Will Grab's ECL provisions exceed 5% of its gross loan book in any 2026 quarter?

The 5% quarterly provision flow threshold is quite high. The current provision stock is ~$100M on $1.3B (~7.7%), but that is cumulative, not quarterly. A 5% quarterly flow on a $1.5-2B book would mean $75-100M in new provisions in a single quarter — a catastrophic credit event. Management claims risk-adjusted returns exceed cost of capital and has proprietary data from ride-hailing/delivery transactions for credit scoring. Absent a major macro shock in Southeast Asia, provisions are more likely to run at 2-3% of the loan book quarterly. However, the untested borrower base (1/3 new-to-credit) creates genuine tail risk that is hard to quantify.

ECL provisions exceeding 5% of the gross loan book in a single quarter would represent a crisis scenario — credit models fundamentally failing. While the credit models are untested through a full cycle, Grab has several protective factors: short-duration loans (most are small personal/micro-business loans), proprietary behavioral data, and the ability to throttle disbursements quickly. The SEA macro environment would need to deteriorate significantly (currency crisis, recession) to trigger 5%+ quarterly provisions. This is a tail risk, not a base case. However, the IFRS 9 framework can produce provision spikes if management revises macro assumptions.

The loan book is growing from $1.3B to $2B+ with disbursements at $3.5B annualized. The velocity of growth means the portfolio is young — most loans are recent and haven't had time to season. Credit losses typically appear with a lag, and the rapid growth dilutes the NPL ratio even as absolute bad debts rise. A provision spike could occur when the book matures — potentially H2 2026 when earlier cohorts season. The 1/3 new-to-credit segment is the wildcard. Given the untested nature and IFRS 9 mechanics, I assign slightly higher probability than pure tail risk but still well below 50%.

5% of gross loan book in a single quarter is an extremely high bar for provisions. On a $2B book that is $100M in one quarter. This would require a severe credit event. Management has shown strong execution discipline and would likely throttle disbursements before allowing provisions to spike that high. The proprietary data advantage (years of transaction history for borrowers) provides better scoring than traditional bureau data. However, untested borrower segments in an untested economic environment create irreducible uncertainty.

The rapid scaling concerns me more than other models suggest. Disbursements at $3.5B annualized with 1/3 new-to-credit borrowers means roughly $1.2B per year going to borrowers with no credit history. Even if loss rates are 'only' 5-8% on this segment, that is $60-96M — which on a quarterly basis could push total provisions near the 5% threshold depending on portfolio concentration. IFRS 9 requires forward-looking macro adjustments, and if any SEA economy shows stress, the provision model could produce a spike. I rate this higher than a pure tail event.

Historical digital lending experience in Southeast Asia (from incumbents like Sea Group's SeaMoney) suggests provision rates of 2-4% of book are typical. Exceeding 5% would be anomalous absent a macro shock. Grab's data advantage from platform usage should produce at least average or better credit outcomes. The question is whether rapid scaling into underbanked segments changes the credit profile materially. I think 5% is a high bar that requires a specific catalyst (recession, currency crisis) to be breached.

5% quarterly provision flow is a very high bar. Management has proprietary data advantage and track record of disciplined execution. Would need a macro shock in SEA to reach this level. Tail risk but not base case.

Rapid loan book growth to $2B means young portfolio that hasn't seasoned. 1/3 new-to-credit borrowers elevate risk. IFRS 9 can produce lumpy provisions. But 5% is still a high bar — more likely 2-3% in normal conditions. Slightly above pure tail risk due to untested borrower base.

Credit models untested through cycle, but base case is provisions staying 2-3%. Management would throttle growth before allowing 5%+ provisions. Short-duration loans limit exposure. Low probability tail risk.