Risk Management
Risk Management tests whether a candidate can name the risk, measure it in the right units, and govern it: the risk taxonomy, appetite and limits, the three lines of defence, market and liquidity risk metrics, operational risk under the revised Basel framework, stress testing and ICAAP.
Taxonomy and governance
- Credit, market (traded), interest-rate risk in the banking book, liquidity and funding, operational, country and transfer, reputational and strategic risk.
- First line owns risk, second line sets framework and challenges, third line (internal audit) provides independent assurance.
- Risk appetite is expressed in limits: exposure caps, VaR limits, stop-loss, liquidity ratios, concentration limits, and key risk indicators with escalation triggers.
Measurement
- Market risk: modified duration gives ΔP ≈ −D x P x Δy; PV01 gives the value of one basis point; VaR is a loss threshold at a confidence level and horizon, not a worst case, and is not sub-additive; expected shortfall averages losses beyond the VaR point.
- Operational risk under the Basel III final reforms uses the standardised approach: a Business Indicator Component multiplied by an Internal Loss Multiplier.
- Liquidity: LCR and NSFR, gap analysis, survival horizon and stress outflows; ICAAP aggregates all risks against available capital.
Worked example
A bond portfolio of 200m has a modified duration of 4.5. Yields rise 60 basis points. ΔP ≈ −4.5 x 200m x 0.006 = −5.4m, a 2.7% mark-to-market loss. Reasoning: duration is linear and local, so it overstates losses for small moves in the other direction and understates them for large moves because convexity offsets; a 300bp shock needs a full revaluation or stress scenario rather than the duration formula.
Traps that cost marks
- Describing VaR as the maximum possible loss.
- Mixing the VaR confidence level or horizon (99%/one-day versus 99%/ten-day).
- Using historical correlations in a stress test, when correlations converge in a crisis.
- Treating banking-book interest-rate risk as market risk and applying trading limits.
- Presenting backtesting exceptions without analysing the model or the data window.