T&S View: cash-flow lending has promise, but rate caps and low account access constrain it
This article is T&S’s view, separate from the facts it cites.
Correction: Confirms the links and dates of the IFC Board statement and the Human Rights Watch article. Adds counterarguments and uncertainties: an illustrative per-loan economics calculation under the rate cap, the reach of account ownership and underwriting data, and industry and supervisor positions. Headline recalibrated to the strength of the evidence.
Key points
- Regulation and international scrutiny are headwinds for land-collateral and over-lending models
- Under a rate cap, small short-tenor loans can lose money per loan, depending on unit costs
- Account ownership is 39.0% (2024); the population reachable by bank-statement underwriting remains limited
Facts
This article is T&S's view. We set out the facts first.
On 24 June 2026 the IFC Board concluded that IFC had not breached its environmental and social policies, in response to the investigation report by the CAO (IFC's independent accountability mechanism) on IFC investments in six Cambodian MFIs. It approved an action plan that includes support such as loan restructuring for the 18 complainants, support for establishing a financial ombudsman, and improved responsible-lending certification for about 65,000 loan officers 1. Human Rights Watch (June 2026; the URL is dated 2026-06-26) criticised the decision, noting that IFC had invested over USD 400m over a decade in Cambodian MFIs.
The NBC–UN plan had completed 10 of 22 actions as of September 2025, with responsible pricing and data sharing outstanding (see separate article).
According to the World Bank Global Findex, adult account ownership was 39.0% in 2024 (up from 33.4% in 2021) 2. ILO estimates put wage and salaried workers at 44.5% of total employment in 2025 3.
Analysis
T&S expects lenders increasingly to be asked to explain how they verified repayment capacity rather than what collateral they took. Starting small on the basis of payroll and account-flow history, and extending limits as repayment history builds, can extend credit while avoiding over-indebtedness.
Three constraints, however, weaken this view.
a) Rate cap and unit economics. The NBC Prakas on Interest Rate Ceiling on Loan (13 March 2017) remains in the published legislation list. The Prakas sets the cap at 18% a year for any loan maturity; it applies to MDIs, MFIs and rural credit operators, for new loan contracts (including restructured and refinanced loans) from 1 April 2017, and its text does not mention fees 4. As an illustration, assume that level (excluding fees), equal instalments, a funding cost of 7% a year, underwriting cost of USD 5 per loan (identity checks, bureau enquiry and assessment), collection and servicing cost of USD 4 per loan and an expected loss of 3% of principal, excluding head-office overheads. For a loan of USD 300 over six months, interest income is about USD 15.9. Against it stand the funding cost of USD 6.2 and per-loan costs of USD 18: underwriting USD 5, collection USD 4 and expected loss USD 9. The result is about −USD 8.3 per loan (interest income minus funding cost minus per-loan costs) 5. At USD 300 over twelve months the loan roughly breaks even (about +USD 0.4); at USD 500 over twelve months it earns about +USD 6.6; and at USD 300 over twelve months with expected loss cut to 1% it earns about +USD 6.4. These are illustrative assumptions, not observed data. Small, short-tenor lending under the cap works only if automation cuts fixed costs and losses stay low.
b) Reach of underwriting data. Account ownership is 39.0%, and most of the employed are not wage earners. The share of workers paid into an account could not be confirmed in primary sources (unverified). The NBC semi-annual report counts 39.6 million deposit accounts and 4.7 million loan accounts in June 2026, but individuals hold multiple accounts, so account counts do not measure people. The population reachable by cash-flow underwriting is limited to a subset of wage earners.
c) Industry and supervisor counterarguments. The IFC Board did not find a policy breach. CMA reports that its members restructured USD 500m of loans for displaced households between mid- and end-2025 and waived USD 10.6m in interest (16 January 2026) 6. The Association of Banks in Cambodia (ABC) and CMA said in a joint statement on 16 March 2026 that sector soundness indicators exceed regulatory requirements. The NBC received 721 complaints and enquiries in 2025, of which 683 were resolved. Any judgement should weigh these industry and supervisor positions, not only NGO criticism.
Implications
Counterpoints and uncertainties
With the payroll-deposit share unconfirmed, the size of the reachable population is uncertain. How the IFC Board's decision will affect development-finance flows and terms is unclear. The timing of data-sharing and DTI rules is also unknown. The NBC Annual Report 2025 says the NBC is preparing and reviewing Prakas and guidelines on interest rate ceilings, so the cap and its treatment of fees could change.
Figures: see definitions
- Loan officers targeted for responsible-lending certification: about 65,000. Loan officers covered by the certification improvement in the IFC action plan. Cambodian microfinance loan officers. Basis date: 24 Jun 2026. Source: IFC Board statement on the CAO investigation report on Cambodia microfinance (2026-06-24)
- Account ownership (adults): 39.0% (up from 33.4% in 2021). Share of adults (age 15+) with an account at a financial institution or mobile-money provider. Population age 15+. Basis date: 31 Dec 2024. Source: World Bank Global Findex: account ownership (FX.OWN.TOTL.ZS), Cambodia
- Wage and salaried workers (% of employment): 44.5% (44.2% in 2024). ILO modelled estimate. Total employment. Basis date: 31 Dec 2025. Source: World Bank WDI (ILO modelled estimate): wage and salaried workers (SL.EMP.WORK.ZS), Cambodia
- Loan interest-rate cap (NBC Prakas, 2017): 18%. Maximum interest rate per annum for any loan maturity (Article 4). New loan contracts, including restructured and refinanced loans, signed from 1 April 2017 by MDIs, MFIs and rural credit operators (Articles 3 and 5). Basis date: 13 Mar 2017. Source: NBC Prakas B7-017-109 Prokor on Interest Rate Ceiling on Loan (13 March 2017), unofficial English translation
The Prakas does not mention fees, commissions or penalties, so whether fees count toward the ceiling cannot be determined from its text.
- Scenario A per-loan costs excluding funding: USD 18. Underwriting USD 5 + collection USD 4 + expected loss USD 9. Illustrative loan, not observed data. Basis date: 27 Sep 2026. Source: T&S illustrative calculation (assumptions, not observed data)
T&S calculation. Equal monthly instalments at an assumed 18% p.a. cap, fees excluded; funding cost accrued monthly on the opening balance; head-office overheads excluded; illustrative assumptions, not observed data.
- Interest waived for displaced households (CMA): USD 10.6m. Interest waived by members. CMA member institutions. Basis date: 31 Dec 2025. Source: CMA press release: support for displaced households (2026-01-16)
Sources
- IFC: Statement from IFC Board on the CAO Investigation Report on Cambodia Microfinance(2026-06-24)primary
- NBC: Prakas on Interest Rate Ceiling on Loan(2017-03-13)primary
- NBC: Annual Report 2025 (2026-01-23)primary
- NBC: Financial Stability Review 2025(2026-03-28)primary
- World Bank: Account ownership, Cambodia(Global Findex)primary
- World Bank: Wage and salaried workers, Cambodia(ILO推計)primary
- CMA: Microfinance sector support for displaced households(2026-01-16)primary
- ABC・CMA: Resilience and stability of Cambodia's financial sector(2026-03-16)primary
- Human Rights Watch: World Bank's IFC rejects microfinance harm findings(2026-06)
- Xinhua: Cambodia's banking sector as of June(2026-07-28、NBC半期報告)
- Cambodia Investment Review: Third NBC–UN convening(2025-09-25)
Reports are for information only and are not investment advice. Methodology: sources, verification and definitions
A Japanese version of this article is available: 日本語版