Citibank·Data Scientist·Technical Phone Screen·Senior

SeniorPrefer not to say

Apr 2026

Summary

Citibank data scientist interview that went deep into credit risk math pretty fast. One big technical case covering the full EL and RWA pipeline, which sounds manageable until you're actually sitting there trying to remember what to do with off-balance-sheet items under pressure.

Questions Asked (1)

Given anonymized loan data with probability of default, loss given default, and exposure at default figures, how would you compute portfolio expected loss and risk-weighted assets? Walk through your assumptions, formulas, aggregation method, how you'd handle off-balance-sheet exposures, and any sensitivity analysis you'd run.

Data ModelingTechnical Trade-offsProduct Analytics & Metrics

Author's notes

This is a meaty one.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by defining the key terms and formulas for expected loss (EL = PD × LGD × EAD) and risk-weighted assets (RWA = 12.5 × capital requirement, where capital requirement = EL + UL, with UL based on Basel IRB formula). Then walk through assumptions, aggregation, off-balance-sheet treatment, and sensitivity analysis in a structured manner, emphasizing data quality and regulatory context.

Pro tip: Demonstrate awareness of regulatory nuances: mention that RWA depends on the Basel framework (e.g., IRB vs. standardized) and that off-balance-sheet exposures require credit conversion factors (CCFs). Also, highlight that sensitivity analysis should include stress testing PD, LGD, and correlations.

1. Define Assumptions and Data Preparation

State assumptions about the data (e.g., PD, LGD, EAD are point-in-time or through-the-cycle, granularity, and whether they are regulatory or economic). Discuss data cleaning, validation, and handling missing values.

2. Compute Expected Loss (EL)

For each exposure, calculate EL = PD × LGD × EAD. Aggregate EL at portfolio level by summing individual ELs, ensuring consistent units and time horizon (typically 1 year).

3. Compute Risk-Weighted Assets (RWA)

Under Basel IRB, compute capital requirement (K) using the formula: K = LGD × N[(1-R)^-0.5 × G(PD) + (R/(1-R))^0.5 × G(0.999)] - PD × LGD, where R is asset correlation. Then RWA = 12.5 × K × EAD. For standardized approach, use regulatory risk weights. Aggregate RWA across exposures.

4. Handle Off-Balance-Sheet Exposures

Convert off-balance-sheet items to on-balance-sheet equivalents using credit conversion factors (CCFs) based on exposure type (e.g., 100% for direct credit substitutes, 50% for commitments). Then apply PD, LGD, and EAD as above.

5. Perform Sensitivity Analysis

Run scenario analyses by varying PD, LGD, EAD, and correlations (e.g., ±10% or stress scenarios). Assess impact on EL and RWA. Consider macroeconomic scenarios and concentration risks.

Key Points to Mention

Formula for expected loss: EL = PD × LGD × EAD, and aggregation by summing across exposures.
Risk-weighted assets calculation under Basel IRB: use of asset correlation, maturity adjustment, and the Vasicek model.
Off-balance-sheet exposure treatment: application of credit conversion factors (CCFs) to derive EAD.
Sensitivity analysis: stress testing PD, LGD, and correlations; scenario analysis for economic downturns.
Data quality and assumptions: importance of validating PD, LGD, EAD inputs; distinction between regulatory and economic capital.
Portfolio aggregation: handling diversification effects and concentration risk; use of segmentation by asset class.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.