Written from the work, not about it
Each of these exists because the question came up while building the tool and the answer had to be settled before anything could be shipped. Every figure on them is the figure the calculator uses, rendered from the same constant — so if the engine’s reading changes, the page changes with it.
- What the RBI's 2026 Directions actually say about provisioning floors — and five places a careful reading changes the answer
The product-wise minimum provisioning schedule in paragraph 82, read against the instrument rather than against an analysis of it, with the five differences that reading found and what each one is worth on a real book.
For: A preparer or model-risk reviewer who has to defend a floor to a supervisor, and wants the paragraph reference beside every rate.
Source: RBI (Commercial Banks — Asset Classification, Provisioning and Income Recognition) Directions, 2026 — RBI/DOR/2026-27/398, paragraph 82.
- Why your base scenario should not sit at Z = 0
Anchoring the central case at the average state of the economy biases the allowance downwards on every book, and the size of it scales with asset correlation — so it is worst on the exposures a bank has most of, and at low correlation it reverses.
For: Anybody setting scenario weights, and anybody who has to defend them to a model-risk committee.
Source: IFRS 9.5.5.17(a), and the single-factor Vasicek model underlying the Basel IRB risk-weight formulas.
- How to calibrate a PD when your book has never defaulted
Zero defaults is data, not an absence of it, and there is a standard supervisory answer to what it is worth — computed here on the same function the calibration screen uses, with the judgement it cannot avoid published rather than buried.
For: Anybody holding a clean book who has to produce a defensible PD, and anybody whose auditor has asked where the number came from.
Source: Pluto & Tasche (2005), Estimating Probabilities of Default for Low Default Portfolios; Clopper & Pearson (1934), the exact binomial interval.
- What the ECL transition does to capital, and how long you have
The day-one adjustment to retained earnings, the CET1 add-back net of tax, and a glide path that runs four years and ends — worked through on an illustrative book by the same engine that would run it on yours.
For: A board or a capital planner who needs the basis-point answer, and the treasurer who has to make the ratio work in 2031 when the relief is gone.
Source: RBI (Commercial Banks — Asset Classification, Provisioning and Income Recognition) Directions, 2026, transitional arrangements — transcribed from published analysis, NOT read from the primary instrument.
- Tie the tape to the ledger before you measure anything
The amounts that go missing from an extract are not random — they correlate with the exposures that are unusual, and unusual exposures are where the loss is. Four worked failures, each diagnosed by the tool itself, and one that ties and is still wrong.
For: Anybody importing a loan tape or a receivables ledger, and anybody who has been asked why their population does not agree with the accounts.
Source: The reconciliation step in this product, and IFRS 9's requirement that expected credit loss be measured on the whole population of financial assets in scope.