How do expected loss and stress loss lead to different decisions?

Expected loss is a probability-weighted average, often estimated as probability of default × loss given default × exposure at default. A $5 million loan with 2% default probability and 40% loss given default has $40,000 expected loss. If the borrower defaults and that loss severity applies, the loss is $2 million. Pricing and provisions may use expected loss, while capital, limits and contingency planning must also consider severe outcomes.

Why can a hedge reduce market risk but increase liquidity risk?

A derivative can gain value when the hedged asset loses value, yet require cash margin before the asset’s offsetting cash flow arrives. For example, a $3 million margin call against a $4 million liquid buffer leaves $1 million immediately available, even if the hedge is economically effective over its full life. Assess hedge value, basis risk, margin timing and available cash separately.

What is the practical difference between credit risk and counterparty credit risk?

A conventional loan exposes the lender to the borrower failing to repay an outstanding balance. A derivative can expose either party as market prices change: the amount at risk is generally the positive replacement value after applicable netting and collateral, plus possible future exposure. Thus a derivative with zero value today can still create future counterparty exposure.

Professional Risk Manager (PRM) Study Guide

Start with concept 1 Choose a topic

A risk manager needs to connect a position’s economics to the losses it can produce, the cash it may demand, and the controls that keep those exposures visible. This guide builds that connection from finance theory and instruments through market practice and quantitative methods, then applies it to enterprise, credit, counterparty, market, liquidity and operational risk. The worked examples use stated assumptions so you can follow each calculation and judge what its result does, and does not, establish. Pay particular attention to distinctions that change decisions: expected loss versus a stress loss, a mark-to-market loss versus a funding outflow, and diversification versus a hedge that fails under stress. The final sections bring those distinctions together in an applied case and in professional governance decisions.

Finance Theory

1. Utility and the price of uncertainty

Expected wealth averages monetary outcomes, while expected utility averages the satisfaction assigned to those outcomes. A risk-averse investor has a concave utility function: gaining a dollar when already wealthy adds less utility than losing a dollar removes. Consequently, a fair gamble can have lower utility than receiving its expected payoff for certain. To compare the two, calculate expected utility, then find the certain wealth level with that same utility. This is the certainty equivalent. Expected wealth minus the certainty equivalent is the investor’s maximum willingness to pay to eliminate the risk under the stated utility function. The result depends on both the payoff distribution and the investor’s preferences; it is not an objective market price. The Professional Risk Manager designation includes finance theory because risk decisions require this distinction.

Worked example: A hypothetical investment pays $120 or $80 with equal probability, so expected wealth is $100. If utility is the square root of wealth, expected utility is (√120 + √80)/2 = (10.9545 + 8.9443)/2 ≈ 9.9494. Certain wealth of 9.9494² ≈ $98.99 gives the same utility. This investor would pay up to about $1.01 to replace the gamble with $100 certain.

Watch the distinction: A tempting shortcut is to call $100 the certainty equivalent because it is the average payoff. That ignores the utility lost when a concave-utility investor bears uncertainty.

Reference: [1]

2. Risk-adjusted performance

Choose a performance measure by asking which risk the decision maker bears and what benchmark matters. Sharpe divides excess return by total volatility; Treynor divides it by market beta. Jensen’s alpha compares return with the security market line, while the information ratio divides active return by tracking error against a specified benchmark. Sortino uses downside deviation below a target; Omega compares gains above a threshold with losses below it, and Kappa uses a chosen lower partial moment. Capital-based measures need especially careful definitions: RAROC relates risk-adjusted return to economic capital, RORAC relates return to risk-adjusted capital, and RARORAC adjusts both. RoVaR relates return to value at risk. Denominators, horizons and capital definitions must match before comparing figures. The PRM Exam 1 guide connects these measures with portfolio and pricing theory.

Worked example: A hypothetical portfolio returns 10%, its risk-free rate is 2%, volatility is 16%, beta is 1.2 and market return is 8%. Sharpe is (10% − 2%)/16% = 0.50. CAPM predicts 2% + 1.2(8% − 2%) = 9.2%, so Jensen’s alpha is 10% − 9.2% = 0.8 percentage points. Sharpe measures reward per unit of total volatility; alpha measures return beyond a beta-based benchmark.

Watch the distinction: A high Sharpe ratio does not establish positive Jensen’s alpha. A portfolio can reward total volatility well while earning less than its market beta would require.

Reference: [1]

3. Diversification, time and hedging

Portfolio variance depends on each holding’s variance and on how their returns move together. For two assets, multiply each variance by its squared weight, then add twice the product of weights, volatilities and correlation. Correlation below one creates diversification because losses need not arrive together, although correlations can change in stress. Time adds another dependence question. If successive returns are positively correlated, multi-period variance exceeds the sum of single-period variances; negative serial correlation reduces it. Scaling volatility by the square root of time assumes suitably independent, stable returns. A hedge should be assessed against the specific exposure it offsets, including currency, maturity and basis. A falling hedge instrument is not automatically a failure if the underlying exposure gains enough to compensate.

Worked example: Two hypothetical assets each receive a 50% weight. Their annual volatilities are 20% and 10%, with correlation 0.2. Portfolio variance is 0.5²(0.20²) + 0.5²(0.10²) + 2(0.5)(0.5)(0.2)(0.20)(0.10) = 0.0145. Volatility is √0.0145 ≈ 12.04%, below the 15% weighted average of individual volatilities because their returns are imperfectly correlated.

Watch the distinction: Averaging the two volatilities to obtain 15% discards covariance. The correct portfolio risk depends on correlation, and that historical correlation may be unreliable during stress.

Reference: [1]

4. Efficient portfolios and the risk-free asset

The efficient frontier contains risky portfolios offering the greatest expected return for each volatility level. A portfolio below that frontier is dominated if another offers more expected return for the same risk. Introducing a risk-free asset lets an investor combine it with a chosen efficient risky portfolio. Because the risk-free asset has zero return volatility, both expected return and volatility change linearly with the weight in that risky portfolio; the resulting line is a capital allocation line. The separation principle says investors with the same opportunity set can choose the same tangency risky portfolio, then vary their risk-free holdings according to risk preference. Borrowing to invest more than 100% in the risky portfolio extends the line only if borrowing at the assumed rate is feasible.

Worked example: Suppose a hypothetical risky portfolio offers 8% expected annual return and 12% volatility, while the risk-free rate is 2%. To target 9% volatility, put 9%/12% = 75% in that portfolio and 25% in the risk-free asset. Expected return is 0.75(8%) + 0.25(2%) = 6.5%. Another available portfolio with 9% volatility and 6% expected return is dominated under these assumptions.

Watch the distinction: Selecting the risky asset with the highest expected return alone ignores its risk. Efficient choice compares attainable return and volatility together, then adjusts overall risk through allocation.

Reference: [1]

5. Single-factor and multifactor pricing

CAPM prices exposure to broad market movements through beta. Diversifiable company-specific risk does not earn a separate premium in the model, so the security market line gives required return as the risk-free rate plus beta times the market premium. A statistical single-index model also relates returns to a market index, but its regression estimates describe observed co-movement; they do not automatically establish CAPM’s equilibrium required return. Multifactor models allow several systematic drivers, such as rate or credit-spread shocks. Arbitrage-pricing reasoning asks whether assets with comparable factor exposures have consistent expected returns, subject to its assumptions and trading frictions. When comparing a forecast with a model benchmark, check factor definitions and estimation error first. Apparent excess return can reflect an omitted exposure rather than a mispriced security.

Worked example: With a hypothetical 2% risk-free rate, 5% market premium and beta of 1.2, CAPM requires 2% + 1.2(5%) = 8%. Suppose a second priced factor adds 0.5 × 2% = 1 percentage point, giving a two-factor requirement of 9%. A 10% forecast exceeds that benchmark by 1 percentage point, conditional on the factors and forecast being credible.

Watch the distinction: A regression beta is not itself proof of a required return. CAPM adds equilibrium assumptions, while a fitted index relationship may only describe the sample used to estimate it.

Reference: [1]

6. Debt, equity and firm value

Debt can increase firm value when interest deductions create usable tax benefits, but leverage also makes promised payments less flexible. As debt grows, the chance and cost of financial distress may rise, even before formal default. Creditors and shareholders can develop conflicting incentives: shareholders may prefer riskier projects when creditors bear part of the downside, while protective covenants and monitoring consume resources. Debt can also discipline managers who might otherwise spend excess cash poorly. The decision therefore concerns the incremental value of financing, not simply the quoted interest rate or accounting return on equity. Compare the present value of expected tax benefits with expected distress and agency costs, and consider how lost borrowing capacity limits future investment. Estimates are scenario-dependent, especially when cash flows weaken.

Worked example: Assume an additional hypothetical debt issue creates $2 million of annual interest and the firm can use a 25% tax deduction. Its annual tax shield is $2 million × 25% = $0.5 million. If the same issue adds an estimated $0.7 million of annual expected distress and agency costs, the stylized annual net effect is $0.5 million − $0.7 million = −$0.2 million, before timing differences.

Watch the distinction: A lower quoted debt rate does not mean more debt always lowers total financing cost. Greater leverage can increase required returns and expected distress costs.

Reference: [1]

7. Yield curves and rate models

A spot yield discounts one cash flow at its own maturity. Start with zero-coupon prices, or bootstrap from coupon bonds by discounting earlier coupons at already determined spot rates and solving for the remaining rate. Forward rates follow algebraically from spot discount factors: investing for two years must match investing for one year and locking in the second year, under the same compounding convention. A forward rate is a price-implied break-even rate, not a guaranteed forecast. Pure expectations theory connects it to expected future short rates; liquidity preference adds a maturity premium. Preferred-habitat theory allows investors to move beyond favored maturities for compensation, while segmentation emphasizes maturity-specific supply and demand. Ho-Lee, Hull-White and Black-Derman-Toy specify different short-rate dynamics and calibration constraints for valuing rate-sensitive claims.

Worked example: Suppose hypothetical one-year and two-year zeros pay $100 and cost $97.087 and $92.456 respectively. With annual compounding, their spot yields are approximately 3% and 4%. The one-year forward rate beginning in year two is 1.04²/1.03 − 1 = 1.0816/1.03 − 1 ≈ 5.01%. That is the rate implied by these prices, subject to the chosen conventions.

Watch the distinction: Reading the 5.01% forward as a certain future short rate confuses a market-implied break-even price with an outcome. Term premiums can also affect it.

Reference: [1]

Financial Instruments

8. Bond cash flows and embedded options

Price a bond by discounting each promised coupon and principal payment at rates appropriate to its timing and risk. A zero has only principal at maturity; a fixed-rate bond has known nominal coupons; a floating-rate bond resets coupons to a reference rate plus any spread. Yield to maturity compresses a bond’s price and promised cash flows into one rate, but it assumes those promised payments and does not explain which part of the curve moved. Duration approximates the first-order price response to a small yield change, and convexity captures curvature. A holder’s put can support value when rates rise or credit weakens because the holder can demand early repayment. An issuer’s call limits upside when falling rates make refinancing attractive to the issuer.

Worked example: A hypothetical two-year bond pays a $5 annual coupon and returns $100 principal. At a flat 4% annual yield, price is $5/1.04 + $105/1.04² = $4.8077 + $97.0784 ≈ $101.89. If the issuer may call it for $101 after year one, a favorable rate fall can make that call likely, limiting the holder’s price gain.

Watch the distinction: Applying straight-bond duration to a callable bond through a large rate fall can overstate its gain. The issuer’s exercise decision changes the expected cash flows.

Reference: [1]

9. Forwards, futures and carry

A forward price follows from comparing delivery later with buying the asset now and carrying it to delivery. Financing and storage raise the carry cost; income or benefits from holding the asset reduce it. State when income is paid and whether it is a cash amount or proportional yield, because these produce different formulas. If a quoted forward materially exceeds replicating carry cost, a trader can compare buying and financing the asset with selling the forward. Execution, borrowing, storage and delivery constraints determine whether the apparent gap is tradable. A forward normally settles at maturity, creating a bilateral exposure as its value changes. A standardized futures contract is marked to market daily, so gains and losses create variation-margin cash flows before final delivery or closeout.

Worked example: For a hypothetical $100 asset, assume one-year financing grows by 5% and the asset provides a proportional 2% holding yield represented by a 1.02 factor. The simple carry price is $100 × 1.05/1.02 ≈ $102.94. A $106 forward is $3.06 higher; compare that gap with transaction, financing and delivery costs before calling it an arbitrage.

Watch the distinction: Subtracting any quoted 2% income from financing without checking its payment timing can misprice the forward. Cash dividends and proportional yields require different carry treatment.

Reference: [1]

10. Swap cash flows and value

A plain interest-rate swap exchanges fixed and floating interest on a stated notional; the notional usually measures payments rather than changing hands. Identify which leg is received, each accrual period, the floating reset and the payment date before determining a cash flow. At valuation, project remaining floating payments from the relevant forward curve and discount both legs using appropriate discount factors. The swap’s value is the present value of receipts minus payments, not simply the next coupon difference. A currency swap can exchange principal and interest in different currencies. Equity and commodity swaps replace a rate leg with returns linked to those assets, while an asset swap combines a bond position with swap cash flows. Each structure adds its own market, funding and counterparty exposures.

Worked example: Suppose a hypothetical swap receiver earns 4% fixed and pays 3% floating on $100 million for a six-month accrual. The net payment is (4% − 3%) × 0.5 × $100 million = $500,000. If the annual simple discount rate for that six-month payment is 2%, its discount factor is 1/1.01, giving present value $500,000/1.01 ≈ $495,050. Other remaining payments determine the full swap value.

Watch the distinction: Using the notional as the amount exchanged in a plain interest-rate swap inflates cash exposure. The stated notional scales interest payments; it is generally not delivered.

Reference: [1]

11. Option payoffs and pricing

A call pays when the underlying finishes above its strike; a put pays when it finishes below. Premium depends on possible future payoffs, so remaining time and expected volatility matter alongside spot, strike, rates and distributions. In a one-step binomial model, construct an up state and a down state, then choose a risk-neutral probability that makes the expected underlying growth equal the period financing return. Discount the option’s probability-weighted payoff. This is a pricing device, not a claim about actual event probabilities. A protective put preserves upside while setting a floor on an owned asset, at the cost of its premium. A vertical spread buys one option and sells another at a different strike, reducing premium while capping some payoff. Check whether exercise style or dividends affect the chosen method.

Worked example: A hypothetical stock starts at $100 and ends at $120 or $90. A $100-strike call pays $20 or $0. With a 10% risk-free return over the period, the risk-neutral up probability is (1.10 − 0.90)/(1.20 − 0.90) = 2/3. Price is [(2/3) × $20 + (1/3) × $0]/1.10 ≈ $12.12, assuming frictionless trading and no distributions.

Watch the distinction: Treating the risk-neutral probability of two-thirds as a forecast of the stock’s actual chance of rising misreads the model. It is chosen to enforce consistent pricing.

Reference: [1]

12. Credit protection and securitization

A credit default swap buyer pays premiums for a defined protection payment following a covered credit event. Its fair spread balances the present value of expected premium payments against expected protection payments. Default probability determines how often a payment may occur; loss given default determines its size after recovery. A quick one-period product of probability and severity is useful intuition, but an actual valuation also needs payment timing, survival, discounting, contract terms and counterparty risk. Buying protection can reduce exposure to a named borrower without selling the underlying loan, though basis and counterparty risk remain. Securitization instead pools exposures and assigns cash flows and losses to tranches with different seniority. It reallocates credit risk among investors; it does not make underlying borrower defaults disappear.

Worked example: Assume a hypothetical one-year default probability of 2% and loss given default of 40% on $10 million. Ignoring timing and discounting, expected protection payout is $10 million × 0.02 × 0.40 = $80,000, or 0.8% of notional, equal to 80 basis points. This is only a rough spread comparison because actual CDS premiums and potential default payments occur at different times.

Watch the distinction: Equating 80 basis points with every traded CDS quote ignores survival, payment schedules and discounting. Market spreads may also include liquidity and recovery uncertainty.

Reference: [1]

13. Caps, floors and swaptions

A rate cap is a series of options that pays when a reference rate exceeds a strike, making it useful for a floating-rate borrower. A floor pays when the rate falls below a strike, protecting a floating-rate lender’s receipts. A swaption grants the right to enter a swap later, useful when a future borrowing or investment is uncertain and a fixed rate may be desirable. For each caplet or floorlet, determine the notional, accrual fraction, observed rate, strike and settlement timing. The payoff is the favorable rate difference times notional and accrual fraction, subject to the contract’s conventions. An option premium buys protection while preserving some favorable rate movement. A swap fixes the exchanged rate exposure more directly but gives up that flexibility.

Worked example: A hypothetical borrower holds a cap on $10 million of floating debt with a 4% strike. The reference rate fixes at 6% for a quarter with a 0.25 accrual fraction. The caplet pays (6% − 4%) × 0.25 × $10 million = $50,000 at its contractual payment date, before any present-value discount. This offsets $50,000 of the borrower’s higher interest cost for that period.

Watch the distinction: A cap payment is not simply the 2-percentage-point rate excess times full-year notional. The accrual fraction determines the period’s cash amount.

Reference: [1]

14. Virtual-asset exposures

A virtual-asset position can fail through several channels even when its quoted price appears adequate. Custody risk concerns control of keys and access to assets; smart-contract risk concerns code behavior and governance; oracle risk concerns the external prices used by automated rules. Collateral quality depends on liquidation speed, market depth, legal enforceability and correlation with the borrower’s condition. A marked token value is therefore only a starting point for estimating realizable collateral. In a stress, the token price may drop, trading depth may vanish and liquidation transactions may be delayed or costly. Trace the sequence from price observation through margin decision, transfer and sale. The resulting shortfall belongs in credit-risk analysis, while outages or faulty code can also create operational loss.

Worked example: A hypothetical borrower posts 150 tokens at $80 each against an $8,000 loan. Marked collateral is 150 × $80 = $12,000, or 150% of the loan. If tokens fall to $40, marked collateral becomes 150 × $40 = $6,000, leaving a $2,000 shortfall before liquidation costs. A delayed oracle update or thin market could make proceeds lower still.

Watch the distinction: Calling the initial $12,000 mark cash-equivalent assumes immediate sale at the displayed price. Liquidation time, trading depth and custody access can change recoverable value.

Reference: [1]

Financial Markets

15. Market participants and trade workflow

A trade begins with an order and execution, then moves through booking, confirmation, clearing where applicable, and settlement. The front office initiates and trades risk; the middle office independently measures positions, limits and valuations; the back office confirms obligations and settles cash or securities. Separation makes errors and unauthorized changes easier to find. Reconciliation compares executed quantities and prices with internal records and external confirmations, ideally before settlement creates a cash or delivery problem. Clearing can reduce bilateral replacement exposure but introduces margin and clearing-member dependencies. Electronic platforms and fintech intermediaries can automate routing or settlement steps, yet they also add interfaces whose records must agree. A price-risk limit tests market exposure; it does not by itself test whether a trade was booked accurately.

Worked example: A hypothetical broker executes 1,000 shares at $50, but the booking system records 100. Reconciliation finds 1,000 − 100 = 900 missing shares, worth 900 × $50 = $45,000 at execution price. The trade record and confirmation must be corrected before settlement. The mismatch is an operational booking error even if the market price never changes.

Watch the distinction: Treating a balanced market-risk report as proof that every trade settled correctly confuses exposure measurement with transaction control. An omitted booking can make the report itself incomplete.

Reference: [1]

16. Bond-market spreads and seniority

A bond’s spread over a reference yield can reflect expected credit loss, uncertainty about that loss, liquidity, structural terms and market risk appetite. An agency rating summarizes an assessment of credit quality, but investors still price the bond’s specific maturity, covenants and trading conditions. Spread duration estimates the percentage price change for a small spread change, holding other yield components fixed. For a widening, multiply spread duration by the spread increase in decimal form and apply a negative sign. Debt seniority concerns claim priority and recovery if the issuer fails; capital structure concerns the issuer’s overall debt and equity financing. A senior claim from a heavily leveraged issuer can still be risky. Check whether the spread move reflects issuer news or a broader market repricing.

Worked example: A hypothetical $10 million bond has spread duration of four years. Its spread widens from 150 to 250 basis points, a 100-basis-point or 0.01 increase, while other yields stay fixed. Approximate value change is −4 × 0.01 × $10 million = −$400,000, leaving about $9.6 million before convexity and other effects. A downgrade may accompany this move but does not mechanically set its size.

Watch the distinction: Assuming two bonds with the same rating must have the same spread ignores maturity, seniority, covenants and liquidity. Ratings inform credit assessment; they are not prices.

Reference: [1]

17. Money-market yields and reference rates

Money-market instruments meet short-term funding and investment needs, but their cash-flow forms differ. A Treasury bill commonly trades at a discount to face value; commercial paper is corporate short-term debt; a certificate of deposit is a bank deposit instrument; and a banker’s acceptance is a bank-accepted payment obligation. For a discount security, the investor earns face value minus purchase price, so an investment-based yield uses purchase price as its denominator. Specify the day-count convention before annualizing. Overnight reference rates describe very short borrowing or swap-linked pricing, while term benchmarks refer to a stated future period and can include different credit or liquidity components. LIBOR and EONIA in older examples should be read as historical benchmarks. The PRM candidate guidebook identifies the instrument and benchmark topics.

Worked example: A hypothetical bill pays $100 in 180 days and costs $98 today. Its 365-day bond-equivalent yield is ($100 − $98)/$98 × 365/180 = 0.04138, or about 4.14%. The $2 gain is measured against the $98 invested. A face-value denominator would answer a different discount-rate question, and a 360-day convention would give another annualized figure.

Watch the distinction: Using $100 as the denominator for an investment yield understates the return on $98 actually paid. Also check whether the requested annualization uses 360 or 365 days.

Reference: [1]

18. Stock-market liquidity

In a primary equity issue, the issuer sells new shares and receives proceeds. In secondary trading, existing investors exchange shares, providing liquidity and price discovery without sending that trade’s proceeds to the issuer. An order book displays prices and available quantities. A buyer seeking immediate execution generally pays the ask; a seller receives the bid. The bid-ask spread therefore measures one component of immediate trading cost. Displayed depth matters as much as the best quote: a large order may consume several price levels, creating slippage. Different trading venues and order types change execution likelihood, disclosure and fees. Midpoint value is a useful reference for estimating execution cost, but it is not necessarily an executable price for the desired size.

Worked example: A hypothetical stock shows a $49.90 bid and $50.10 ask, with at least 1,000 shares offered at the ask. Buying 1,000 costs 1,000 × $50.10 = $50,100. The midpoint is ($49.90 + $50.10)/2 = $50.00, giving a $50,000 midpoint value. Immediate purchase costs $100 above that reference, before fees; insufficient ask depth could make the actual cost higher.

Watch the distinction: Valuing an immediate purchase at the midpoint assumes access to a price that may not be offered. Use the ask and available depth when estimating execution cost.

Reference: [1]

19. FX quotation and cross rates

An FX pair states how many units of the quote currency buy one unit of the base currency. EUR/USD of 1.10 therefore means $1.10 per euro, not euros per dollar. Cross rates follow by multiplying or dividing pairs so the intermediate currency cancels. Write the units beside each quote to choose the correct operation and catch an inverted answer. A spot transaction converts currencies on its settlement terms; a forward locks an exchange rate for a later date and creates exposure to rate movements and counterparty performance before settlement. Real dealing quotes have separate bid and ask rates, so a synthetic cross formed from midpoints is only an illustration. Funding-rate differences help determine forward points under covered interest parity, subject to market conventions and costs.

Worked example: Suppose hypothetical midpoint quotes are EUR/USD = 1.10 dollars per euro and USD/JPY = 150 yen per dollar. Multiplying cancels dollars: 1.10 × 150 = 165 yen per euro, so EUR/JPY is 165. Buying €2 million costs 2,000,000 × 165 = ¥330 million at these illustrative rates, before bid-ask spreads or fees.

Watch the distinction: Dividing 150 by 1.10 here leaves the wrong currency logic. Track units explicitly: dollars per euro times yen per dollar gives yen per euro.

Reference: [1]

20. Energy and commodity markets

A commodity forward reflects the economics of holding physical inventory until delivery. Financing and storage costs tend to raise the forward price, while the convenience of having inventory available can lower it. That convenience yield may change sharply when stocks are scarce. Energy prices also respond to seasonal demand, transport bottlenecks and delivery location. The quoted benchmark contract specifies a particular grade, place and date; a physical user may need something different. The price difference between its exposure and the hedge contract is basis. Even a well-sized hedge can leave losses if that basis changes. Before using a carry formula, state whether costs and convenience yield are simple annual rates or continuously compounded, and whether the asset can actually be stored and delivered as assumed.

Worked example: For a hypothetical commodity worth $80 now, assume one-year simple financing of 5%, storage of 2% and convenience yield of 3%, all measured against spot. A simplified carry estimate is $80 × (1 + 0.05 + 0.02 − 0.03) = $83.20. A local grade may trade differently at delivery, so hedging it with this benchmark forward leaves basis risk despite the $83.20 estimate.

Watch the distinction: Treating a benchmark forward as a perfect hedge for every physical grade ignores location, quality and delivery timing. Changes in those differences can survive the hedge.

Reference: [1]

21. Futures exchanges and clearing

Exchange-traded futures specify the underlying asset, contract size, delivery terms and expiry. Standard terms make positions easier to offset, while the clearinghouse becomes the counterparty to each cleared trade. A trader posts initial margin as a performance deposit, not as the purchase price of the underlying asset. Each day, the exchange marks the contract to its settlement price: losses are paid and gains received as variation margin. If the margin balance falls below the applicable maintenance level, more cash may be required. This changes a hedger’s cash timetable. An inventory holder can gain economically when the commodity price rises yet still face a cash call on a short futures hedge before selling the inventory. When assessing a hedge, check its price offset and the liquidity available to meet margin. The PRM candidate guidebook includes futures markets in its scope.

Worked example: A firm holds 10,000 barrels of oil and shorts ten futures contracts covering 1,000 barrels each. The futures price rises by $2 per barrel in one day. Its short position owes 10 × 1,000 × $2 = $20,000 in variation margin. If the inventory price also rises $2, its marked value gains about $20,000. The hedge offsets the price move economically, but the inventory gain supplies no cash until the firm sells or borrows against it.

Watch the distinction: A matched futures hedge does not eliminate cash needs. Daily settlement can require payment before a gain on the physical asset becomes spendable.

Reference: [1]

Mathematical Foundations of Risk Measurement

22. Time value and algebra

Money received later is worth less than money available now when the current amount can earn a positive return. With annual compounding, a present amount PV becomes FV = PV(1 + r)^n after n years, so PV = FV/(1 + r)^n. Solve the same equation for an unknown rate with r = (FV/PV)^(1/n) − 1. Before calculating, match the rate period to the cash-flow period: a monthly rate needs a monthly number of periods, and a nominal annual quote needs its compounding convention. Cash flows at different dates must each be discounted to the same valuation date before they are added. Algebra also lets you test an implied return against a quoted one. Check that the starting amount is the invested cash, the terminal amount includes all promised proceeds, and any interim payments are valued at their actual dates.

Worked example: A zero-coupon claim pays $100 in two years. At an assumed 5% effective annual rate, its present value is $100/1.05² = $90.7029, or $90.70. If its exact price is $90.7029, the implied annual rate is (100/90.7029)^(1/2) − 1 = 5%. The total two-year return is 10.25%; dividing that by two gives 5.125%, which is not the compounded annual rate.

Watch the distinction: Do not apply a one-year discount factor to a two-year payment. Both the exponent and the rate convention must describe the payment’s timing.

Reference: [1]

23. Sensitivity through calculus

A derivative measures how a value changes for a small change in an input. For a two-year zero-coupon payment, P(y) = F(1 + y)^−2, so dP/dy = −2F(1 + y)^−3. Multiplying this slope by a yield change gives a first-order price estimate. Yield changes must be entered as decimals: one percentage point is 0.01. The slope is negative because a higher discount rate lowers present value. A straight-line estimate loses accuracy as the move grows because the price-yield curve bends. Its second derivative is positive for this bond, so adding half the second derivative times the squared yield change supplies a curvature correction. For a portfolio, distinguish a derivative with respect to one yield from sensitivity to a parallel shift across the curve. Always state which input moved and what stayed fixed.

Worked example: For F = $100 and y = 5%, P = 100/1.05² = $90.703. The slope is −200/1.05³ = −$172.77 per unit of yield. A rise to 6% gives a first-order change of −172.77 × 0.01 = −$1.73. Exact repricing gives 100/1.06² = $89.000, a $1.70 fall. The estimate is slightly too negative because it omits positive curvature.

Watch the distinction: A derivative quoted per unit of yield is not the loss for a one-percentage-point move. Multiply the slope by 0.01 for that move.

Reference: [1]

24. Summary statistics and tail behavior

The mean describes a distribution’s center; variance averages squared deviations from that center, and standard deviation returns dispersion to the original units. Decide whether the observations represent the whole population or a sample before choosing the variance denominator. Correlation standardizes covariance and shows the direction and strength of linear co-movement, but it cannot establish how often two positions suffer extreme losses together. Skewness describes asymmetry: positive skew in a loss distribution points toward a longer high-loss side. Kurtosis helps describe the weight of extreme observations relative to central variation. Extreme-value methods focus on observations beyond a high threshold or on block maxima, then model the tail under explicit assumptions. They require care with threshold choice, dependence and sparse data. A risk report should therefore show central tendency, dispersion and tail information separately when rare large losses matter.

Worked example: Treat four hypothetical losses, $1 million, $2 million, $2 million and $5 million, as the entire population. Their mean is $10 million/4 = $2.5 million. The squared deviations sum to 2.25 + 0.25 + 0.25 + 6.25 = 9 million². Population variance is 9/4 = 2.25 million², and standard deviation is $1.5 million. The $5 million observation warrants separate tail attention despite the modest mean.

Watch the distinction: A low average loss does not imply a mild worst case. Squared deviations and tail observations answer questions that the mean cannot.

Reference: [1]

25. Conditional probability and dependence

Conditional probability changes a risk estimate when new information arrives. Bayes’ rule combines a prior event rate with a signal’s sensitivity and false-positive rate; the base rate matters even when the signal looks strong. Dependence is a separate question: joint losses cannot generally be found by multiplying marginal probabilities when events share drivers. Distribution choice also matters. A normal variable can be negative, whereas a lognormal variable is strictly positive and often suits a modeled positive amount. A Poisson model describes counts under its arrival assumptions. Probit and logit link predictors to event probabilities, using different response curves. A copula links marginal distributions into a joint distribution, but its chosen form can strongly affect estimated joint tail risk. Identify whether the task asks about a count, an amount, a conditional event or dependence before selecting a model.

Worked example: Suppose 2% of firms default, a warning flags 80% of defaulters, and it also flags 10% of survivors. Among 10,000 firms, expect 200 defaulters, of whom 160 are flagged; expect 9,800 survivors, of whom 980 are flagged. Of 1,140 flagged firms, 160 default. The conditional default probability is 160/1,140 = 14.04%, consistent with 0.02(0.80)/[0.02(0.80) + 0.98(0.10)].

Watch the distinction: The 80% flagging rate among defaulters is sensitivity, not default probability among flagged firms. False positives and the low prior default rate change the answer.

Reference: [1]

26. Numerical and state-transition methods

Choose a numerical method according to the unknown you need. Optimization searches for the best feasible decision, such as portfolio weights subject to risk and budget constraints. A Lagrange multiplier incorporates a constraint into the objective; at a solution, it can describe how much the optimum would improve if that constraint relaxed slightly. Root finding solves an equation, such as the yield that equates discounted cash flows with a bond price. Newton-Raphson uses a function’s value and slope to update a trial solution, so a poor starting point or flat slope can cause trouble. Monte Carlo simulation samples specified risk-factor paths to estimate a range of outcomes; the result is only as sound as its inputs and dependence assumptions. A multistate Markov model instead moves exposures among defined states using transition probabilities. Test that each transition row sums to one and that absorbing states behave as intended.

Worked example: Assume a borrower has a 90% probability of remaining performing over each year conditional on having survived so far; default is permanent. Two-year survival is 0.90 × 0.90 = 0.81. Two-year default probability is therefore 1 − 0.81 = 19%. Adding 10% + 10% gives 20% because it treats second-year default as possible even for borrowers who defaulted in year one.

Watch the distinction: Transition probabilities are conditional on the current state. Adding annual default probabilities double-counts paths when default is an absorbing state.

Reference: [1]

27. Matrices, correlated shocks and PCA

Matrix notation keeps a many-factor risk calculation organized. An exposure vector maps changes in market factors to approximate profit or loss; a covariance matrix records each factor’s variance and every pair’s covariance. Portfolio variance is the exposure vector multiplied by the covariance matrix and then by the vector’s transpose. To simulate correlated normal shocks, factor the correlation matrix as LLᵀ and apply the Cholesky factor L to independent standardized shocks. The resulting scenarios reproduce the target correlation across many draws, provided the matrix is valid and the underlying assumptions hold. Principal components analysis rotates correlated variables into orthogonal directions ordered by explained variance. A dominant first component may resemble a broad level move in rates, but that economic label is an interpretation, not a property guaranteed by the algebra. Inspect the component loadings before using it to hedge.

Worked example: Two standardized factors have correlation 0.6. A valid Cholesky factor is L = [[1, 0], [0.6, 0.8]], since 0.6² + 0.8² = 1. Applying L to independent shocks (1, −1) gives (1, 0.6 × 1 + 0.8 × −1) = (1, −0.2). For exposures of $2 million and $1 million per standardized shock, approximate combined P&L is $2 million − $0.2 million = $1.8 million.

Watch the distinction: A principal component is a mathematical direction of variation. Calling it an economic cause without checking its loadings can produce a misleading hedge.

Reference: [1]

28. Regression, inference and model fit

A regression coefficient estimates how the outcome changes with a predictor while the other included predictors are held fixed. In a time series, check whether residuals remain correlated across dates; otherwise a conventional standard error may overstate precision. In multiple regression, highly correlated predictors can make individual coefficients unstable even when the model fits historical data well. A confidence interval combines an estimate and its uncertainty under stated sampling assumptions. To test a zero-coefficient hypothesis, compare the estimate with its standard error or check whether zero lies in the interval. Neither a small p-value nor a narrow interval proves causation: omitted variables, reverse effects and selection can remain. For a machine-learning forecast, reserve data that were not used to fit or tune the model, respect time order, and compare performance against a simple benchmark. Data leakage can make apparent accuracy disappear in use.

Worked example: Suppose a fitted risk-factor sensitivity is 1.5 with a standard error of 0.2. An approximate 95% confidence interval using a normal critical value is 1.5 ± 1.96(0.2) = [1.108, 1.892]. It excludes zero, supporting a nonzero association under the model’s assumptions. It does not show that changing the factor causes the outcome or that the estimate will hold in a later market regime.

Watch the distinction: An in-sample fit or significant coefficient is not proof of predictive skill. Test forecasts on genuinely withheld periods and investigate leakage.

Reference: [1]

Enterprise Risk Management Frameworks and Operational Risk Management Practices

29. Who owns and challenges risk

Risk governance works when the party taking an exposure remains accountable for it and another party can challenge it. The business proposes, approves within delegated authority and monitors its positions. Independent risk sets or interprets limits, measures aggregate exposure, challenges assumptions and escalates breaches through a reporting line that can withstand commercial pressure. Compliance addresses applicable obligations, while internal audit gives later independent assurance about the design and operation of controls. Senior management and the board need timely information about material risks, exceptions and decisions. Examine incentives as well as organization charts: a risk officer whose pay or reporting depends entirely on a desk’s revenue may struggle to challenge that desk. A useful escalation record states the exposure, limit, approver, rationale and action taken. The PRM Exam 2 study guide can help you place these roles alongside operational and credit-risk decisions.

Worked example: A sales head authorizes a $12 million loan against a hypothetical $10 million desk limit and asks risk to report only $10 million. The exposure exceeds the limit by $2 million, or 20% of the limit. The business must own its decision. Risk records the full $12 million, challenges the exception and escalates it to the authorized decision-maker. Internal audit may later test how approval and reporting controls operated.

Watch the distinction: Deleting a limit breach from a report does not reduce exposure. It removes information needed for independent challenge and an authorized response.

Reference: [1]

30. Appetite, tolerance and capacity

Expected-loss appetite, stress-loss tolerance and loss-absorbing capacity answer different questions. Appetite describes the level of expected risk the organization chooses to take in pursuit of its objectives. A stress tolerance sets a boundary for a defined adverse scenario, including the horizon and assumptions. Capacity reflects the resources available to survive loss while continuing required activities. A proposal can meet expected-loss appetite and remain below capacity yet exceed stress tolerance. That is a genuine boundary breach because tolerances are designed to prompt action before survival resources are exhausted. First confirm that the figures use comparable exposures and time horizons. Then identify which boundary is crossed, how sensitive the result is to assumptions, and who has authority to reduce, transfer, hedge or explicitly accept the exposure. Record the decision and retest after the change; raising a threshold merely to erase a breach would defeat its purpose.

Worked example: A proposed portfolio has hypothetical expected annual loss of $1.5 million and stress loss of $6 million. The stated expected-loss appetite is $2 million, stress tolerance is $5 million and loss capacity is $7 million. Expected loss is $0.5 million below appetite and stress loss is $1 million below capacity, but stress tolerance is breached by $1 million. Reducing the concentrated position or obtaining effective protection addresses that breach.

Watch the distinction: Capacity is not permission to use every available loss-absorbing dollar. A stress tolerance can require action while the firm could still survive the modeled loss.

Reference: [1]

31. Operational-risk assessment

An operational-risk assessment starts with an activity and the process needed to complete it. Trace a specific cause through a possible event to its financial, customer or service impact. Identify preventive controls that make the event less likely and detective controls that reveal it quickly; then judge how reliably each works. Inherent risk describes exposure before those controls, while residual risk reflects their expected effect. The distinction matters because a low incident count may reflect either strong controls or incomplete reporting. Treatments can avoid the activity, reduce frequency or severity, transfer part of the financial consequence, or accept the residual exposure within authority. After choosing a treatment, assign an owner, test that the control operates, and reassess the remaining risk. Keep cause, event and impact separate so a one-time recovery does not get mistaken for a durable fix.

Worked example: A payment system accepts the same vendor invoice twice and sends two $200,000 payments. The extra $200,000 is cash at risk pending recovery; the initiating cause is absent duplicate validation, and the event is duplicate processing. Recovering the second payment restores cash if successful. Blocking repeated invoice identifiers, with review for legitimate exceptions, reduces recurrence. Reconcile outgoing payments afterward to detect failures that bypass the block.

Watch the distinction: Recovering a mistaken payment addresses its consequence. Unless the processing weakness is corrected and tested, the same event can recur.

Reference: [1]

32. Operational loss data and capital

Operational losses arise from both how often events occur and how severe they are. If annual event frequency and severity are modeled separately, expected annual loss is expected count times expected loss per event, assuming the model’s conditions hold. Capital for rare severe outcomes needs the shape of the aggregate-loss distribution, including dependence, tail severity and uncertainty; the mean alone cannot describe it. Internal loss data should be checked for reporting thresholds, missing small events, changes in process volume and shifts in control quality. External incidents and scenarios can illuminate severe events absent from a firm’s own record, but must be adjusted for differences in size and controls. Near misses may reveal control weakness without being booked losses. When comparing periods, normalize counts by activity where appropriate and distinguish a genuine risk change from a change in reporting behavior.

Worked example: A process has a hypothetical average of four incidents per year, each averaging $50,000. Its expected annual loss is 4 × $50,000 = $200,000. A separate plausible event could cost $8 million, forty times that annual mean, so a tail assessment must consider its likelihood and controls. If small events were excluded from the database, even the four-event frequency estimate could be biased downward.

Watch the distinction: Expected annual loss is an average, not a capital amount that covers rare shocks. Missing events can distort the estimated frequency as well.

Reference: [1]

33. Loss investigation and indicators

A sound loss investigation reconstructs what happened before assigning blame. Establish the event timeline from system logs, confirmations and handoffs; compare it with the intended process and identify the first failed or bypassed control. Test competing explanations, such as a late market feed versus manual entry, against evidence. Separate the trigger from weaknesses that allowed the impact to grow. A key risk indicator should give enough lead time for action: unmatched trades per volume, ageing unresolved breaks or repeated overrides may be more useful than last quarter’s loss total. Define its numerator, denominator, measurement frequency, owner and hypothetical action trigger. Investigate whether rising values reflect more activity, better detection or deteriorating controls. Then choose a control change and check whether the indicator improves without simply suppressing reports. PRM practice questions can reinforce this cause-to-control reasoning.

Worked example: Unmatched trades rise from 2 to 20 per 10,000 trades, a tenfold increase. A hypothetical trigger is 10 per 10,000, so the current rate exceeds it by 10 per 10,000 and is twice the trigger. Compare trade timestamps with feed receipts and reconciliation logs. If delayed messages explain the breaks, repair and retest the interface; hiding unmatched trades from the count would defeat the indicator.

Watch the distinction: A historical loss total may arrive too late to prevent the next event. Choose an indicator linked to a process weakness and an actionable response.

Reference: [1]

34. Operational resilience

Operational resilience concerns continued delivery of an important service during disruption, not simply restoration of one system. Map the people, technology, facilities, vendors and data needed to deliver the service end to end. Define a recovery objective or impact tolerance for the service, then test a severe but plausible disruption against it. A workaround must have enough capacity, trained staff, current data and a route for clearing backlogs. Dependency mapping often exposes a shared provider or authentication service whose failure disables both the primary and backup channels. During an event, track customer outcomes and unfinished transactions as well as server status. After a test or outage, compare actual service performance with the objective, identify the bottleneck and retest the remedy. The programme’s current scope expressly includes operational resilience, but the numerical objective in a scenario must be treated as that scenario’s assumption.

Worked example: A payment service has a hypothetical four-hour recovery objective. Its primary channel is unavailable for six hours, exceeding the objective by two hours. An alternate channel processes only 60% of urgent payments, leaving 40% queued even while it operates. Restoring the main server does not establish that service has recovered: the firm must verify successful payments, clear the queue and check whether any obligations missed their due time.

Watch the distinction: A running backup system is not sufficient evidence of resilience. Judge the service by completed transactions, capacity and time to recover.

Reference: [1]; [2]

35. Compliance controls across the enterprise

A compliance control connects a specific obligation to a responsible owner and evidence that it was met. Translate the obligation into a process step, then identify a preventive control, such as approval before release, and a detective control, such as a review of released items against an approval register. Record who acts, what evidence is retained, how exceptions are escalated and how remediation is verified. Enterprise-wide analysis looks for shared causes across products or units rather than treating every exception as isolated. A failed control means the required evidence or step is missing; deciding whether a law was breached requires the applicable rule and full facts. Sampling gives a useful diagnostic but introduces uncertainty. When failures appear, identify affected records, preserve the original trail, expand testing appropriately and stop or correct the weak process while the facts are assessed.

Worked example: In a hypothetical sample, four of 20 product disclosures lack a recorded approval. The observed sample failure rate is 4/20 = 20%. Identify those four products and check whether approval occurred but was poorly recorded, or never occurred. Preserve the release and approval logs, then test the wider population. The 20% sample result alone proves neither a 20% organization-wide failure rate nor a particular legal violation.

Watch the distinction: Missing approval evidence is a control finding, not automatically a legal conclusion. The obligation, affected records and actual approval history still matter.

Reference: [1]

Credit Risk and Counterparty Credit Risk

36. The credit life cycle

Credit risk begins before money is advanced. Underwriting assesses repayment sources, leverage, cash-flow resilience and collateral; approval sets terms, authority and limits. After origination, monitoring follows payments, covenant compliance, borrower condition and collateral value. Deterioration may prompt tighter monitoring, amendment, restructuring or recovery, each with different effects on timing and ultimate loss. Track a funded balance separately from an undrawn commitment. The funded amount is already owed, while an undrawn line creates potential future exposure if the borrower draws it, possibly when credit quality is worsening. Repayments reduce principal only when applied to principal; interest payments do not. For a facility-level view, distinguish committed limit, amount drawn, accrued amounts and available capacity. A credit decision should consider the borrower's total obligations and linked facilities, not merely the balance of one loan.

Worked example: A borrower has a $1 million term loan and pays $200,000 of scheduled principal. Ignoring other draws and fees, its funded term-loan balance becomes $1 million − $200,000 = $800,000. It also has a separate $300,000 revolver that is wholly undrawn. Current funded principal across these facilities is $800,000, while the revolver adds $300,000 of possible future draw, subject to its terms.

Watch the distinction: An undrawn commitment is not current funded principal, but it can become exposure. Interest received also should not be subtracted from principal.

Reference: [1]

37. Probability, severity and exposure

A simple expected credit loss calculation multiplies probability of default (PD), loss given default (LGD) and exposure at default (EAD) over a stated horizon. PD concerns whether the borrower defaults; LGD is the fraction of EAD lost after recoveries and recovery costs; EAD includes amounts expected to be outstanding when default occurs. Borrower earnings, leverage and sector stress mainly inform PD. Collateral quality, enforceability, seniority and recovery timing inform LGD, though a shared shock can also affect default and collateral value. Drawdown behavior and amortization affect EAD, especially on commitments. Check that inputs describe the same horizon and scenario. For changing future exposures or multiple periods, a single PD × LGD × EAD product may be inadequate; timing, survival and discounting matter. Expected loss is an average across outcomes, while the conditional loss on an actual default may be far larger.

Worked example: A $5 million exposure has a hypothetical one-year PD of 2% and LGD of 40%. Expected loss is $5 million × 0.02 × 0.40 = $40,000. If enforceable collateral is expected to reduce LGD to 25%, with PD and EAD unchanged, expected loss becomes $5 million × 0.02 × 0.25 = $25,000. The $15,000 reduction depends on the collateral assumption holding under default.

Watch the distinction: Do not treat PD as the fraction of principal lost. Default likelihood and loss severity are distinct inputs, and collateral does not necessarily change both.

Reference: [1]

38. Credit evaluation and migration

Credit evaluation methods reveal different aspects of vulnerability. Financial ratios compare debt service and leverage with cash generation, but historical accounts can lag deterioration. A rating-transition model tracks movement among grades and default over a defined period; observed transitions depend on the cohort, period and rating process. A structural model relates default to asset value falling below obligations, making capital structure central to its assumptions. A reduced-form model instead specifies a default arrival process using observable information without requiring an asset-value threshold. Compare outputs only after aligning horizon, definition of default and treatment of changing conditions. A one-year cohort frequency is an observation, not automatically a forecast for another year. Lifetime default probability requires the sequence of conditional survival and default probabilities across future periods. A downgrade can raise concern without being a default. The related FRM study guide offers another route for practising these shared credit concepts.

Worked example: Follow 100 borrowers initially rated BBB for one hypothetical year. Five move to lower nondefault grades and one defaults; assume these are distinct borrowers. The observed downgrade frequency is 5/100 = 5%, and default frequency is 1/100 = 1%. These are one-year cohort outcomes. To estimate lifetime default probability, subsequent-year transitions and survival are needed; multiplying 1% by an arbitrary number of years would ignore changing states and conditions.

Watch the distinction: A downgrade is not a default, and one observed year is not a lifetime forecast. Keep the transition definition and horizon explicit.

Reference: [1]

39. Counterparty exposure

Counterparty credit exposure on a derivative depends on replacement value: if the counterparty defaults now, how much would it cost to replace the positive value owed to the firm? A negative-valued trade is a liability, not a receivable. Values may be offset only where enforceable netting covers the relevant trades and default conditions. Eligible collateral then reduces the positive net amount after applying the agreement’s terms, valuation and any applicable adjustments. Current exposure is a point-in-time amount; potential future exposure concerns how much the replacement cost could grow before closeout. For a limit decision, examine both, along with concentration and stressed exposure. A small current mark does not imply a safe counterparty position if the trades can move sharply. Check whether collateral value could fall as exposure rises, and whether calls and settlement can occur quickly enough.

Worked example: Two trades with one counterparty are worth +$8 million and −$5 million to a bank. Assuming enforceable netting across both, net replacement value is $3 million. With $2 million of eligible collateral, current exposure is max($3 million − $2 million, 0) = $1 million. Without enforceable netting, the negative trade cannot offset the receivable; exposure before collateral would be $8 million.

Watch the distinction: Netting is a legal and contractual condition, not just arithmetic. Offsetting trades without an enforceable agreement can materially understate current exposure.

Reference: [1]

40. Credit portfolios and tranches

Portfolio credit risk depends on how borrowers fail together as well as on individual expected losses. Large single-name or sector positions can dominate outcomes, and a common economic shock can make defaults more dependent precisely when recoveries weaken. Concentration measures, scenario analysis and portfolio loss distributions expose risks that an average PD cannot. Securitization reallocates cash flows and losses through a priority structure. A first-loss tranche absorbs initial pool losses; once exhausted, losses move to the next tranche and eventually to senior investors. Subordination can protect senior claims up to a point, but it does not remove pool credit risk. To analyze a tranche, identify its attachment and detachment points, allocate pool loss in order, and compare the result with each tranche’s size. Also assess servicing, collateral quality and correlation: apparently diversified loans can share a severe common driver.

Worked example: A $100 million loan pool incurs an $8 million realized loss. Its first-loss tranche covers the first $5 million and is fully wiped out: $5 million loss. The next tranche covers the following $10 million and absorbs the remaining $3 million, leaving $7 million of that tranche intact. Senior investors take no loss from this $8 million event, although pool losses above $15 million would reach them under the stated structure.

Watch the distinction: Securitization changes who bears each layer of loss. It does not make the underlying loans safer or guarantee that senior investors avoid larger losses.

Reference: [1]

41. Valuation adjustments

A derivative’s clean value assumes the promised cash flows arrive and can be financed without friction. Valuation adjustments then isolate departures from that assumption. CVA reflects the counterparty’s possible default on amounts owed to you, so it depends on expected positive exposure, default probability and loss given default. DVA reflects your own possible default on amounts you owe; it is not a cash benefit you can freely realize. FVA addresses the cost or benefit of funding the position, while MVA addresses funding initial margin posted to support it. For a simplified unilateral CVA at one future date, multiply expected positive exposure by the counterparty’s default probability and loss given default, then discount if appropriate. A full calculation uses exposure and default timing across dates. Check the valuation framework before adding adjustments so the same funding or credit effect is counted once. The PRM candidate guidebook places these adjustments within counterparty credit risk.

Worked example: Suppose a derivative has $10 million expected positive exposure at one future date, a hypothetical 2% counterparty default probability and 60% loss given default. Ignoring discounting and the timing of exposure within the period, unilateral CVA is $10 million × 0.02 × 0.60 = $120,000. This is an expected valuation deduction, not a certain loss or a complete derivative price. FVA and MVA need separate funding and margin assumptions.

Watch the distinction: Adding CVA, FVA and MVA mechanically can count the same funding effect twice. Identify which cash flows and costs the clean valuation already includes before applying each adjustment.

Reference: [1]

42. Wrong-way risk and central clearing

Wrong-way risk arises when the exposure owed by a counterparty tends to increase as its ability to pay deteriorates. The dependence matters because an expected loss calculated from average exposure and average default probability can miss their coincidence in stress. An oil producer that owes more on a derivative after oil prices collapse may also be less creditworthy after that collapse. Right-way risk runs in the opposite direction. Central clearing replaces bilateral exposure to the original trading partner with exposure managed through a central counterparty, subject to its rules, margin and safeguards. Variation margin reduces accumulated current exposure; initial margin protects against potential losses during closeout. Neither makes the transaction riskless. Margin calls require cash promptly, and concentrated members or weak collateral can still create stress. Evaluate counterparty loss and liquidity demands together, using conditional scenarios when dependence is material.

Worked example: A bank expects a $5 million derivative receivable from an oil producer after a price collapse. At 60% loss given default, expected loss using a 2% default probability is $5 million × 0.02 × 0.60 = $60,000. If conditional default probability rises to 10% in that same scenario, loss rises to $300,000. Clearing may reduce bilateral exposure, but required margin can create an immediate cash need.

Watch the distinction: A lower average default probability does not rule out wrong-way risk. The relevant question is whether default becomes more likely precisely when the receivable becomes large.

Reference: [1]

Market Risk, Asset Liability Management, and Funds Transfer Pricing

43. Trading-book and banking-book rate risk

A trading position is commonly assessed through changes in its marked value as market yields move. For a banking book, management also asks how asset and liability repricing changes future interest income and expense. The two views can diverge: a fixed-rate loan may keep paying the same coupon while its economic value falls when yields rise, and short-term funding may become more expensive. Key-rate sensitivities locate value changes at selected maturities on the yield curve. Express each sensitivity as value change per basis point for a rise at that point, then multiply by the scenario move and add the contributions. The approximation assumes small moves and the stated curve-shock construction; convexity, optionality and changing customer behavior can matter for larger shocks. A parallel-duration number compresses the curve into one movement and can conceal a twist where short and long yields move differently.

Worked example: Suppose a portfolio loses $1,000 per basis point when the two-year yield rises and $3,000 per basis point when the ten-year yield rises. A 20-basis-point two-year rise contributes −$20,000. A 10-basis-point ten-year fall contributes +$30,000. The approximate combined value change is +$10,000. This measures value sensitivity, not the change in next year’s interest income.

Watch the distinction: Applying one parallel-shift duration to a curve twist can give the wrong sign. Reprice or use key-rate sensitivities when maturities move in different directions.

Reference: [1]

44. VaR, expected shortfall and stress

Gross exposure shows position size before offsets; net exposure shows what remains after selected offsets. Neither alone describes a loss distribution, and concentration reveals when apparently diversified positions depend on the same name, sector or factor. Value at risk, or VaR, identifies a loss quantile for a stated horizon and confidence level. Expected shortfall averages losses beyond a specified tail cutoff, so it responds to the severity of the worst observations. Both depend on the distribution, data and modeling choices; a scenario loss instead answers what happens under one specified shock, whether or not that shock appears in recent data. When comparing measures, first align their horizons and position assumptions, then ask whether the decision concerns ordinary fluctuations, extreme severity or forced liquidation. Expected shortfall is especially useful when two books share a VaR figure but have different tail losses.

Worked example: Consider five equally likely hypothetical losses, sorted as $1 million, $2 million, $3 million, $4 million and $10 million. Under the stated nearest-rank convention, 80% VaR is the fourth observation, $4 million. The worst 20% is the single $10 million observation, so 80% expected shortfall is $10 million. A different sample or quantile convention can change the estimate.

Watch the distinction: VaR of $4 million does not cap losses at $4 million. It marks a distributional cutoff; the $10 million observation shows why tail severity needs separate attention.

Reference: [1]

45. Liquidity gaps and contingency funding

Liquidity analysis places expected cash receipts and payments into time buckets, because cash available next month cannot settle a payment due today. Start with contractual maturities, then adjust for plausible behavior such as deposit withdrawals, credit-line drawings, collateral calls and delayed customer payments. For each horizon, net outflows equal outflows minus inflows; a usable liquidity buffer can be compared with that amount. Count assets as usable only to the extent they can be monetized in time after relevant haircuts and operational constraints. A coverage ratio above one is reassuring only for the assumptions and horizon stated. Contingency funding planning then names executable sources, their likely capacity, timing and dependencies. Potential actions include drawing committed facilities, selling or pledging liquid assets, and reducing discretionary cash uses. Stress the plan for correlated failures: an expected inflow and an emergency funding source may both weaken during the same market disruption.

Worked example: Over a stated horizon, a bank expects $30 million of inflows and $50 million of outflows. Net outflow is $20 million. With a $25 million usable liquid buffer, coverage is $25 million ÷ $20 million = 1.25 times, leaving $5 million. If $10 million of inflows fail, net outflow becomes $30 million and the same buffer is $5 million short; a timely alternative is needed.

Watch the distinction: Treating every security’s market value as immediately available cash overstates coverage. Sale time, collateral eligibility and stressed haircuts determine how much the buffer can actually supply.

Reference: [1]

46. Balance-sheet funding and transfer pricing

Funds transfer pricing, or FTP, assigns a funding charge or credit to a business line based on the cash-flow profile it creates for the bank. A long-lived fixed-rate loan should bear a matched-maturity funding and liquidity charge, even if treasury happens to borrow overnight today. That charge makes the lending decision reflect the cost of supporting the asset through its term and discourages a desk from earning an apparent spread by relying on unstable short funding. The loan desk’s contribution equals asset yield minus FTP, expected credit cost and operating costs, before any other specified charges. Treasury separately manages the actual funding mix, interest-rate hedges and resulting consolidated funding profit or loss. FTP is an internal allocation, so adding the desk’s FTP charge to the bank’s external interest expense as if both were independent costs would double count. Assumptions about prepayment and deposit behavior affect matching.

Worked example: A hypothetical loan earns 7% annually. Its matched-maturity FTP charge is 4.5%, expected credit cost is 1% and operating cost is 0.5%, all on the same balance and annual basis. Illustrative contribution is 7% − 4.5% − 1% − 0.5% = 1% of balance. An overnight charge below 4.5% would make the desk appear more profitable while leaving longer-term funding risk elsewhere.

Watch the distinction: FTP is not the bank’s final external funding bill. It attributes funding costs to products; treasury’s actual borrowing costs and hedging results determine the consolidated outcome.

Reference: [1]

47. Commodity hedge basis

A commodity hedge reduces exposure only to the extent its contract matches the quantity, timing, location and grade of the physical need. A futures gain on crude oil can offset an airline’s fuel-cost increase, but fuel and crude prices need not move by the same amount. That difference is basis risk. Contract size also matters: hedging fewer barrels than the expected purchase leaves an explicit unhedged quantity. Calculate the physical cost change as required units times the fuel price change, then calculate the hedge gain or loss using the derivative’s covered units and reference-price move. Net the two using consistent units and signs. A statistical hedge ratio may reduce historical variance, but its estimated relationship can fail when refining margins or delivery conditions change. Check whether the exposure is a purchase or sale before choosing a long or short hedge; a future buyer benefits from a position that gains when the relevant price rises.

Worked example: An airline plans to buy one million barrels of fuel. Fuel rises $5 per barrel, increasing cost by $5 million. Its long crude hedge covers 700,000 barrels and gains $4 per barrel, or $2.8 million. The net cost increase is $5 million − $2.8 million = $2.2 million. Both incomplete quantity coverage and the $1 per barrel difference in price moves contribute.

Watch the distinction: A 70% quantity hedge does not guarantee a 70% offset of the cash loss. The reference commodity’s price move and the physical price move can differ.

Reference: [1]

48. Stress limits and market liquidity

A stress scenario should translate a market move into a position loss and include the cost of exiting or reducing a position when trading depth deteriorates. For a long equity holding, the first component is position value times the assumed price decline. A liquidation allowance can add estimated spread, price impact or fees, provided those costs are not already embedded in the shocked sale price. Compare the total with a stated management trigger, and report both the size of any breach and the assumptions behind it. A trigger is an action point, not a claim that losses cannot exceed it. A sensible response sequence checks the scenario inputs, confirms the exposure, escalates the breach and evaluates risk reduction that is feasible in the stressed market. If selling itself moves prices, a nominally immediate exit may cost more than the original estimate. This links market risk measurement to market liquidity.

Worked example: A $10 million equity position falls 20% in a hypothetical stress, producing a $2 million mark loss. Assume selling costs another 1% of the initial position, or $100,000, and that this cost is excluded from the price shock. Total stress loss is $2.1 million. Against a $1.5 million action trigger, the breach is $600,000; the response should consider available market depth.

Watch the distinction: Stopping at the 20% price loss ignores the cost of exiting in a thin market. Conversely, adding a separate sale cost when the shocked price already includes it double counts.

Reference: [1]

Case Study Practicum

49. Case: map the risks before calculating

In an applied case, classify each event before putting numbers into a single worksheet. A borrower’s failure to pay creates credit loss on the loan. Funding a fixed-rate asset with overnight debt creates interest-rate and funding sensitivity because the debt reprices while the loan coupon stays fixed. A derivative margin call consumes cash immediately; its amount is not automatically a realized trading loss. Delayed receipts create a timing shortfall, and the payment-system outage causing the delay is an operational event. Keep three views: expected or stressed economic loss, annual earnings change, and cash available at the payment deadline. Each has a different horizon and decision use. The PRM Exam 2 study guide is a useful companion for following a case across these risk types. In the next four lessons, use the same Harbor Bank facts and preserve the units and horizons.

Worked example: Harbor Bank holds a €100 million fixed-rate loan earning 6%, funded by €80 million overnight debt at 3% and €20 million equity. Borrower PD is 3% and LGD 40%. Funding rises to 5%, and a derivative requires €7 million margin. Cash is €10 million; €8 million is due today, while a €5 million receipt is delayed by a payment-system outage. Classify default, repricing, margin and outage separately.

Watch the distinction: Calling the €7 million margin call a €7 million trading loss confuses collateral cash movement with valuation. The call creates immediate liquidity pressure even if collateral is later returned.

Reference: [1]

50. Case: expected loan loss

Expected credit loss combines the amount exposed at default, the probability of default over a specified horizon, and the fraction of exposure lost if default occurs. For a simple one-period loan, ECL = EAD × PD × LGD. This is a probability-weighted average across default and survival outcomes, not the cash loss that occurs in the default state. A stressed decision may need both figures: expected loss for pricing or risk comparison, and conditional loss for capital, concentration or contingency analysis. Check that PD and exposure refer to compatible horizons and that LGD accounts for recoveries without assuming instant collection. In Harbor Bank’s case, the full loan balance is assumed exposed and the stated 40% LGD applies if the borrower defaults. No extra collateral or amortization is supplied. The next step is to keep this credit result separate from interest-rate repricing and the day’s cash bridge.

Worked example: Harbor Bank’s €100 million loan has a hypothetical 3% default probability and 40% loss given default over the same stated period. Conditional loss if default occurs is €100 million × 0.40 = €40 million. Expected loss is €40 million × 0.03 = €1.2 million, equivalently €100 million × 0.03 × 0.40. The €1.2 million is an average, not the amount payable upon default.

Watch the distinction: Comparing a €1.2 million expected loss with a stress-loss tolerance can conceal a €40 million default outcome. Match the loss measure to the decision being made.

Reference: [1]

51. Case: funding repricing

A fixed-rate asset pays a coupon that does not change with the current funding rate over its fixed period. Overnight debt generally reprices much sooner. When funding rates rise, the bank pays more on the liability while receiving the same loan coupon, reducing net interest income. Measure this earnings effect using the repricing liability amount multiplied by the rate change and the length of time it persists. State the horizon: an annualized calculation assumes the higher rate and balances remain in place for a full year. This is distinct from the loan’s credit loss and from any change in the asset’s economic value. A bank could reduce the mismatch by terming out funding, changing product pricing where feasible, or using an appropriate interest-rate hedge, each with costs. Before recommending one, compare its cash timing, basis and effect on the balance sheet.

Worked example: Harbor Bank’s €80 million overnight debt rises from 3% to 5%, while its €100 million fixed-rate loan continues earning 6%. Annual funding expense rises by €80 million × 0.02 = €1.6 million if rates and balances persist for a year. Loan interest remains €6 million annually. Ignoring other items, the loan less debt interest spread falls from €3.6 million to €2 million.

Watch the distinction: Applying the 2-percentage-point rate increase to the €100 million loan misstates the effect. The loan coupon is fixed; the €80 million repricing debt drives this earnings change.

Reference: [1]

52. Case: margin and delayed cash

A cash bridge begins with money available at the relevant deadline, adds receipts expected by that deadline and subtracts payments due by then. Timing is decisive. A receipt delayed until tomorrow cannot fund today’s obligations, even if the bank remains solvent on a longer horizon. Derivative margin is a cash use in this bridge because collateral must be delivered when called; do not infer that its full amount is an economic loss. Build an on-time baseline, then remove the delayed inflow to isolate the outage’s impact. If the balance turns negative, identify a source that can actually deliver cash before the payments fall due. A funding line, eligible collateral or operational payment workaround may help, but its availability must be tested. This shortfall may coexist with the annual earnings decline and the borrower’s credit risk without being summed into either measure.

Worked example: Harbor Bank starts today with €10 million cash. It owes €7 million derivative margin and €8 million scheduled payments, and expects €5 million from a customer. If the receipt arrives, closing cash is €10 million − €7 million − €8 million + €5 million = €0. A payment-system outage delays it, leaving −€5 million today. The bank needs €5 million of timely cash or a valid payment contingency.

Watch the distinction: Netting tomorrow’s expected €5 million against today’s payments hides the deadline shortfall. Liquidity is measured by when cash can be used, not merely by eventual receipts.

Reference: [1]

53. Case: choose the management response

A risk response should prioritize deadlines while keeping the measures conceptually separate. First, secure or release enough cash to meet today’s €5 million gap; a margin call and delayed receipt explain that liquidity need. Next, escalate the severe conditional credit loss against the stated stress tolerance and review borrower concentration, collateral and possible risk transfer. Then address the continuing fixed-asset versus overnight-funding mismatch, since the higher funding rate erodes annual earnings while it persists. A combined scenario may place the €40 million conditional default loss alongside one year’s €1.6 million incremental funding expense, but only if the scenario assumes default and the higher rate over compatible periods. The €7 million margin transfer does not belong in that economic-loss sum. Management should also investigate the payment-system failure and test whether its workaround can deliver funds. Decisions should be recorded with assumptions, owners and deadlines.

Worked example: Assume Harbor Bank’s borrower defaults with 40% LGD on €100 million and elevated funding costs persist for a year. Conditional credit loss is €40 million; incremental annual funding expense is €1.6 million, giving €41.6 million across those stated components. This exceeds a hypothetical €6 million stress-loss tolerance by €35.6 million. Separately, today’s delayed receipt leaves a €5 million cash gap; €7 million margin is already in that bridge.

Watch the distinction: Adding the €7 million margin call to €41.6 million treats collateral posted as a permanent loss. Keep the cash bridge and economic-loss scenario distinct unless a separate loss is demonstrated.

Reference: [1]

54. Case pathology and root cause

A case pathology reconstructs what happened in time: the initial exposure, concealment or control failure, trigger, detection and eventual loss. A market move can reveal a problem without being its root cause. Ask whether authorized limits, independent trade capture, position reconciliation and escalation would have detected the exposure before the move. Then distinguish a control designed to prevent an unauthorized position from one designed to discover it after booking. Historical risk failures can suggest patterns such as hidden positions or weak independent checks, but their names do not establish facts in a new scenario. Use only the evidence provided for the case at hand. Quantify what is known, identify uncertain amounts and avoid claiming that the eventual loss was entirely caused by one event if losses had already accumulated. A good remedy closes the pathway that allowed risk to build, not just the account or market position where it surfaced.

Worked example: In a hypothetical case, a trader hides a $4 million loss in an unreviewed account. A later market fall reveals a $15 million total loss. At least $4 million existed before the fall; the additional reported loss is $11 million if the figures use the same valuation basis. The fall was the trigger for discovery, while unauthorized positions and failed reconciliation enabled accumulation. Independent position checks address that pathway.

Watch the distinction: Blaming only the market fall overlooks the concealed $4 million and the control failures. A sound root-cause account separates the trigger from the conditions that let exposure grow.

Reference: [1]

PRMIA Standards and Governance

55. Group of Thirty derivatives practice

Sound derivatives practice requires a desk’s reported value to be challenged by a function with access to independent prices, model review and trade records. Start by confirming contract terms and legal enforceability, including netting and collateral arrangements. Then compare the desk mark with independently sourced inputs, examine model assumptions and explain material differences before relying on the figure for profit, limits or counterparty exposure. Exposure measurement should consider current replacement value and how the value could change before closeout. Operational controls should keep trading, confirmation, settlement and valuation responsibilities appropriately separated. Legal review matters because an assumed netting or collateral benefit may fail if documentation is defective or unenforceable. These principles are associated with the Group of Thirty derivatives recommendations covered in the PRM scope; they are practical control tests rather than a claim that every discrepancy proves misconduct. Record unresolved valuation uncertainty and escalate it in proportion to its effect.

Worked example: A derivatives desk reports a $2 million gain, while independent valuation gives $1.7 million. The difference is $300,000, or 15% of the desk mark. The reviewer checks trade terms, market inputs and model choices, then determines the supportable value before the gain enters reported profit or risk limits. The discrepancy alone does not identify which mark is correct.

Watch the distinction: Accepting the trading desk’s mark because the trade was profitable bypasses independent challenge. A valuation difference must be explained before it is used in decisions or reports.

Reference: [1]

56. Governance principles in action

Governance turns risk information into accountable decisions. A business can take exposure within delegated authority, while independent risk provides challenge and governing bodies set direction and oversee exceptions. Reports must let those bodies see where limits were overridden, who approved each override, why it was justified and whether the exposure remains open. An aggregate number can hide repeated exceptions even when the total appears stable. Good information flow therefore includes trends, concentrations and unresolved actions as well as headline measures. If business incentives reward volume, the process needs independent escalation so a manager cannot silently redefine a breach as ordinary activity. The PRM designation covers practical risk management; the guidebook specifically includes PRMIA governance principles. Applying them requires a clear decision owner, an informed reviewer and evidence that challenge affected the decision, rather than a committee name alone.

Worked example: A risk committee sees total exposure but no detail on five limit overrides. The total cannot show whether one exception is large or whether approvals are becoming routine. A useful report lists all five with size, date, approver, rationale and current status. The committee can then challenge recurring causes, confirm delegated authority and request a reduction or formal limit review.

Watch the distinction: A committee’s receipt of an aggregate exposure report does not prove effective oversight. It needs exception details and decision rights to identify and address repeated breaches.

Reference: [1]

57. Bylaws and institutional authority

An association’s bylaws establish its governance architecture and the authority through which member rights and institutional decisions are handled. A desk procedure or internal memo may guide staff work, but it cannot by itself create powers that the bylaws reserve elsewhere. When a proposed action affects members, identify the right or decision at issue, find the applicable governing provision, determine the authorized decision-maker and follow the specified approval and communication process. This sequence avoids treating convenience or seniority as a substitute for authority. The guidebook identifies PRMIA bylaws as protecting member rights and assets, but the supplied excerpt does not set out particular amendment votes or notice periods. Therefore, do not infer a numerical voting threshold or a named approval body from this lesson. The transferable reasoning is the hierarchy of authority: governing rules control, while operating documents implement decisions made under those rules.

Worked example: An association officer wants to alter a member right through an internal desk memo. The memo can propose wording but cannot establish authority to change the right. The officer first identifies the relevant bylaw and authorized decision-maker, then obtains the approval the governing rule requires and communicates the effective decision. No vote count can be inferred from the facts given.

Watch the distinction: Assuming an officer’s title makes a memo binding confuses administrative responsibility with governing authority. A member-right change must follow the applicable bylaw process.

Reference: [1]

58. Conflicts of interest

A conflict of interest exists when a financial, personal or organizational interest could compromise professional judgment or reasonably appear to do so. The question is not only whether the final analysis is wrong; it is whether the decision process remains credible and independently testable. Identify the interest, disclose it to the appropriate decision-maker, and remove or control the conflicted person’s influence. For a material direct payment from an issuer, an independent analyst can perform the credit assessment; recusal may be needed from rating decisions and related approval discussions. If the person remains involved, document why, define limits on that involvement and retain review evidence. A disclosure alone is insufficient when the conflicted person still controls key assumptions or suppresses adverse findings. Equally, an unfavorable rating does not automatically prove impartiality. Professional conduct is assessed through transparent handling of the conflict and reliable analysis, rather than the direction of the conclusion.

Worked example: An analyst receives $20,000 from an issuer whose credit quality the analyst is assigned to assess. The analyst discloses the payment and steps away from the assessment; an independent analyst reviews the issuer’s cash flows, debt and repayment capacity. A favorable rating may still be justified, but the separate review makes its basis testable. Keeping the original analyst as sole approver would leave the conflict active.

Watch the distinction: Disclosing a payment does not neutralize its influence if the recipient still controls the rating. Disclosure should lead to an independent decision or effective recusal.

Reference: [1]

59. Confidentiality and responsible disclosure

Risk managers need to communicate material findings accurately while limiting access to client information to those with a legitimate need. Removing names is only one step. A report may still reveal a client through an unusually large balance, a distinctive transaction, a small group or information already known to recipients. Before sharing an aggregate, ask whether a reader could subtract known amounts or match the figure with public facts to identify an individual client. If so, use broader grouping, suppress sensitive breakdowns, restrict the audience or change the level of detail while preserving the risk message. The choice should also protect decision quality: hiding a concentration entirely can prevent appropriate oversight. A restricted report to authorized decision-makers may provide the precise exposure, while a wider report uses a less identifying summary. Confidentiality and accurate escalation can coexist when information is routed to the right audience at the right granularity.

Worked example: A report totals three client loans, but one client represents 90% of the balance. Someone who knows the other two amounts could infer the dominant client’s loan by subtraction. The risk manager provides the precise concentration to authorized reviewers and uses a broader aggregate for wider circulation. Merely deleting the client’s name from the three-loan total would not remove the identification risk.

Watch the distinction: An anonymous total is not necessarily confidential. Small groups and dominant positions can let readers reconstruct a client’s exposure from information they already hold.

Reference: [1]

60. Independent challenge and escalation

Independent challenge protects risk reporting when a business head has an incentive to minimize a breach. Begin with the reported measure, its limit and the calculation inputs. If an adjustment is proposed, require a documented analytical reason and test it through a reviewer who is not dependent on the position’s owner. While the dispute is examined, report the breach shown by the current approved method and distinguish any alternative estimate clearly. Escalate through the organization’s established governance route so an authorized decision-maker can decide whether to reduce exposure, approve a justified exception or change the model after validation. A verbal request cannot alter the underlying risk or retroactively erase a limit crossing. The reviewer should preserve both the original result and the evidence supporting any correction. This sequence supports accurate information flow and fair treatment of the trading desk without allowing unsupported overrides to become routine.

Worked example: A model reports an $8 million stress loss against a hypothetical $5 million limit, so the breach is $3 million. A trading head requests a $3 million reduction without evidence; that would merely make the reported loss $5 million. Record the $8 million result and breach, validate the disputed input independently, and escalate. If later evidence supports a revised result, document the calculation and approval trail.

Watch the distinction: Changing a model result to equal the limit does not reduce exposure. Until a supported correction is validated, the original $3 million breach remains the reportable finding.

Reference: [1]

Official references and further reading

All study guides

Choose your exam version

PRM Exam 1

PRM Exam 2

Finance Theory

1. Utility and the price of uncertainty

2. Risk-adjusted performance

3. Diversification, time and hedging

4. Efficient portfolios and the risk-free asset

5. Single-factor and multifactor pricing

6. Debt, equity and firm value

7. Yield curves and rate models

Financial Instruments

8. Bond cash flows and embedded options

9. Forwards, futures and carry

10. Swap cash flows and value

11. Option payoffs and pricing

12. Credit protection and securitization

13. Caps, floors and swaptions

14. Virtual-asset exposures

Financial Markets

15. Market participants and trade workflow

16. Bond-market spreads and seniority

17. Money-market yields and reference rates

18. Stock-market liquidity

19. FX quotation and cross rates

20. Energy and commodity markets

21. Futures exchanges and clearing

Mathematical Foundations of Risk Measurement

22. Time value and algebra

23. Sensitivity through calculus

24. Summary statistics and tail behavior

25. Conditional probability and dependence

26. Numerical and state-transition methods

27. Matrices, correlated shocks and PCA

28. Regression, inference and model fit

Enterprise Risk Management Frameworks and Operational Risk Management Practices

29. Who owns and challenges risk

30. Appetite, tolerance and capacity

31. Operational-risk assessment

32. Operational loss data and capital

33. Loss investigation and indicators

34. Operational resilience

35. Compliance controls across the enterprise

Credit Risk and Counterparty Credit Risk

36. The credit life cycle

37. Probability, severity and exposure

38. Credit evaluation and migration

39. Counterparty exposure

40. Credit portfolios and tranches

41. Valuation adjustments

42. Wrong-way risk and central clearing

Market Risk, Asset Liability Management, and Funds Transfer Pricing

43. Trading-book and banking-book rate risk

44. VaR, expected shortfall and stress

45. Liquidity gaps and contingency funding

46. Balance-sheet funding and transfer pricing

47. Commodity hedge basis

48. Stress limits and market liquidity

Case Study Practicum

49. Case: map the risks before calculating

50. Case: expected loan loss

51. Case: funding repricing

52. Case: margin and delayed cash

53. Case: choose the management response

54. Case pathology and root cause

PRMIA Standards and Governance

55. Group of Thirty derivatives practice

56. Governance principles in action

57. Bylaws and institutional authority

58. Conflicts of interest

59. Confidentiality and responsible disclosure

60. Independent challenge and escalation

Official references and further reading

Frequently Asked Questions

Related Study Guides

Financial Risk Manager (FRM)

Associate in Risk Management (ARM)

Certified Enterprise Risk Analyst (CERA)