IFRS 9: PD Model Development for Low Default Portfolios Incorporating External Data and Expert Judgment

IFRS 9 Expected Credit Loss (ECL) demands forward-looking assessments of credit risk, with Probability of Default (PD) as a critical component. While PD modeling is straightforward for high-default portfolios, a significant challenge emerges with Low Default Portfolios (LDPs).

These portfolios, often comprising high-quality corporate loans or sovereign debt, inherently experience very few, if any, historical defaults internally. This scarcity of data creates a hurdle for traditional statistical modeling. So, how do actuaries and risk professionals effectively estimate PDs for LDPs under IFRS 9?

The LDP Conundrum: Why Internal Data Falls Short

Statistical models thrive on observed events. For LDPs with few or no defaults, building a statistically significant model based solely on internal experience is difficult. This can lead to volatile, unreliable, or even zero PD estimates, which misrepresent true credit risk.

IFRS 9, however, insists on robust, forward-looking, and unbiased PDs. This necessitates a blend of external insights and considered qualitative overlays.

Harnessing External Data for a Strong Foundation

When internal data is sparse, external sources become essential for LDP PD modeling. These include:

Industry Default Rates:

Data from credit rating agencies (e.g., S&P, Moody's) on rated entities across sectors and rating grades. These offer broad benchmarks of default probabilities for different risk profiles.

Publicly Available Market Data:

Yields on corporate bonds, Credit Default Swap (CDS) spreads, or equity prices can provide market-implied PDs, reflecting market sentiment and forward-looking views.

Peer Group Analysis:

Data from similar institutions or industry bodies can offer proxies for default behaviour.

Using external data requires careful selection and robust calibration: mapping external characteristics to your obligors and adjusting for differences in definition, economic conditions, and specific portfolio traits.

The Indispensable Role of Expert Judgment

While external data provides a strong quantitative anchor, it rarely offers a perfect fit. Here, rigorously applied expert judgment bridges the gap. This isn't arbitrary guessing; it's a structured application of informed insights from experienced credit professionals, actuaries, and economists.

Expert judgment is crucial for:

Adjusting for Unique Circumstances:

Reflecting specific borrower characteristics, industry nuances, or collateral structures not captured by generic external data.

Incorporating Forward-Looking Information:

Accounting for anticipated changes in economic conditions, regulatory environments, or geopolitical factors.

Dealing with Data Gaps:

Where no suitable external proxy exists, well-reasoned expert judgment, supported by qualitative assessments, becomes the primary input.

Crucially, all applications of expert judgment must be thoroughly documented, transparent, and subject to robust governance and periodic review to ensure consistency, auditability, and prevent arbitrary adjustments.

Blending Data and Judgment for Robust ECLs

Effective PD modeling for LDPs under IFRS 9 is an iterative process, combining quantitative analysis with qualitative insights. It typically involves:

1. Starting with a base PD derived from the most relevant external data.

2. Calibrating this base PD using any available internal loss experience, even if scarce, and adjusting for differences in risk drivers.

3. Applying structured expert judgment to overlay specific adjustments for forward-looking information, idiosyncratic risks, or portfolio specifics.

This blended approach ensures resulting PDs are statistically sound where possible, and reflect the LDP's unique risk profile and IFRS 9's forward-looking requirements.

Conclusion

At Lux Actuaries, we understand PD modeling for Low Default Portfolios is one of IFRS 9's trickiest challenges. By strategically integrating relevant external data with well-governed expert judgment, financial institutions can develop robust, compliant, and defensible PD estimates. This hybrid approach is not just a regulatory necessity; it’s a smart way to gain deeper insights into credit risk and build resilience in financial reporting.