1. Move from a position to its risk drivers
Treasury risk management connects market positions, counterparty obligations, funding and operational controls. A bond purchase can create interest-rate sensitivity, issuer exposure, settlement obligations and a refinancing need. A derivative can reduce one market exposure while creating collateral demands or counterparty risk. The first task is therefore to identify what the bank has promised and what can change its value or cash flows. A single product label or an aggregate profit number cannot explain the full risk of the activity.
Fictional Meridian Bank holds fixed-rate bonds, quotes foreign exchange to customers and uses interest-rate swaps. Its report shows a favourable daily profit, but independent risk notices a growing bond sensitivity and an unresolved trade confirmation. These facts need different responses. The sensitivity may require a position or hedge decision, while the confirmation issue requires operational investigation. Treating both as harmless because the day was profitable would confuse an outcome with the quality of the underlying risk process.
This chapter applies the risk concepts introduced in Module B to treasury decisions. It covers supervision, control, market and credit exposure, duration, value at risk and derivative use. Calculators use transparent educational models rather than claim to implement a complete regulatory capital method. The student should be able to explain the direction, units and limitations of a result. That explanation matters because a correct calculation can still support an unsuitable decision if the position, horizon or assumptions were misunderstood.
2. Establish an independent control structure
The board and senior management establish permitted activities, appetite and oversight. Treasury executes within delegated authority. Independent risk oversight challenges measures and monitors limits, while settlement and accounting functions confirm and reconcile transactions. Internal audit assesses the framework and its operation. Actual organisational names vary; the principle is a clear separation of conflicting duties with accountable decisions. An institution should not rely on one dealer to originate a position, value it, approve its limit and resolve every resulting reconciliation difference.
Independent valuation uses appropriate market evidence and controlled methods. A dealer's quote can be informative but may be influenced by incentives. Illiquid positions require particular attention because a small indicative quote may not represent an executable price for the full holding. Valuation adjustments and uncertainty should be visible to decision makers. A model can generate a precise number without reliable inputs, so the review should challenge data sources, assumptions and changes as well as arithmetic implementation.
Trade completeness is equally important. Late capture, duplicate records or unauthorised amendments can invalidate every subsequent risk calculation. Reconcile dealing systems, confirmations, accounting and settlement records, investigate material breaks and preserve an audit trail. Emergency access needs boundaries and review. A sophisticated risk report based on an incomplete position file can be less useful than a simple accurate report, because its apparent precision may discourage the questions that would reveal the missing exposure.
3. Design limits that match the exposure
Limits translate appetite into usable boundaries. Notional limits constrain size, position limits constrain net or gross exposure as defined, sensitivity limits constrain response to specified market movements, and counterparty limits constrain relevant credit exposure. VaR and stress limits add other views. A stop-loss threshold responds to a defined loss measure but does not guarantee execution at that amount. Each limit needs a definition, owner, measurement frequency and escalation process. A number without an action rule is weak operational guidance.
Suppose a desk has an assumed DV01 limit of INR5 lakh and current magnitude INR4 lakh. Utilisation is 80%, leaving INR1 lakh headroom under that measure. A proposed trade adding INR1.5 lakh sensitivity would raise the total to INR5.5 lakh if the sensitivities add in the stated scenario. The transaction should not proceed merely because notional headroom remains. Different limits address different dimensions, and passing one does not authorise a breach of another.
Warnings should permit action before a hard boundary is crossed. Breaches require timely explanation, containment and authorised resolution. Raising a limit retrospectively should not become routine loss avoidance or a means of hiding weak controls. Temporary exceptions need rationale, conditions and expiry. Management should also examine repeated near-breaches or rapid utilisation growth: a desk can become vulnerable before a formal breach occurs, particularly where liquidity deteriorates and the capacity to reduce exposure is shrinking.
4. Measure bond sensitivity with duration and DV01
Modified duration estimates the percentage price response to a small yield change. Approximate monetary change equals minus market value times modified duration times the yield change expressed as a decimal. For INR80 crore market value, duration 3.5 and a 20-basis-point yield rise, the change is minus 80 times 3.5 times 0.002, or minus INR0.56 crore. The negative sign describes the usual inverse price-yield relationship for the assumed fixed-rate position. It is not an arbitrary convention to discard after calculation.
DV01 is the approximate value change for one basis point, often reported as a positive magnitude. In the example, 80 times 3.5 times 0.0001 gives INR0.028 crore, or INR2.8 lakh per basis point. A twenty-basis-point rise therefore implies twenty times that magnitude of loss under the linear approximation. State both the monetary unit and reporting convention. Confusing one percentage point with one basis point creates a hundredfold error, while mixing lakh and crore can create another major scale error.
Duration is local and incomplete. A large shock, non-parallel curve movement, spread change or embedded option can cause actual revaluation to differ. A portfolio can have a small aggregate duration while retaining substantial exposure to the relative movement of different maturities. Use bucketed sensitivities or suitable full revaluation where needed. The simple tool teaches the first-order relationship; it should not be presented as a complete pricing engine for callable bonds, structured instruments or a stressed yield curve.
Follow the process
No autoplay. Read each step, or use Play, Pause and Next. Reduced-motion settings remove transitions.
Direction and units come first.
Duration is local; VaR is a quantile.
Monitor residual risk and cash demands.
All steps are available above.
5. Build a hedge using sensitivity rather than face value
A hedge should address the exposure's economic driver. Equal face values do not ensure equal and opposite sensitivity. Suppose an asset portfolio has DV01 magnitude INR2.8 lakh and one hedge contract provides INR0.07 lakh of opposite DV01. A simplified full sensitivity hedge requires forty contracts. The direction of the hedge must be established separately. For a long fixed-rate bond position harmed by rising yields, an appropriate short price exposure may offset part of that sensitivity under the assumed relationship.
Rounding can leave residual risk. If the ideal contract count is 40.4 but only whole contracts are available, compare forty and forty-one using the resulting net sensitivity and policy. The choice should not be made solely by rounding to the nearest integer without considering over-hedging constraints or other exposures. A hedge also needs liquidity, documentation, counterparty and margin assessment. A reduction in DV01 can coexist with greater cash demands if the derivative requires collateral before the asset's economic benefit is realised.
Basis risk remains when the hedge and asset do not move identically. Different reference rates, maturities, credit spreads or deliverable instruments can weaken the offset. Monitor the combined position and reassess when the underlying changes. If a bond is sold but its hedge remains, the derivative can become a new exposure. A hedge register should therefore connect the transaction to its objective, amount, horizon and review conditions rather than treat an initial approval as permanent evidence of effectiveness.
6. Interpret VaR as a conditional quantile
Value at risk describes a loss threshold for a specified horizon and confidence level under a model. A one-day 99% VaR of INR2 crore means the model assigns about 1% probability to exceeding that loss over a day. It is not a maximum loss, an average daily loss or a guarantee that a position can be closed at the threshold. Always name the portfolio, horizon, confidence and method. Without those details, comparisons between two VaR numbers can be meaningless.
For a zero-mean normal-return illustration, portfolio value INR100 crore and daily volatility 1% give a one-day 99% VaR of approximately INR2.32635 crore. The formula multiplies value, decimal volatility and the normal quantile 2.32635. Under independent identically distributed daily returns and constant volatility, ten-day scaling uses the square root of ten, producing approximately INR7.35656 crore. Multiplying by ten assumes a different relationship. The stated shortcut should not be extended automatically to serial dependence, changing volatility or nonlinear portfolios.
Parametric, historical and simulation methods depend on different modelling choices. Historical simulation still assumes that the selected past movements are informative. Monte Carlo requires a suitable model and reliable revaluation; more paths do not repair a bad specification. Model governance should cover data, conceptual validity, implementation and use. A simple transparent measure with understood limitations can support better decisions than a complex estimate whose users cannot explain what loss distribution or position set it represents.
Work the cases yourself
7. Challenge estimates with outcomes and stress
Backtesting compares forecasts with realised outcomes under a defined profit-and-loss measure. A 99% daily model does not require exactly one exception in every hundred observations. Counts vary, and patterns matter. Investigate whether exceptions arise from market behaviour, data errors, new trades, valuation differences or use outside the model's intended scope. An apparently good exception count does not prove that every assumption is sound, especially if the observation period contains little stress or the portfolio changes materially.
Stress testing examines specified adverse conditions beyond routine estimates. A coherent treasury scenario might combine a yield rise, currency movement, widening spreads, lower market depth and larger collateral calls. Explain the economic links before calculating. A random collection of unrelated worst observations can be difficult to interpret, while mild recent history can understate vulnerability. Show results before and after feasible management actions, and do not assume that a large position can be sold instantly at its previous mid-market valuation.
Reverse stress testing starts from a critical outcome such as exhausting a loss allowance or available liquidity and asks what conditions could cause it. This is a vulnerability analysis, not a prediction of certain failure. It can reveal dependence on one market relationship or funding source. Management should identify early warnings and actions with credible authority, capacity and timing. A scenario report that produces no changes or follow-up may add information without improving the institution's ability to respond.
8. Measure counterparty exposure separately
Treasury credit risk includes issuer risk on securities, interbank exposure and derivative counterparty risk. A positive derivative replacement value represents an amount potentially lost if the counterparty fails, subject to the applicable netting, collateral and exposure method. Future market changes can increase exposure. Notional principal is a reference amount and should not automatically be treated as current loss. Conversely, a zero-value contract at inception can still create substantial future exposure and cash obligations.
Suppose Meridian has a positive replacement value of INR3 crore and an explicitly assumed future-exposure add-on of INR2 crore in a simplified teaching case. Gross illustrative exposure is INR5 crore before any recognised offsets. If eligible collateral of INR1 crore is applied under the supplied model, residual exposure is INR4 crore. This is not a complete regulatory method. Legal enforceability, margin timing, disputes, wrong-way risk and the governing exposure rules need separate assessment before a real capital or limit calculation.
Wrong-way risk occurs when exposure increases as the counterparty's credit quality worsens. A hedge can be least reliable when needed if its provider is closely linked to the same adverse driver. Collateral can also lose value in the same scenario. Counterparty assessment should therefore examine common dependencies rather than assume that a contract with another institution necessarily diversifies risk. Limits, documentation, collateral practices and monitoring work together; none alone guarantees collection after a default.
9. Incorporate trading and funding liquidity
Market liquidity determines the cost and time needed to reduce a position. A quoted price for a small quantity may not apply to the full holding. Bid-ask spreads, market depth and concentration affect achievable proceeds. If a bank must sell quickly, the loss can exceed a model based on ordinary daily price changes. Reports should distinguish marked value from estimated stressed execution value and identify assets that are unavailable because of encumbrance or settlement commitments.
Funding liquidity concerns cash needed to maintain or settle positions. A hedge that stabilises eventual economic value can require margin today. A bond financed overnight may face rollover pressure even when its market price is unchanged. Treasury should connect price stress with collateral, funding and payment schedules. A position can be economically sensible over its horizon but impossible to hold through an interim cash squeeze. This is why market-risk mitigation and liquidity planning must be reviewed together rather than by disconnected reports.
For an illustrative marked holding of INR50 crore, a 3% stressed discount and INR0.10 crore execution cost imply proceeds of INR48.40 crore. The INR1.60 crore reduction is separate from any loss already recognised in the mark; avoid double counting the same price movement. State the starting valuation and scenario assumptions clearly. A stress calculation should explain whether the discount represents additional execution cost, a market shock or both, so that users can reconcile it with other reported losses.
10. Report decisions and residual risk
A treasury report should connect exposure, limits, changes, causes and required actions. Include significant sensitivities, counterparty positions, VaR or approved measures, stress results, liquidity assumptions and unresolved exceptions. A large gain can conceal a growing risk, while a loss can arise from an authorised hedge offset by an underlying benefit elsewhere. Profit attribution and risk reporting should therefore reconcile across the relevant perimeter. Neither an isolated derivative gain nor an isolated trading loss is enough to judge hedge effectiveness.
Model and limit changes need controls. A change that lowers measured risk should be independently reviewed before it creates apparent headroom for more business. Preserve the prior method, rationale, testing and approval. Staff should know when a model is unsuitable and which fallback applies. A report that silently changes definitions across periods can make a trend look favourable without any economic improvement. Clear documentation supports both operational use and later examination of the decisions made under uncertainty.
For revision, identify the position and driver, establish the units, select a suitable measure, calculate and interpret, then test the limitations and response. Duration is a local approximation, VaR a modelled quantile and a hedge a conditional offset. Counterparty and liquidity exposures remain even when market sensitivity falls. A strong case answer explains the result and the action it supports, while recognising what further information is required before the bank can approve or maintain the position.
11. Conduct a morning risk review
Meridian's desk proposes increasing its bond holding after a quiet week. Current DV01 is INR2.8 lakh against an assumed INR5 lakh limit. The proposed addition contributes INR1.5 lakh, producing INR4.3 lakh and 86% utilisation if sensitivities add. That passes this particular limit, but the review should also examine concentration, stress loss, funding and settlement capacity. A quiet week may lower historical volatility without reducing the consequences of a large yield movement. Available headroom is permission to consider a trade, not proof that it is desirable.
If the bank chooses to hedge part of the addition, calculate the opposite sensitivity and examine basis risk and margin needs. Record the intended combined exposure rather than only the number of contracts. Independent risk should verify the inputs, while operations confirms that documentation and settlement arrangements support the plan. Any unresolved confirmation or data issue should be addressed before it undermines the position report. Risk management relies on a complete transaction record as well as sound models.
The final recommendation should specify position size, limits, hedge objective, residual exposure and review triggers. After execution, compare actual positions and cash flows with the approved plan. Investigate differences and reassess if market depth, funding or the underlying exposure changes. This closes the loop from identification to control and reporting. The calculators provide a way to practise sensitivity, hedge sizing, VaR, counterparty exposure and stressed proceeds, but their outputs remain useful only when the assumptions fit the actual treasury decision.
25 original questions
Choose one best answer. Explanations address the answer and distractors; no negative marking.
Browser-local saving is optional.
Ready: 25 questions.
Sources and version notes
- IIBF CAIIB syllabus 2026: Module C, PDF page 9; seven topic groups. IDs are editorial navigation.
- RBI commercial-bank ALM direction: 2025 direction retrieved in Module B; conceptual governance reference. Tools here do not implement regulatory ALM returns.
- RBI government-securities primer: 1 April 2020; conceptual mechanics only, not a complete current market-rule register.
Prepared 7 November 2026. Review edition; independent academic approval and current rule-consolidation verification remain pending. Original examples are illustrative. External source links require internet.