Value at Risk (VaR) in Banking: CAIIB BFM Guide

Every trading desk and treasury unit in a bank runs on one question every morning: how much could we lose today if the market turns against us? Value at Risk (VaR) is the statistical answer banks rely on. For CAIIB BFM candidates, VaR ties together market risk measurement, regulatory capital and the treasury's day-to-day risk dashboard — and it is a favourite examiner topic precisely because it sits at the intersection of statistics, regulation and real trading-book decisions.

📊 What Is Value at Risk (VaR) and Why Banks Use It

Value at Risk estimates the maximum expected loss in the value of a portfolio over a defined holding period, at a stated confidence level, under normal market conditions. A "1-day 99% VaR of ₹5 crore" means the bank expects its trading portfolio to lose more than ₹5 crore on only 1% of trading days — roughly 2-3 days a year. VaR is a single number that compresses interest rate, forex, equity and commodity risk in the trading book into one comparable figure, which is why treasuries, ALCOs and boards use it to set trading limits and monitor risk appetite daily.

VaR is primarily a trading-book market-risk tool, unlike duration and interest rate risk in the banking book (IRRBB), which measure the banking book's exposure to rate movements over the medium term. A bank's dealing room uses VaR for bond, forex and derivative positions that are marked to market daily, while ALM committees use gap and duration analysis for the banking book. Both feed into the same overall risk appetite framework but answer different questions — VaR asks "how much can we lose tomorrow," while banking-book measures ask "how much does our net worth move if rates shift over a year."

🧮 Key VaR Methodologies Every Banker Should Know

Three approaches dominate practice. Historical Simulation reorders actual past returns — typically the last 250 trading days — into a loss distribution and reads off the required percentile. It makes no assumption about how returns are distributed and captures fat tails reasonably well, but it reacts slowly when volatility regimes change because it is anchored to history. The Variance-Covariance (parametric/delta-normal) method assumes portfolio returns are normally distributed and calculates VaR from the portfolio's volatility and the correlation matrix of its risk factors. It is fast and easy to compute but understates risk for portfolios with options or other non-linear payoffs, since it linearises exposures.

Monte Carlo simulation generates thousands of random price paths from an assumed stochastic process and revalues the portfolio along each path. It handles non-linear instruments and complex derivatives best but is the most computationally demanding of the three. Two inputs decide the final number regardless of method: the confidence level (95% or 99% are standard) and the holding period (1-day for desk monitoring, 10-day for regulatory capital). Basel guidance lets banks scale a 1-day VaR to a 10-day figure using the square-root-of-time rule when a full 10-day historical series is impractical.

🏦 VaR in the Regulatory Capital Framework

Under the Basel market risk framework that RBI has adopted for Indian banks, capital for trading-book exposures can be computed either through the Standardised Duration Method or, subject to specific regulatory approval, through an Internal Models Approach (IMA) built on VaR. Banks permitted to use IMA must compute VaR at 99% confidence over a 10-day holding period and hold capital equal to a multiple of the average of the previous 60 days' VaR. That multiplier — set between 3 and 4 — is not fixed; it rises if the bank's backtesting shows the model underestimated risk too often, under what is commonly called the traffic-light approach (green, yellow, red zones based on the number of exceptions in a year).

💡 Exam Tip: Remember the Basel IMA default settings — 99% confidence, 10-day holding period, minimum 250-day historical data, and a multiplication factor starting at 3. These numbers are exam favourites.

RBI's master directions on capital adequacy and market risk management, available at rbi.org.in, lay down the eligibility criteria banks must meet before using internal VaR models for regulatory capital, including independent model validation and a robust stress-testing programme alongside the VaR engine.

⚖️ VaR Limitations, Backtesting and Stress Testing

VaR has a well-known blind spot: it says nothing about the size of a loss beyond the confidence threshold. A 99% VaR tells you a loss bigger than the stated figure will happen roughly 1% of the time, but not whether that loss is marginally bigger or catastrophically bigger. This gap is exactly why Basel III's Fundamental Review of the Trading Book (FRTB) shifted the standard risk metric from VaR to Expected Shortfall (Conditional VaR) at a 97.5% confidence level, since Expected Shortfall averages the losses in the tail instead of stopping at a single cut-off point.

⚠️ Common Mistake: Candidates often assume VaR caps the maximum possible loss. It does not — it only estimates the threshold that will be breached with a given probability; the tail loss beyond that threshold can be much larger, especially in a crisis.

Because VaR models are calibrated on "normal" market conditions, they routinely fail during crises when correlations between asset classes break down. Regulators therefore mandate Stressed VaR, computed using data from a period of significant market stress, in addition to routine VaR. Backtesting — comparing each day's actual profit or loss against the previous day's predicted VaR — is mandatory and any cluster of exceptions triggers both a capital multiplier increase and a formal model review.

📌 Remember: VaR + Stressed VaR + backtesting + Expected Shortfall form the complete market-risk toolkit examiners expect you to connect, not VaR in isolation.

🌍 Applying VaR to Forex and Bond Portfolios

In practice, a bank's overall trading-book VaR is built up desk by desk. The forex desk's VaR is driven by the volatility of exchange rates and the bank's net open position across currency pairs — a topic covered in depth in the Exchange Rates and Forex Business chapter, which explains how spot, forward and cross-rate movements translate into a measurable exposure. Trade-finance-linked currency exposures arising from letters of credit, ECBs and import-export facilities also flow into this VaR calculation, and the Case Study Forex chapter works through a full worked example of how such exposures are quantified end to end.

On the fixed-income side, a bond portfolio's parametric VaR is built directly from the price sensitivity concepts covered under bond pricing formula — the more price-sensitive the portfolio, the higher its volatility input and, therefore, its VaR. Banks also lay off part of this market risk using instruments explained in our guide on currency futures and options hedging, and treasuries that route forex settlement through nostro accounts should also revisit correspondent banking relationships, since settlement risk sits alongside market risk in a well-run treasury's overall risk map.

Sound treasury risk management does not stop with market risk alone. Just as VaR quantifies how much a trading book can lose, banks need an equally disciplined credit management lifecycle to control losses on the lending book — the two disciplines together form the backbone of enterprise risk management tested across CAIIB ABM and BFM.

🧠 Practice MCQs: Value at Risk (VaR)

Want chapter-wise mock tests with 100+ MCQs? Start practising free →

VaR turns market risk into one comparable, dashboard-ready number — but as this guide shows, it is only reliable when paired with backtesting, stress testing and Expected Shortfall. Master this topic alongside the full CAIIB course, browse more Bank Financial Management chapter guides, and put your understanding to the test with free chapter-wise mock tests before exam day.