Vietnam’s banks should test cash-flow underwriting for small and medium-sized enterprises (SMEs) through limited portfolios, with loan amounts based on verified cash flows and stress-tested repayment capacity. Expansion should depend on mature credit outcomes and commercial viability. SME credit exceeded VND 4.1 quadrillion ($158 billion) by 28 August 2026, up about 12.4% from end-2025. Resolution 138/NQ-CP encourages lending based on payment and cash-flow data. The test is whether verified transaction data support viable lending with less collateral.

The OECD’s 2025 Economic Survey of Viet Nam identifies inadequate collateral as a barrier to SME borrowing, although this does not establish that affected firms are otherwise creditworthy. My survey-weighted calculations from the World Bank’s 2023 Vietnam Enterprise Survey show that 33.3% of small establishments used bank finance for working capital, against 58.3% of large establishments, using its employment-size definitions. The joint test of credit constraints across small, medium and large establishments was statistically inconclusive; lower borrowing alone does not establish unmet demand. The survey covers registered establishments with at least 1% private ownership and five or more workers in manufacturing and selected services, excluding informal businesses. Most financial responses concern 2022, so these results cannot assess recent reforms.

Assess repayment capacity before approval

Regional experience offers an operational lesson. Alliance Bank in Malaysia describes checking bank statements for tampering and reconstructing balances before credit assessment. In a February 2026 interview with The Asian Banker, the bank reported gross impairment of around 1% in its Digital SME portfolio. That portfolio includes different loan types and guarantee arrangements, so the figure cannot establish the effect of cash-flow assessment alone. Vietnam’s pilots should similarly establish the reliability of transaction records. More data help only if banks can interpret and verify them.

Start with one borrower segment and product, such as working-capital facilities for established SMEs with verifiable transaction histories but limited acceptable collateral. Assess each applicant under existing rules and record whether cash-flow assessment changes approval, amount or terms. Reconcile bank statements with invoices or other available records, investigate unexplained flows and distinguish operating receipts from loan proceeds, owner injections and transfers between accounts. Withhold approval where repayment capacity cannot be substantiated. Results from firms with reliable records should not be generalised to all SMEs.

Assess cash available after operating costs, taxes and essential investment, accounting for existing debt payments and information from credit registries. Test repayment capacity against plausible falls in receipts, delayed customer payments and higher costs. Set loan amounts, tenor and repayment schedules against those scenarios and the business’s cash-conversion cycle, with buffers calibrated to the segment. For a seasonal distributor, instalments based on peak-season receipts could exceed repayment capacity in quieter months. Collateral or guarantees can supplement the assessment but should not replace a credible repayment plan.

Before launch, obtain the bank’s required approvals for eligibility criteria, exposure and concentration limits, and thresholds for pausing expansion when arrears, fraud or losses breach risk appetite. Independently validate any scoring model and record manual overrides. Monitor receipts and payment behaviour after disbursement, with human review when risks deteriorate. Preserve normal classification and provisioning requirements, allow data correction and human reconsideration of adverse decisions, and handle personal data under applicable Vietnamese law.

Expand only on comparable evidence

Define a comparison portfolio at launch using contemporaneous conventional lending in similar borrower and product segments. Compare origination cohorts at equal months on book, accounting for differences in borrower selection, collateral and guarantees. Such comparisons can inform scaling but cannot isolate the underwriting method’s causal effect. Track arrears, restructuring, write-offs, recoveries and net losses, allowing time for deterioration to emerge.

More approvals alone would not demonstrate success. Track disbursements, borrowing costs and whether borrowers are new to the bank or have no previous business loan in available credit records. Verification and servicing costs matter: additional checks may improve assessment while making small loans uneconomic. Compare returns after funding, operating costs and credit-loss expense with the bank’s required return on allocated capital. Better access is meaningful for banks only if the additional lending remains viable after these costs and risks.

The State Bank of Vietnam could coordinate common reporting definitions and publish aggregated access, pricing and credit outcomes separately for cash-flow lending pilots and comparable conventional SME portfolios, subject to disclosure controls. This would allow comparisons without disclosing bank-level model diagnostics or profitability. Where public support applies, report guarantee claims, recoveries and subsidy outlays separately from bank returns. Otherwise, guarantees may make bank results appear stronger by absorbing losses without improving underlying underwriting. Expansion should depend on useful customer outcomes, acceptable credit losses and commercial viability after adequate loan seasoning and supervisory review.