How Exposure at Default Is Actually Computed in the Real World (A practical guide that connects PD and LGD into a full loss story)

-

In the previous guides, we did two important things:

  • We showed that PD is about loans migrating into 90+ DPD

  • We showed that LGD is about cash recovery after default, adjusted for time

There is one missing piece that quietly holds both together.

Before a loan defaults.
Before recovery even begins.
Before loss can be measured.

We must answer one simple but slippery question:

How much money will actually be outstanding when default happens?

That number is Exposure at Default (EAD).

This guide explains EAD the way it is built and used in real ECL models, not the way it is defined in textbooks.

First, kill the most common misunderstanding

EAD is not:

  • Sanction amount

  • Original disbursed amount

  • Current outstanding

  • A single fixed number

In real models:

EAD is a month-wise path of outstanding balances, not a point estimate.

Why?

Because default does not happen “today” for all loans.
It happens somewhere in the future, and outstanding changes every month.

Where EAD sits in the loss flow

Think of credit loss in this strict sequence:

  1. Loan is alive today

  2. Outstanding reduces every month due to amortisation

  3. Default may happen in any future month

  4. Whatever is outstanding in that month becomes the base for loss

  5. PD, LGD, and discounting act on that base

EAD lives between today and default.

Step 0: Fix the universe (again)

EAD is calculated only for loans that are alive today.

So your starting table always looks like this:

Table 1: Alive portfolio snapshot (today)


At this stage:

  • No one has defaulted yet

  • PD has not been applied

  • LGD is irrelevant

This is pure exposure mechanics.

Step 1: Accept the core truth about EAD

Exposure changes every month.

For most retail and MSME loans:

  • EMI is fixed

  • Principal repayment increases over time

  • Outstanding comes down

So EAD cannot be “picked”.
It must be projected.

Step 2: Build the amortisation path (this is the heart)

In real models, this is done using:

  • Weighted IRR

  • Remaining tenure

  • PPMT logic

Conceptually, the output looks like this.

Table 2: Monthly outstanding path (one loan)



This table is not yet ECL.
This is just exposure evolution.

But this table is critical.
Everything else multiplies onto this.

Step 3: EAD is this entire path, not one row

A very common mistake is to ask:

“Which row is EAD?”

That question itself is wrong.

EAD is the whole column of future outstanding balances.

Why?

Because:

  • Default can happen in Month 2

  • Or Month 7

  • Or Month 11

Each possibility has a different exposure.

So EAD must exist for every future month.

Step 4: Portfolio-level EAD (how models actually scale)

At portfolio level, we don’t track one loan.

We track thousands of loans, bucketed by:

  • DPD bucket

  • Tenure band

  • Product type

So EAD is computed as average outstanding paths.

Table 3: Average EAD path by DPD bucket


Notice something subtle but important:

  • Riskier buckets amortise less before default

  • Their EAD remains closer to today’s outstanding

This is why EAD is behavioural, even though it looks mechanical.

Step 5: What EAD is not trying to do

EAD does not try to:

  • Predict the exact month of default

  • Guess borrower behaviour explicitly

  • Override PD

Those jobs belong elsewhere.

EAD only answers:

If default happens in a given future month, what is the outstanding then?

Nothing more.

Step 6: How EAD quietly connects to PD and LGD (without mixing)

At this stage:

  • PD has not been applied

  • LGD has not been applied

  • Discounting has not been applied

EAD remains clean and independent.

This separation is intentional.

Because once EAD is polluted with probability or loss assumptions, it becomes untestable.

Step 7: Why this approach works in real ECL builds

This EAD construction:

  • Works for Stage 1 and Stage 2

  • Extends naturally from 12-month to lifetime

  • Aligns cleanly with marginal PD (later)

  • Avoids point-estimate traps

  • Is auditable and explainable

That’s why this is the industry default.

Step 8: Common shortcuts (and when they break)

Some teams shortcut EAD as:

“EAD = current outstanding”

This only works when:

  • Tenure is very short, or

  • Defaults happen almost immediately

For most portfolios, it:

  • Overstates exposure for safe loans

  • Understates exposure for risky loans

Which silently distorts ECL.

The one idea to take home

EAD is not today’s number.
It is tomorrow’s outstanding, mapped month by month.

It is the bridge between probability and money, but it must stay neutral until the very end.

Why this guide stopped exactly here

This article intentionally did not show:

  • How marginal PD is applied month-wise

  • How LGD multiplies onto EAD

  • How discount factors enter

  • How Stage 1 vs Stage 2 differs numerically

Because the moment we do that,
we are no longer explaining EAD.

We are computing ECL.

That is the next guide.

Comments

Post a Comment

Popular posts from this blog

What Expected Credit Loss (ECL) Really Means — Through My Journey in Credit

-
I didn’t start my career in risk modelling. I’m an accountant by training. My early years were spent in accounting and finance roles, working on things like OPEX and CAPEX planning, budgets, and variance tracking. Numbers, yes. But not credit models. Statistically, my toolkit was basic. Mean, median, standard deviation. Some idea of correlation, regression, and probability from textbooks. Nothing fancy. No machine learning. No advanced econometrics. So when I say this, I mean it honestly: If I could understand ECL and work with it comfortably, almost anyone in finance can. ECL looks complex mainly because of how it’s presented, not because of what it’s trying to do. At its core, ECL is built around one very practical question: If this borrower fails to pay in the future, how much will I realistically lose? Everything else exists only to answer this question in a structured, regulator-acceptable way. Before we talk about ECL, let’s talk about provisioning When I was in core f...
Read more

Credit Cost in Loan Pricing: A Practical, End-to-End Explanation

-
Image
At a high level, loan pricing looks like this: We have fair idea about all the components except for credit cost. Let’s assume everything except credit cost  is already known. This article is only about how to think about, estimate, and sanity-check credit cost — in a way that actually works in real lending. What credit cost really represents Credit cost answers one very boring but very important question: “Out of all the money I lend, how much will I not get back?” It is not : GNPA, write-offs, stress loss, worst-case loss. It is the average, expected loss baked into pricing. If you don’t price this correctly, everything else in loan pricing becomes meaningless. The irreducible formula (don’t fight it) Credit cost has only two moving parts: Credit Cost = PD × LGD Where: PD (Probability of Default) Out of 100 similar loans, how many will default at least once? LGD (Loss Given Default) If a loan defaults, what % of the outstanding amount will I ultimate...
Read more