CAIIB Paper 2 Module B Quiz 2

Market risk arises from adverse movements in market variables, including rates, foreign exchange, equity and commodity prices. It can affect trading and banking-book positions depending on the exposure.

For a fixed-rate bond, higher market yields generally reduce the present value of its fixed cash flows, so its market price falls. The magnitude depends in part on duration and convexity.

An open foreign-currency position gains or loses value as exchange rates move. The direction depends on whether the bank is long or short and on the currency pair’s quotation convention.

VaR estimates a loss quantile over a specified holding period and confidence level, subject to model assumptions and data. It does not describe the size of losses beyond the threshold and is not a worst-case guarantee.

At the stated confidence and horizon, the model estimates a 1% tail probability of losses exceeding the VaR threshold. It is not a certainty, exact forecast or maximum possible loss.

Historical simulation revalues current positions using observed historical changes in risk factors. It avoids assuming a specific parametric distribution but depends heavily on the chosen historical window and may miss unprecedented events.

The variance-covariance approach estimates portfolio volatility using risk-factor volatilities and correlations, often with a distributional assumption such as normality. Those assumptions can understate tail risk in stressed markets.

Monte Carlo simulation generates scenarios from specified models for risk factors and revalues exposures. It is flexible but computationally demanding and sensitive to model design and calibration.

VaR identifies a quantile, not the average or severity of losses in the tail beyond that quantile. Expected Shortfall and stress testing can provide complementary information about tail losses.

Expected Shortfall, also called Conditional VaR in common usage, measures the average loss in the tail beyond a selected quantile, subject to the methodology used. It better reflects tail severity than VaR alone.

Backtesting compares VaR forecasts with realised or appropriately defined hypothetical trading outcomes and counts exceptions. Too many exceptions may indicate model weakness, poor calibration or changed market conditions.

An exception occurs when the relevant realised loss exceeds the VaR estimate for that horizon and confidence level. Exception patterns should be analysed rather than treated as a mere administrative count.

Stress tests examine the effect of extreme or adverse but plausible scenarios, including events outside the normal model distribution. They complement rather than replace VaR, limits and scenario analysis.

Sensitivity analysis estimates how a portfolio value or earnings measure changes when a risk factor moves, such as a one-basis-point rate change or a 1% FX move. It is often local and may not capture nonlinear effects fully.

BPV, often called PV01 in relevant contexts, measures the monetary value change for a one-basis-point rate movement. The sign and size depend on the position and rate-shock convention.

Modified duration estimates the percentage price sensitivity to a small yield change: approximately ΔP/P ≈ −Dmod × Δy. It is a local linear approximation and becomes less accurate for large changes or material convexity.

Duration approximates interest-rate sensitivity. A longer-duration bond generally experiences a larger percentage price change for the same small yield movement, assuming comparable conventions and other factors.

Basis risk arises when hedging instruments and the exposure being hedged do not move perfectly together. The hedge can therefore be imperfect even when positions are economically related.

A long equity position generally gains when the equity price rises, while a short position generally loses, ignoring transaction costs and other factors.

Limits constrain risk-taking within approved appetite, for example through position, sensitivity, VaR or stop-loss limits. Breaches require defined escalation and remediation, not concealment.