Expected Credit Loss (ECL) Model
Why a lender now has to book a loss on a loan that's being repaid perfectly on time, every single month, without a single missed payment — just because the loan exists.
In plain English
Imagine a lender has given out 10,000 loans. On any given day, the vast majority are being repaid perfectly on schedule — no missed payments, no warning signs. Old-fashioned accounting logic said: don't book a loss on a loan until there's actual evidence something has gone wrong — a missed payment, a bounced cheque, a default. Modern accounting takes a very different, forward-looking view: even a perfectly-performing loan carries some statistical probability of going bad in the future, purely because lending inherently involves risk, and that expected future loss should be recognised TODAY, as soon as the loan is made — not only after the first missed payment. This shift, from waiting for bad news to pricing in expected bad news upfront, is what the Expected Credit Loss model is about.
Words you'll need first
The estimated likelihood, usually a percentage, that a borrower will fail to repay within a given period, based on historical data, credit scores, and current economic conditions. A well-secured home loan to a salaried borrower might carry a PD of well under 1%; an unsecured personal loan to a higher-risk borrower might carry a PD many times higher — this number is the starting point for estimating how much loss to provide for, even on a loan that hasn't missed a payment.
If a borrower DOES default, how much of the outstanding amount does the lender actually expect to lose, after whatever it can recover by selling collateral or pursuing legal recovery? A fully secured loan against a property might have a low LGD; an unsecured loan might have a much higher LGD. Multiplying PD × LGD × the outstanding loan amount gives the Expected Credit Loss for that loan — the number that actually gets provided for.
Before vs. now
How it used to happen
Under the older 'incurred loss' model (used under previous Indian GAAP and the old international standard IAS 39), a lender only recognised a provision once there was objective evidence that a loss event had already occurred — a missed payment, a covenant breach, financial difficulty at the borrower. Perfectly-performing loans, however statistically risky the overall book, carried little or no provision at all. This approach came under heavy criticism after the 2008 global financial crisis, when banks worldwide were seen recognising loan losses 'too little, too late' — provisions only showed up once the damage was already visible.
How it's done now
Ind AS 109 (aligned with the global post-crisis reform, IFRS 9) replaced this with the forward-looking Expected Credit Loss model. From the moment a loan is originated, the lender must estimate and provide for expected future losses using Probability of Default and Loss Given Default, even if the loan is performing perfectly. As a loan's credit risk deteriorates over its life — even before an actual default — the required provision increases in stages (often summarised as Stage 1, 2, and 3, reflecting performing, significantly-deteriorated, and credit-impaired loans respectively), meaning provisions build up progressively as warning signs emerge, well before an actual missed payment.
A worked example, with numbers
| Item | Value |
|---|---|
| Outstanding loan book | ₹1,000 crore |
| Probability of Default (PD) | 1.5% |
| Loss Given Default (LGD) | 25% |
| Expected Credit Loss = ₹1,000 cr × 1.5% × 25% | ₹3.75 crore |
- ₹3.75 crore gets provided for and expensed TODAY, even though every one of these home loans is being repaid on time, with zero actual missed payments — a purely forward-looking, statistical estimate of future loss on an otherwise healthy loan book.
- If economic conditions worsen — say the estimated Probability of Default rises from 1.5% to 3% — the required provision roughly doubles to about ₹7.5 crore, hitting the P&L immediately, even though not a single additional loan has actually defaulted yet.
- If a specific pool of loans shows early warning signs, those loans can move from Stage 1 to Stage 2, at which point provisioning often shifts from a 12-month expected loss estimate to a full-lifetime expected loss estimate — well before any of them are technically in default.
- This is why lenders' provisioning charges can rise sharply during an economic slowdown even while reported Gross NPA percentage — which only counts loans that have ACTUALLY gone bad — barely moves: ECL is designed to front-run the bad news, not just report it after the fact.
What it does to the financial statements
Impact on the P&L
- Expected Credit Loss provisions are charged to the P&L as soon as a loan is originated, and re-estimated every reporting period — a lender's profit is directly affected by changes in its own credit risk assumptions, not just by actual defaults.
- A deteriorating macroeconomic outlook can force a lender to increase ECL provisions and take a P&L hit purely on forward-looking grounds, even while its existing loan book continues to perform normally.
- Because ECL is inherently model-and-assumption-driven, provisioning charges can be genuinely volatile quarter to quarter, and are one of the more heavily scrutinised line items in a lender's results — different lenders can reach different ECL conclusions on economically similar loan books.
- A loan moving into Stage 3 (credit-impaired) typically triggers a much larger, often near-full, provision against that specific exposure, on top of whatever general ECL provision already existed.
Impact on the Balance Sheet
- Loans are shown on the balance sheet net of their ECL provision — the reported 'Loans and Advances' figure is already reduced by the lender's own best estimate of expected future losses, not the full contractual amount owed.
- A rising ECL provision reduces net loan assets and, through the matching P&L expense, reduces shareholders' equity — directly affecting a lender's capital adequacy ratios, closely watched by regulators like the RBI.
- Because provisioning is forward-looking, a lender's balance sheet can look weaker during a period of deteriorating economic expectations, even before any borrowers have actually stopped paying — a genuinely different signal from the old incurred-loss approach.
- Comparing two lenders' balance sheets purely on gross loan book size, without checking their respective ECL provisioning levels and assumptions, can be misleading — more conservative provisioning holds back more of a lender's reported net worth against future risk.
Which standard covers this
In India this is governed by Ind AS 109 – Financial Instruments, notified under the Companies (Indian Accounting Standards) Rules and applicable to NBFCs, housing finance companies and other Ind AS-reporting lenders. Notably, commercial BANKS in India have not yet transitioned to Ind AS at all — the Reserve Bank of India has repeatedly deferred Ind AS implementation for banks, so Indian banks currently continue to follow their own regulator-specific Income Recognition and Asset Classification (IRAC) provisioning norms under RBI guidelines, a genuinely different, though directionally similar, framework from the Ind AS 109 ECL model used by NBFCs and other Ind AS filers.
How it's recognised globally
Globally, the equivalent is IFRS 9 – Financial Instruments, issued as a direct response to criticism of the pre-2008 incurred-loss model, and Ind AS 109 closely mirrors its ECL framework, including the three-stage approach described above. The United States took a related but distinctly different path: US GAAP's equivalent, under ASC 326, is known as CECL (Current Expected Credit Losses), and while it shares the same forward-looking philosophy, it doesn't use IFRS 9's staged (12-month vs lifetime) approach — CECL generally requires lifetime expected credit losses to be recognised from day one for essentially all in-scope financial assets, a somewhat more front-loaded and conservative approach than IFRS 9/Ind AS 109's staged model. A US bank and an Indian NBFC holding economically similar loan books can end up with genuinely different Day-1 provisioning levels, purely from this structural difference.
Real example — Indian listed company
Among large Indian NBFCs that report under Ind AS and therefore apply the full Expected Credit Loss model described in this article, Bajaj Finance is one of the most closely tracked by analysts, given the scale and diversity of its lending book across consumer durables financing, personal loans, and other retail credit products. Its quarterly results routinely draw attention to provisioning trends and Stage 1/2/3 loan classifications, since shifts in these — driven by changes in the assumed Probability of Default, Loss Given Default, or macroeconomic overlay adjustments applied on top of the base model — can move reported profit meaningfully in either direction, independent of how the underlying loan book is actually performing on a day-to-day collections basis. This is a useful real-world lens for the core lesson of this article: an NBFC's headline profit can be shaped as much by its own forward-looking credit risk assumptions and provisioning philosophy as by the literal repayment behaviour of its borrowers that quarter — exactly why analysts covering lenders read the provisioning and asset-quality notes as closely as the profit number itself.
Where you'll see this
Related concepts
Hedge Accounting
Why an airline that locks in its jet fuel price months in advance can end up reporting a 'loss' on that smart, protective decision — unless it uses a specific accounting technique designed to stop that mismatch from happening.
Fair Value Hierarchy (Level 1, 2 and 3)
Why two investments on the same balance sheet, both labelled 'fair value', can carry wildly different levels of confidence — one priced off a stock ticker updated every second, the other based on a spreadsheet model nobody outside the company can fully verify.
Preference Shares — Equity or Liability?
Why something literally called a 'share', sitting in a company's own share capital register, can be accounted for as a LOAN on its balance sheet — not as part of shareholders' equity at all.