Market Risk

Introduction

Market risk is the possibility that changes in market prices, rates, or volatility will reduce the value of a financial position or weaken an institution’s earnings and capital. It is central to banking, finance, and insurance because institutions continuously transform, hold, and trade exposures to interest rates, equity prices, foreign-exchange rates, commodities, and derivatives.

Market risk differs from credit risk. Credit risk concerns failure of a counterparty or borrower; market risk can produce losses even when every counterparty performs exactly as promised. It also differs from liquidity risk, although the two interact: a position may have a modest price risk under normal trading conditions but a large loss when it must be liquidated in a stressed market.

A sound market-risk framework combines economic understanding, quantitative measurement, limits, independent control, stress testing, and governance. No single measure is sufficient. Sensitivity measures explain the source of risk, while Value at Risk (VaR), Expected Shortfall (ES), backtesting, and stress testing assess the scale and reliability of potential losses.

Learning Objectives

By the end of this chapter, a student should be able to:

1. Explain the main types and drivers of market risk.
2. Calculate and interpret duration, convexity, beta, and other sensitivity measures.
3. Describe foreign-exchange, commodity, volatility, and basis risks.
4. Calculate and critique VaR and Expected Shortfall.
5. Explain historical, variance–covariance, and Monte Carlo risk models.
6. Use backtesting and stress testing to evaluate market-risk models.
7. Recommend appropriate controls, limits, hedges, and governance practices.

1. The Architecture of Market Risk

A market-risk position can be represented as a portfolio whose value depends on risk factors. Examples include a bond’s dependence on interest rates, an equity portfolio’s dependence on share prices, and an option’s dependence on price, volatility, time, and interest rates.

A useful local approximation is:

Here, ∆V is the change in portfolio value, xᵢ and xⱼ are market risk factors, the first derivatives represent delta or linear sensitivity, and the second derivatives represent curvature. The approximation is most reliable for small changes and sufficiently smooth pricing functions. Large shocks, discontinuities, and poorly behaved instruments require full repricing.

Main categories

Risk factors may be systematic, affecting many assets, or specific, affecting an individual issuer or instrument. They may also be linear or nonlinear. A forward contract is often approximately linear; an option is nonlinear because its sensitivity changes as the underlying price moves.

2. Interest-Rate Risk

Interest-rate risk arises because the value and income of instruments change when market rates change. Banks face repricing risk, yield-curve risk, basis risk, and optionality risk.

Bond pricing and yield

For a fixed-rate bond with cash flows Cₜ at times t = 1, …, n and yield y per period:

P is the bond’s present value, F is face value, and y is the yield per compounding period. The formula assumes known cash flows and a single yield for discounting. In practice, a full yield curve may be used, with each cash flow discounted at its appropriate spot rate.

When yields rise, the present value of fixed future cash flows falls. The relationship is inverse and nonlinear.

Macaulay and modified duration

Macaulay duration is the weighted average time to receive a bond’s cash flows:

where PV(CFₜ) is the present value of the cash flow at time t and P is bond price. If y is an annual effective yield with annual compounding, modified duration is:

For a small parallel yield change ∆y, the percentage price change is approximately:

Duration is measured in years, while ∆y is expressed as a decimal. The negative sign reflects the inverse price–yield relationship. The approximation assumes a parallel shift, small yield change, unchanged cash flows, and a sufficiently smooth bond price function.

A related measure is dollar value of a basis point (DV01), the approximate price change for a one-basis-point increase in yield:

The sign is convention-dependent. Risk reports often show positive DV01 as the magnitude of loss for a rate increase.

Convexity

Duration gives a linear approximation, but bond prices are curved. Convexity measures the second-order effect:

C is convexity, normally expressed in years squared. For a standard option-free bond, positive convexity means that the price rises by more when yields fall than it falls when yields rise by the same amount. The approximation becomes more useful for larger yield changes, although callable and puttable bonds may exhibit negative or changing convexity.

Worked example: duration and convexity

Suppose a bond is worth 1,000 monetary units, has modified duration of 4.5 years, and convexity of 30 years². If the yield rises by 50 basis points, ∆y = 0.005:

The estimated price decline is 2.2125%, or approximately 22.125 monetary units. The estimated new price is 977.875. A duration-only estimate would have predicted a 2.25% decline, illustrating the correction supplied by positive convexity.

Yield-curve and basis risk

A parallel-shift assumption is often unrealistic. A portfolio may be exposed to changes in the level, slope, or curvature of the yield curve. Key-rate duration measures sensitivity to a particular maturity segment, such as two-year or ten-year rates. Basis risk occurs when instruments intended to hedge one another respond differently, for example when a loan reprices using a different benchmark from the interest-rate swap used as its hedge.

Banks also face repricing-gap risk. If assets reprice later than liabilities, a rise in rates can compress net interest income. If assets reprice sooner, falling rates may have the same effect. Duration of equity extends this analysis by comparing the interest-rate sensitivity of the market value of assets and liabilities.

3. Equity Risk

Equity risk is the possibility that stock prices, equity indices, or issuer-specific values will decline. It includes market-wide risk, sector risk, country risk, and idiosyncratic risk.

Beta and the market model

Beta measures an asset’s sensitivity to market returns:

Rᵢ is the asset return, Rₘ is the market return, Cov denotes covariance, and Var denotes variance. In the single-index model:

αᵢ is the intercept and εᵢ is the asset-specific residual. The model assumes a stable relationship over the estimation period and separates market-related from residual variation. Beta is not a complete risk measure: a low-beta security can still have substantial issuer-specific or liquidity risk.

For a portfolio, beta is approximately the value-weighted average of constituent betas when the portfolio is fully invested and measured consistently:

where wᵢ is the portfolio weight. Hedging index exposure with futures can reduce systematic beta but cannot eliminate basis, sector, or stock-specific risk.

Equity sensitivities

For an equity holding, a simple price shock estimate is:

where N is the number of shares and ∆S is the change in share price. For an equity option, delta approximates the change in option value for a one-unit change in the underlying price, while gamma measures the change in delta. A portfolio containing options therefore requires nonlinear analysis and revaluation under substantial shocks.

4. Foreign-Exchange Risk

Foreign-exchange (FX) risk arises when the value of assets, liabilities, revenues, or expenses changes because exchange rates move. It has three forms:

• Transaction exposure: a contracted foreign-currency cash flow changes in domestic-currency value.
• Translation exposure: foreign subsidiaries’ accounts change when consolidated.
• Economic exposure: exchange rates alter future competitiveness and cash flows.

If an institution holds Q units of foreign currency and the domestic-currency price is S, the domestic value is V = QS. A small exchange-rate change gives:

For a portfolio with multiple currencies, exposures should be measured by currency, maturity, and instrument. A net long position in a currency gains when that currency appreciates against the reporting currency, subject to quotation convention.

FX forwards, futures, swaps, and options can hedge exposure. A forward hedge may remove much of the exchange-rate risk but introduce counterparty, settlement, rollover, and basis risks. Currency options provide protection with retained upside but require a premium and expose the institution to volatility and time-value effects.

5. Commodity Risk

Commodity risk concerns changes in the price of energy, metals, agricultural products, or other physical inputs. It affects producers, airlines, manufacturers, trading firms, banks, and insurers.

Commodity prices are influenced by inventories, weather, geopolitics, transportation, storage, seasonality, and convenience yield. Futures prices may differ from spot prices because of financing and storage effects. A hedge using a related but non-identical commodity creates basis risk. For example, an airline hedging jet fuel with crude-oil futures may reduce broad energy exposure without perfectly matching its actual fuel cost.

A futures position’s approximate profit or loss is:

N is the number of contracts, M is the contract’s quantity per unit price, and ∆F is the futures-price change. Units must be checked carefully; a contract quoted in dollars per barrel and covering 1,000 barrels has a different exposure from a contract covering 100 barrels.

6. Volatility Risk and Derivative Sensitivities

Volatility risk is the risk that realised or implied volatility changes. It is especially important for options, whose values depend not only on the underlying price but also on volatility, time to maturity, interest rates, and strike price.

Common option sensitivities, often called Greeks, include:

A local option approximation may be written as:

∆S is the underlying-price change, ∆σ is the volatility change, ∆r is the interest-rate change, and ∆t is the passage of time. The equation assumes small changes and locally stable sensitivities. It is inadequate for large shocks or instruments with discontinuous payoffs.

Volatility surfaces contain information across strikes and maturities. Risk managers must consider not only a parallel volatility shift but also skew and term-structure changes. A portfolio can be delta-neutral yet lose money because of gamma, vega, or correlation exposure.

7. Value at Risk

Value at Risk estimates a loss threshold over a specified horizon and confidence level. A one-day 99% VaR of 10 million means that, under the model and assumptions, the loss should exceed 10 million on approximately 1% of comparable days. VaR does not state how large losses may be beyond that threshold.

VaR depends on:

1. Horizon, such as one trading day or ten trading days.
2. Confidence level, such as 95% or 99%.
3. Data and model, including historical window, distributions, correlations, and valuation method.
4. Portfolio scope, including trading positions and treatment of hedges.

Parametric VaR

For a single normally distributed return with portfolio value V, standard deviation σ, and confidence quantile zα:

For a portfolio of linear positions with exposure vector w and covariance matrix Σ:

and VaR is approximately zα σₚ when w is expressed in monetary exposure units. For a one-tailed 99% normal model, zα is approximately 2.326. The approach is fast but relies on linearity and distributional assumptions.

Historical simulation

Historical simulation applies actual historical changes in risk factors to today’s portfolio and revalues the resulting scenarios. It makes fewer distributional assumptions but depends strongly on the chosen historical window. It may omit unprecedented events and may be slow for complex portfolios.

Monte Carlo simulation

Monte Carlo generates many simulated risk-factor changes from specified statistical or economic models and revalues the portfolio. It can represent nonlinear instruments and dependencies, but results depend on model specification, calibration, computational quality, and the number of simulations.

8. Expected Shortfall

Expected Shortfall measures the average loss in the tail beyond the VaR threshold. For a continuous loss variable L and confidence level α:

This definition assumes losses are expressed as positive numbers and that the conditional expectation exists. For discrete empirical distributions, a tail average or a mathematically coherent quantile-based definition should be used, especially where observations have equal probabilities.

ES answers a question VaR cannot: once the threshold is breached, how severe is the average loss? It is more informative for tail risk and, under suitable conditions, has the desirable property of subadditivity. Nevertheless, ES is sensitive to tail data, model assumptions, liquidity effects, and the quality of extreme scenarios.

Illustrative risk calculation

Assume a portfolio’s simulated one-day losses, in millions, are 1, 2, 3, 4, 5, 6, 7, 8, 12, and 20. At a simplified 80% confidence level, the worst 20% consists of the two largest observations. The empirical VaR is 12 million under the convention that the 80th percentile is the ninth ordered observation, while:

The result is illustrative; professional systems must specify interpolation, tail inclusion, horizon, confidence level, and scenario weighting.

9. Backtesting and Stress Testing

Backtesting

Backtesting compares model forecasts with realised outcomes. If a model produces a 99% one-day VaR, exceptions—days on which the actual loss exceeds VaR—should occur approximately 1% of the time under stable conditions.

The simplest diagnostic is the exception rate:

Too many exceptions suggest underestimation, model deterioration, data problems, or changing portfolio composition. Clustering of exceptions suggests that losses are not independent or that volatility changes are not captured. A model may have the correct average exception rate but still fail because exceptions cluster during stress. Backtesting should therefore examine both frequency and independence, not merely count breaches.

Backtesting is limited by finite samples. A 99% daily model requires many observations to distinguish random variation from systematic failure. It also tests the joint forecasting and valuation process, not only the statistical distribution.

Stress testing

Stress testing evaluates portfolio outcomes under severe but plausible scenarios. Scenarios may be historical, hypothetical, reverse stress tests, or sensitivity-based. Examples include a sharp yield-curve steepening, a currency devaluation, an equity-market crash, a commodity supply shock, or a simultaneous volatility and correlation increase.

Stress testing should include nonlinear repricing, market illiquidity, basis movements, collateral calls, margin requirements, and management actions. A reverse stress test begins with an unacceptable outcome and asks which combinations of events could produce it. Stress tests complement, rather than replace, VaR and ES because they can address events absent from the estimation sample.

10. Integrated Case: Meridian Commercial Bank

The following is an illustrative case, not a report of an actual institution.

Background

Meridian Commercial Bank holds a portfolio containing fixed-rate government bonds, an equity-index position, a US-dollar loan funded partly in domestic currency, and energy-sector loans. It also sells options to corporate customers. The treasury department reports a low daily VaR, but the chief risk officer is concerned that the result masks concentration and nonlinear risks.

Problem

The portfolio has duration exposure to a steepening yield curve, a net long dollar position, equity beta of 0.85, and short volatility exposure from sold options. Its VaR model uses a short historical window that excludes a previous energy shock. The bank’s limit framework focuses on aggregate VaR and does not separately limit vega, key-rate duration, or commodity basis risk.

Analysis

Risk management decomposes the portfolio into risk factors. Key-rate duration shows that the ten-year bond exposure dominates the interest-rate risk. DV01 indicates that a 25-basis-point increase in ten-year yields would produce a material loss, even though a parallel-shift duration report appears moderate. The FX report shows that a domestic-currency depreciation would generate a gain on the dollar loan, but the gain is partly offset by dollar funding and expected credit losses among unhedged borrowers.

The equity position’s beta implies substantial market exposure, while the option book is approximately delta-neutral but short gamma and vega. Consequently, a large equity decline accompanied by a volatility spike can generate losses much greater than a linear beta report suggests. Commodity stress testing identifies a simultaneous energy-price fall and deterioration in energy borrowers’ credit quality.

Approach and solution

Meridian introduces:

• key-rate duration and DV01 limits;
• separate equity beta and concentration limits;
• gross and net FX exposure limits;
• commodity basis and stress limits;
• delta, gamma, vega, and volatility-surface reporting;
• a longer and more representative historical window;
• daily VaR backtesting and independent price verification;
• ES and scenario-based stress testing;
• escalation procedures for limit breaches.

The bank hedges part of the ten-year exposure with interest-rate futures, reduces the short-vega option inventory, and uses FX forwards while preserving a documented residual-risk budget. It also evaluates hedge effectiveness under non-parallel curve and basis movements.

Outcome

Aggregate VaR increases initially because the improved model captures previously hidden risks. Management nevertheless regards the result as better because it is more transparent and supported by stress scenarios. The bank reduces tail losses under a combined rate, equity, FX, and volatility shock, although it accepts some residual basis and liquidity risk.

Learning points

The case demonstrates that aggregate VaR can conceal concentrations, that delta neutrality does not eliminate option risk, and that hedging must be assessed under realistic joint movements. Effective market-risk management is not the pursuit of the smallest model number; it is the disciplined identification, measurement, challenge, and control of economically important exposures.

11. Governance, Regulation, and Professional Practice

Market-risk governance requires clear ownership. The front office manages positions, risk management independently measures and challenges risk, finance validates valuation and reserves, and internal audit evaluates controls. Senior management and the board should approve risk appetite, limits, model policies, and escalation arrangements.

Important controls include independent market-data sourcing, price verification, valuation adjustments, model validation, limit monitoring, documentation, and breach escalation. Valuation uncertainty is material when instruments are illiquid or models depend on unobservable inputs. Collateral and margin calls should be included in liquidity and stress analysis.

Regulatory frameworks generally require capital for market risk and expect robust risk measurement, stress testing, governance, and disclosure. Exact regulatory requirements vary by jurisdiction and change over time; institutions must therefore consult applicable supervisory standards rather than treating a textbook formula as a regulatory rule.

Ethically, risk professionals must avoid manipulating assumptions to achieve desired VaR, suppressing exceptions, or presenting hedges as riskless. Model uncertainty should be communicated clearly. AI and machine learning may assist with anomaly detection, volatility forecasting, and scenario generation, but they introduce risks of opacity, unstable relationships, data leakage, and procyclicality. Human oversight, validation, explainability, and controlled use remain essential.

12. Limitations and Best Practices

Market-risk models are conditional estimates, not forecasts with certainty. Common weaknesses include non-stationary markets, inadequate tail data, unstable correlations, omitted liquidity effects, model risk, stale prices, incorrect position data, and failure to capture behavioural responses.

Best practice is to:

1. combine sensitivities, VaR, ES, and stress tests;
2. use risk-factor decomposition rather than relying only on aggregate figures;
3. revalue nonlinear instruments under material scenarios;
4. validate models independently and backtest regularly;
5. include liquidity, basis, concentration, and wrong-way effects;
6. maintain conservative limits and escalation triggers;
7. document assumptions, data sources, and known weaknesses;
8. review scenarios after structural market changes;
9. connect risk measures to capital, pricing, hedging, and compensation;
10. ensure that management understands uncertainty and residual risk.

Chapter Summary

Market risk is the potential loss arising from adverse movements in rates, equity prices, exchange rates, commodity prices, volatility, and related risk factors. Interest-rate risk can be assessed using duration, DV01, convexity, key-rate duration, and repricing analysis. Equity risk includes systematic exposure measured by beta and specific risk that beta does not capture. FX and commodity positions require careful attention to currency, contract size, maturity, and basis risk. Volatility risk is particularly important for options and is assessed through Greeks such as delta, gamma, vega, theta, and rho.

VaR estimates a loss quantile but does not describe losses beyond that point. Expected Shortfall measures the average tail loss and provides additional information about extreme outcomes. Historical, parametric, and Monte Carlo methods involve different assumptions and weaknesses. Backtesting evaluates forecast performance using realised exceptions, while stress testing examines severe plausible scenarios, including nonlinear, liquidity, and joint-factor effects. Effective governance combines quantitative models with independent validation, limits, stress analysis, ethical conduct, and informed management judgment.

Key Terms

Market risk: Risk of loss from adverse market movements.

Risk factor: A market variable that influences portfolio value.

Duration: Approximate sensitivity of a bond price to yield changes.

Macaulay duration: Present-value-weighted average timing of bond cash flows.

Modified duration: Duration measure used to estimate percentage price change for a yield change.

DV01: Approximate price change for a one-basis-point yield movement.

Convexity: Measure of the curvature of the bond price–yield relationship.

Basis risk: Risk that related instruments or prices do not move together.

Beta: Sensitivity of an asset’s return to market return.

Foreign-exchange exposure: Sensitivity of value or cash flow to exchange-rate movements.

Commodity risk: Risk from changes in commodity prices and related markets.

Volatility risk: Risk from changes in realised or implied volatility.

Delta: First-order sensitivity of a derivative to its underlying price.

Gamma: Sensitivity of delta to the underlying price.

Vega: Sensitivity of an option to volatility.

Value at Risk: A specified loss quantile over a defined horizon and confidence level.

Expected Shortfall: Average loss beyond a specified VaR threshold.

Backtesting: Comparison of risk-model forecasts with subsequent outcomes.

Stress testing: Evaluation of losses under severe, specified scenarios.

Model risk: Risk that a model is unsuitable, incorrectly implemented, or misused.

Review Questions

5 Short Answer Questions

1. Why does a rise in yield generally reduce the price of a fixed-rate bond?
2. Define DV01 and state its usual unit of measurement.
3. What does an equity beta of 1.2 indicate in the market model?
4. Distinguish transaction exposure from translation exposure in foreign exchange.
5. Why can a delta-neutral option portfolio still experience substantial losses?

5 Long Answer Questions

1. Explain Macaulay duration, modified duration, and convexity, including their assumptions and limitations.
2. Compare parametric VaR, historical simulation, and Monte Carlo simulation.
3. Discuss the principal forms of interest-rate risk faced by a commercial bank.
4. Explain how Expected Shortfall complements VaR and identify practical difficulties in estimating it.
5. Describe an effective governance framework for market-risk measurement and control.

3 Analytical / Application Questions

1. A bond portfolio is worth 20 million monetary units, has modified duration of 5.2 years, and convexity of 42 years². Estimate the percentage and monetary price change for a 40-basis-point yield increase using the duration–convexity approximation.
2. A bank is long US dollars through a foreign-currency loan but has partially matching dollar funding. Analyse the residual FX risks and recommend suitable measurements and hedges.
3. A portfolio records the correct number of VaR exceptions overall, but the exceptions cluster during periods of market stress. What does this indicate, and what additional tests or risk controls should management introduce?

Multiple Choice Questions with Answers

Question 1

A bond has modified duration of 4 years. For a small yield increase of 0.25%, the approximate percentage price change is:

A. −1.00%

B. −0.25%

C. +1.00%

D. +4.25%

Correct Answer: A. −1.00%

Question 2

Which measure captures the second-order curvature of a bond’s price–yield relationship?

A. Beta

B. Convexity

C. Vega

D. DV01

Correct Answer: B. Convexity

Question 3

A beta of 1.3 most directly indicates that an asset:

A. Has no issuer-specific risk

B. Moves exactly opposite to the market

C. Has estimated market sensitivity of 1.3

D. Has 1.3% guaranteed return

Correct Answer: C. Has estimated market sensitivity of 1.3

Question 4

Which risk measure estimates the average loss beyond a specified loss quantile?

A. Duration

B. Expected Shortfall

C. Beta

D. DV01

Correct Answer: B. Expected Shortfall

Question 5

Under a one-tailed normal model, a 99% VaR primarily represents:

A. The average loss on all trading days

B. The maximum possible loss

C. A loss threshold exceeded approximately 1% of the time under the model

D. The expected profit at the 99th percentile

Correct Answer: C. A loss threshold exceeded approximately 1% of the time under the model

Question 6

Which Greek measures an option’s sensitivity to implied volatility?

A. Gamma

B. Vega

C. Theta

D. Rho

Correct Answer: B. Vega

Question 7

A futures contract covers 1,000 barrels and its price falls by 2 monetary units per barrel. For one long contract, the approximate loss is:

A. 2 monetary units

B. 20 monetary units

C. 1,000 monetary units

D. 2,000 monetary units

Correct Answer: D. 2,000 monetary units

Question 8

The main purpose of VaR backtesting is to:

A. Guarantee that no future loss will exceed VaR

B. Compare predicted risk thresholds with realised outcomes

C. Calculate a bond’s coupon

D. Remove all model assumptions

Correct Answer: B. Compare predicted risk thresholds with realised outcomes

Question 9

Which statement about stress testing is most accurate?

A. It only uses normally distributed daily returns

B. It replaces all sensitivity measures

C. It evaluates portfolio outcomes under severe specified scenarios

D. It proves that a portfolio is risk-free

Correct Answer: C. It evaluates portfolio outcomes under severe specified scenarios

Question 10

A portfolio can have low aggregate VaR while remaining vulnerable because:

A. VaR always includes every nonlinear risk perfectly

B. Risk concentrations, basis risk, and tail events may be hidden by aggregation

C. Low VaR automatically implies high liquidity

D. VaR measures only accounting profit

Correct Answer: B. Risk concentrations, basis risk, and tail events may be hidden by aggregation