank of Uzbekistan, official rates archive, rate on the snapshot date.
Renames: the Microcreditbank and Saderat Bank Iran series are consolidated under their current names. Other banks that changed names (e.g., Kishlok Kurilish Bank → Business Development Bank, 2023) are shown as separate series.
02 – Year over year
The NPL share has barely changed over the year (3.75% → 3.70%), though in January it dipped to 2.99%
Uzbekistan's banks: NPL share in the loan portfolio, %, September 2022 – September 2026; independent of currency
In January 2024, 2025 and 2026 the share dipped and then rose again in the following months; there is no January 2023 snapshot in the archive. The central bank does not disclose the reason; an annual balance-sheet cleanup is our hypothesis. From April to September the share fell in 2024 and 2025, and in 2026 it rose by 0.5 pp; short windows in different years give different signs, so we compare year over year.
Source: Central Bank of the Republic of Uzbekistan, “Information on non-performing loans of commercial banks”, monthly data 1 Sep 2022 – 1 Sep 2026; historical – from the public web archive.
03 – Bank contributions · interactive chart
The six banks with the largest increases in NPL amounts added $326m of NPL over the year – more than the entire system ($289m)Change in bank NPL, 01.09.2025 → 01.09.2026
Change in bank NPL from 01.09.2025 → 01.09.2026, USD (equiv.)Change in bank NPL from 01.09.2025 → 01.09.2026, sum
- NPL growth
- NPL decline
- other banks, net
Dollar amounts are an indicative conversion at the central bank's rate on each snapshot date; in dollars a change may differ from the change in sums in magnitude and in direction.
The reasons for NPL declines (collection, write-off, sale, restructuring) are not disclosed in the central bank's data. A change in the NPL amount describes a bank's portfolio and does not assess the bank itself.
Source: Central Bank of the Republic of Uzbekistan, commercial bank NPL table for the selected dates.
04 – Share and volume · interactive chart
The largest NPL amounts are at the two biggest banks, with a share of 3.1–3.4%; a share above 8% occurs only at banks with loan portfolios below $2bnThe largest NPL amounts are at the two biggest banks, with a share of 3.1–3.4%; a share above 8% occurs only at banks with loan portfolios below 24tn UZS
Uzbekistan’s banks on 01.09.2026: loan portfolio (log scale, USD equivalent) and NPL share; circle area is the NPL amountUzbekistan’s banks on 01.09.2026: loan portfolio (log scale, sums) and NPL share; circle area is the NPL amount
- labelled banks: NPL share above 8% and the two largest NPL amounts
- other banks
- Агробанк 3.4% · $272m NPL
- Узнацбанк 3.1% · $269m NPL
- Микрокредит банк 9.2% · $181m NPL
- Анор банк 8.8% · $94m NPL
- Тибиси банк 13.4% · $87m NPL
- Пойтахт банк 13.2% · $8m NPL
- Мадад инвест банк 10.0% · $5m NPL
Zero in the source is shown on the zero line; it may mean either no NPL or an undisclosed value.
Source: Central Bank of the Republic of Uzbekistan, commercial bank NPL table on the snapshot date; latest data – as of 1 Sep 2026.
05 – Microfinance organisations
MFOs with an NPL share above 5% account for 7% of the sector's portfolio and 32% of its problem loans
Uzbekistan's MFOs as of 1 Sep 2026, groups by NPL share: share in the sector's portfolio ($1.58bn) and in its NPL ($39m)Uzbekistan's MFOs as of 1 Sep 2026, groups by NPL share: share in the sector's portfolio (18.6tn UZS) and in its NPL (456bn UZS)
- share in the sector's portfolio
- share in the sector's NPL
Of 141 MFOs, 35 did not disclose their NPL amount: 24 report zero in the source with a non-zero portfolio, and 11 have neither a portfolio nor NPL. The portfolio of the former is shown as a separate row.
Source: Central Bank of the Republic of Uzbekistan, “Share of problem loans of microfinance organisations” as of 1 Sep 2026. Bank and MFO figures are not summed.
Avenex hypotheses · to be tested on a lender's data
What this may mean for collections
- The collections workload is set by the amount. Over four years the share fell by 1.2 pp while the amount grew 27%.Over four years the share fell by 1.2 pp while the amount grew 37%.
- The January dip. If this is an annual write-off, then early in the year more written-off debts go to off-balance-sheet collections or sale. This can only be verified with a lender's data on the movement of overdue debt.
- Kazakhstan. Since April 2024, the NPL90+ amount has grown from 0.9 to 1.9 trillion tenge, while loans to the economy grew 49%. If this continues, the flow of collection cases will grow faster than the portfolio.
- A $169m reduction in NPL at four Uzbek banks.A 2.4tn UZS reduction in NPL at four Uzbek banks. Open data do not show how much was collected, written off or sold, and that determines the assessment of the debt-sale market.
Sources
Where the data comes from
- cbu.uzCentral Bank of the Republic of Uzbekistanproblem loans of banks and MFOs, loan portfolio, official exchange rates
- nbkr.kgNational Bank of the Kyrgyz Republicclassification of banking sector loans
- gov.kz/memleket/entities/ardfmAgency of the Republic of Kazakhstan for Regulation and Development of the Financial Marketstate of the banking sector: loans, NPL90+
- web.archive.orgInternet Archivehistorical copies of Central Bank of Uzbekistan publications
AI analyticsHow this material was prepared and how to use it
- AI analyticsThe data were collected by automated tools from open sources; the list is in the “Sources” section. The calculations, charts, headlines and conclusions were prepared and interpreted by automated tools, including artificial-intelligence models. Manual verification of every number and every wording is not guaranteed.
- Research purposeThe material was prepared to study open data and to practise methods and wording for analysing them. It is informational and may contain inaccuracies, extraction and conversion errors, and incomplete or outdated data.
- Sources and completenessHistorical files were obtained from the public web archive; some months are missing. Indicators, periods and groupings were chosen by the author; a different sample may lead to different conclusions.
- Not official statisticsThe material is not a publication of a central bank or of any other government body. In case of discrepancy, the primary source prevails.
- Not a recommendation or an assessment of organisationsThe material is not investment, credit, legal or tax advice, an offer, a rating, or a conclusion on the financial position, reliability, solvency or business reputation of any organisation or other person. A change in an organisation's published indicators reflects only the indicators themselves and does not characterise the quality of its management, its stability or its integrity. Organisations are highlighted in charts by formal criteria: the value of an indicator or its change. Do not use the material as a basis for decisions.
- Mention of organisationsOrganisation names are given as published by the regulator. A mention does not imply an assessment of an organisation or any affiliation with it. Neither the organisations that published the underlying data nor the organisations mentioned took part in preparing the material or approved it. Trademarks belong to their owners.
- CurrencyDollar amounts are an indicative conversion at official exchange rates on each snapshot date; the sources state them in national currency. In dollars a change may differ from the change in national currency in magnitude and in direction.
- HypothesesConclusions about what the indicators may mean for the market are editorial hypotheses about the market as a whole; they do not relate to specific organisations. Hypotheses and model estimates are labelled separately; they are not facts.
- IndependenceThe material was prepared in a personal capacity and solely from open data, without official, confidential or insider information and without anything constituting banking or commercial secrets. It contains no personal data. This is an independent Avenex project; it does not reflect the position of any organisations the author works for or cooperates with.
- No guarantee of timelinessThe material does not claim completeness or timeliness. The data are given as of the snapshot date and may be outdated; sources may revise published values, and page updates are not guaranteed. Collection, verification and interpretation of the data were performed automatically, so inaccuracies, omissions and outdated values are possible. Interactive charts and CSV exports are derivative data; when quoting, cite the primary source and the snapshot date. If the language versions diverge, the Russian version prevails.
- LiabilityThe material is provided “as is”, without guarantees of completeness, accuracy or currency. To the extent permitted by applicable law, Avenex is not liable for the consequences of its use.
- CorrectionsFound an error – write via the form at avenex.xronocode.com; the organisations mentioned can report an inaccuracy the same way. Corrections are made where possible after verification against the primary source.