If you work in finance or accounting, you’ve probably heard about IFRS 9 and how it’s changed the way we handle credit losses. One of the trickiest parts? Calculating the Probability of Default, or PD for short. Don’t worry—it sounds more complicated than it actually is. Let me walk you through it in plain English.
What Exactly Is Probability of Default?
Think of Probability of Default as your best educated guess about whether a borrower will fail to pay back their loan. It’s basically asking: “What are the chances this person or company won’t be able to meet their payment obligations?”
Under IFRS 9, you can’t just wait until someone defaults to recognize a loss. You need to estimate potential losses upfront, and PD is a crucial piece of that puzzle.
Why PD Matters for IFRS 9
IFRS 9 introduced a three-stage model for recognizing credit losses. Your PD calculation helps determine which stage a loan falls into:
- Stage 1: Everything looks fine. The credit risk hasn’t increased significantly since you first made the loan.
- Stage 2: Warning signs are appearing. Credit risk has increased, but no default yet.
- Stage 3: The borrower has defaulted or is showing serious signs they can’t pay.
Getting your PD right means you’re setting aside the correct amount of money for potential losses, not too much, not too little.
How to Calculate PD: The Basics
Let’s break this down into manageable steps.
Step 1: Gather Your Historical Data
Start by looking at your past lending experience. How many borrowers actually defaulted? This historical data is your foundation.
For example, if you lent money to 1,000 customers over the past five years and 50 of them defaulted, your basic historical default rate is 5%.
Step 2: Define What “Default” Means
This might seem obvious, but you need a clear definition. Most institutions consider a borrower in default when:
- Payments are more than 90 days overdue
- There’s clear evidence they can’t meet their obligations
- You’ve had to write off part or all of the loan
Make sure everyone in your organization uses the same definition.
Step 3: Consider the Time Horizon
PD isn’t just one number—it changes depending on your timeframe. You’ll typically need:
- 12-month PD: The probability of default within the next year (used for Stage 1 loans)
- Lifetime PD: The probability of default over the entire life of the loan (used for Stage 2 and Stage 3)
A borrower might have a 2% chance of defaulting in the next 12 months, but a 15% chance over a 10-year mortgage. Both numbers matter.
Step 4: Segment Your Portfolio
Not all borrowers are created equal. A young entrepreneur starting their first business carries different risk than an established corporation with 50 years of steady profits.
Break your portfolio into meaningful groups:
- Industry sectors
- Geographic locations
- Credit ratings
- Loan sizes
- Collateral types
Each segment will have its own PD characteristics.
Step 5: Adjust for Current Conditions
Here’s where it gets interesting. Historical data tells you what happened in the past, but IFRS 9 requires forward-looking estimates.
Ask yourself: Is the economy stronger or weaker now? Are interest rates rising? Is your specific industry facing headwinds?
If you’re a bank lending to restaurants and a pandemic hits (we all remember 2020), your historical PD from 2019 won’t cut it. You need to adjust upward.
Practical Methods for Calculating PD
Now let’s talk about actual calculation approaches.
The Simple Frequency Method
This works well if you have lots of data and stable conditions.
Formula: PD = Number of Defaults / Total Number of Borrowers
If 20 out of 400 borrowers in a particular segment defaulted over three years, your annual PD would be roughly 1.67%.
Pros: Easy to understand and calculate
Cons: Doesn’t account for changing conditions or individual borrower characteristics
Credit Rating Migration Method
If you assign credit ratings to borrowers (like A, B, C ratings), track how they move between ratings over time.
Watch how many “B-rated” borrowers eventually default. This gives you a PD for each rating category.
Pros: More sophisticated and accounts for changing creditworthiness
Cons: Requires consistent rating systems and good record-keeping
Statistical Models (Logistic Regression)
For the statistically inclined, logistic regression models can predict PD based on multiple factors simultaneously.
You might look at: debt-to-income ratio, payment history, industry, economic indicators, and more.
Pros: Captures complex relationships between factors
Cons: Requires expertise and solid data infrastructure
Adding the Forward-Looking Element
IFRS 9 specifically requires you to incorporate reasonable forecasts. Here’s how:
Use Economic Scenarios
Develop 2-3 economic scenarios:
- Base case: What you think will most likely happen
- Optimistic: Economy grows faster than expected
- Pessimistic: Economy slows or enters recession
Calculate PD under each scenario, then weight them based on their likelihood.
For example:
- Base case PD: 3% (probability 60%)
- Optimistic PD: 1.5% (probability 20%)
- Pessimistic PD: 7% (probability 20%)
Weighted PD = (3% × 0.6) + (1.5% × 0.2) + (7% × 0.2) = 3.5%
Monitor Leading Indicators
Keep an eye on metrics that predict trouble before it arrives:
- Rising unemployment in your borrowers’ industries
- Declining property values (if you hold mortgages)
- Increasing interest rates
- GDP growth or contraction
- Commodity price changes
When these indicators deteriorate, adjust your PD estimates accordingly.
Common Mistakes to Avoid
After years in this field, I’ve seen people trip over the same issues repeatedly.
Mistake 1: Using Stale Data
Your five-year-old default data might not reflect today’s reality. Refresh your data regularly and weight recent information more heavily.
Mistake 2: Ignoring Small Portfolios
“We only have 50 loans in this segment” isn’t an excuse. You might need to use external benchmarks or pool similar segments together, but you still need a reasonable PD.
Mistake 3: Set-It-and-Forget-It Mentality
PD isn’t a one-time calculation. Review and update your estimates at least quarterly, or more often when conditions change rapidly.
Mistake 4: Overly Complex Models
Yes, sophisticated models can be more accurate, but if your team doesn’t understand them, you’ll make mistakes. Sometimes a simpler approach that everyone comprehends beats a complex model nobody trusts.
Mistake 5: Forgetting About Stage Transfers
A borrower’s PD will change as their circumstances change. Someone who loses their job should move to a higher PD category (and possibly Stage 2). Update these assessments regularly.
Practical Tips for Implementation
Let me share some real-world wisdom:
Start Simple, Then Improve
Don’t try to build the perfect model on day one. Start with basic historical frequencies, get that working, then add sophistication over time.
Document Everything
Write down your methodology, assumptions, and data sources. Your auditors will ask, and future you will thank present you.
Validate Your Results
Does your PD make intuitive sense? If your model says blue-chip corporate borrowers have a higher PD than risky startups, something’s wrong. Sanity-check your outputs.
Use Technology Wisely
Excel works for small portfolios, but if you’re managing thousands of loans, invest in proper IFRS 9 ECL software. The time saved and error reduction pays for itself quickly.
Talk to Your Risk Team
The people managing credit risk in your organization probably already calculate something similar to PD. Don’t reinvent the wheel leverage their expertise and data.
Bringing It All Together
Calculating PD for IFRS 9 compliance doesn’t have to be overwhelming. Yes, it requires thought and effort, but break it down into steps:
- Collect your historical default data
- Segment your portfolio meaningfully
- Calculate basic historical default rates
- Adjust for current and forecasted conditions
- Document your approach
- Review and update regularly
Remember, perfection isn’t the goal, reasonable accuracy is. IFRS 9 recognizes that you’re making estimates about uncertain future events. What matters is that your approach is logical, consistent, and based on the best information available.
The more you work with PD calculations, the more intuitive they’ll become. You’ll start to develop a feel for when numbers look right and when something needs a second look.
Final Thoughts
IFRS 9 fundamentally changed how we think about credit losses, shifting from “incurred loss” to “expected loss.” PD calculations sit at the heart of this change.
Is it more work than the old way? Absolutely. But it also gives a more realistic picture of your credit risk and helps you make better lending decisions.
Start where you are, use what you have, and improve continuously. That’s the practical approach to PD calculation, and to most things in IFRS 9 finance, really.
If you’re just starting your IFRS 9 journey, don’t panic. Take it one step at a time. And if you’re already calculating PD but struggling with certain aspects, revisit your methodology with fresh eyes. Sometimes a small tweak makes a big difference.
Got questions about your specific situation? The beauty of the finance community is that someone somewhere has probably faced a similar challenge. Don’t hesitate to reach out to peers, attend industry conferences, or consult with specialists when needed.
After all, we’re all navigating these regulatory waters together.
About the Author: With over a decade of experience in financial reporting and regulatory compliance, I’ve helped numerous organizations navigate the complexities of IFRS 9 implementation. This guide reflects real-world lessons learned from working with banks, finance companies, and corporate treasury teams across different industries.
 for IFRS 9 compliance with our practical guide. Includes step-by-step methods, formulas, and real-world tips for accurate credit risk assessment.){kind=link}