DexLab Analytics

A lot of credit risk discussion treats "will this default?" as the whole question. It's only half of it.

The other half — how much will we actually lose if it does — is Loss Given Default, and it's arguably the harder number to get right.

We laid out why this matters, with the Lehman Brothers recovery story as a real-world example of how fast these assumptions can move.
Full piece linked below — curious what others in risk and banking think.
https://www.dexlabanalytics.com/blog/why-is-loss-given-default-important

Loss Given Default (LGD): How much does a lender actually lose when a borrower defaults?

A borrower defaulting does not necessarily mean the lender loses the entire outstanding amount.

The eventual loss depends on factors such as:

• Recovery rates
• Collateral value
• Time to recovery
• Workout and collection costs
• Default and recovery characteristics

This is where LGD modeling becomes an important component of credit risk assessment and expected loss estimation.

In our latest practical guide, we explore how LGD can be estimated, how recovery rates influence the model, and the key considerations involved in building an effective LGD framework.

📖 Read the complete guide on DexLab Analytics.
https://www.dexlabanalytics.com/blog/lgd-recovery-rates-estimation-guide

Freedom inspires.
Knowledge transforms.
India progresses.

Happy 80th Independence Day 🇮🇳

Indian banks have under 8 months until ECL.

RBI's directions take effect 1 April 2027. Most teams are scoping it as a provisioning change.

It's a modelling change — and that's a different project entirely.

Basel asked for one number: a 12-month PD, through-the-cycle, used once a quarter for capital.

IFRS 9 asks for a curve: lifetime PD for every Stage 2 exposure, point-in-time, flowing into the P&L every quarter.

The logistic scorecard most banks spent years validating produces one number at one horizon. Getting to a lifetime curve needs a survival layer for the term structure and a macro overlay for point-in-time conditioning.

Three models where there used to be one.

Full breakdown on the blog. Link in bio.
https://www.dexlabanalytics.com/blog/pd-estimation-methods-for-credit-risk

Bad loans in Indian banks just hit their lowest level in decades. Here's what that number isn't telling you.

RBI's own data shows unsecured retail lending — personal loans, credit cards — now drives over half of all new loan defaults. At private banks, it's nearly 3 in 4.

We broke down the numbers straight from RBI's Financial Stability Reports — no guesswork, no inflated stats.

See the full picture: https://www.dexlabanalytics.com/blog/credit-risk-in-indian-banking-rbi-data-analysis

Quick question: can you explain the difference between counterparty risk, concentration risk, and settlement risk — without pausing to think?
If not, you're not alone. Most people equate "credit risk" with loan defaults. That's only one piece of it. It shows up in derivatives, bond markets, trade settlements, and interbank lending too — anywhere one party depends on another to deliver.
Full guide here: https://www.dexlabanalytics.com/blog/what-is-credit-risk-understanding-credit-risk-definition-and-types

Worth bookmarking if you're building your foundation in risk analytics.

A bank that underestimates default risk lends into losses it never provisioned for. A bank that overestimates it loses good customers to competitors with sharper pricing. Credit risk modeling is the only thing standing between those two failure modes.