Explanation

When dealing with limited historical data for operational risk loss severity estimation, the most appropriate approach is to supplement the internal data with external data from other banks, after making appropriate scale adjustments.

Why Option D is correct:

• External data supplementation: Operational risk events are relatively rare, so internal data alone is often insufficient for robust statistical modeling
• Scale adjustments: External data needs to be scaled to account for differences in bank size, business mix, and risk profile
• Industry practice: This approach is widely used in operational risk management and is recommended by regulatory frameworks like Basel II/III

Why other options are incorrect:

• Option A: Monte Carlo simulation generates synthetic data but doesn't address the fundamental lack of real loss observations
• Option B: Poisson distribution is typically used for loss frequency modeling, not loss severity modeling
• Option C: Credit rating agencies primarily focus on credit risk information, not operational risk loss data

This approach helps create a more comprehensive dataset for estimating the loss severity distribution while maintaining relevance to the specific bank's characteristics.