How Banks Calculate Derivatives XVA: CVA, DVA, FVA, MVA, Exposure, Credit, Funding, Margin and Capital Valuation Adjustments

Quick answer: a derivative’s textbook or “clean” value is not always the amount a bank regards as its full economic value once counterparty credit, the bank’s own credit, funding, collateral, initial margin and capital are considered. XVA is a family name for valuation adjustments that add those effects around the clean derivative price. CVA reflects counterparty credit risk; DVA reflects the bank’s own non-performance risk in fair-value frameworks; FVA represents funding effects under a bank’s chosen valuation policy; MVA represents the cost of funding initial margin; and KVA is an internal economic concept used by some institutions to represent the cost of capital. These terms are related but are not interchangeable, and not every XVA has the same accounting or regulatory status.

A derivative can have the same contractual cash flows and a different economic value because the parties, collateral, funding route and capital burden are different.

Page role: How Banks Calculate Counterparty Credit Risk owns exposure profiles, netting sets, collateral and default exposure. This article begins with those exposures and asks a different question: how are credit, funding, margin and capital effects translated into valuation adjustments around a derivative’s clean price?

1. Start with the clean price

Suppose a collateralised interest-rate swap has a clean model value V_clean. In a simplified framework, the bank may represent an economic value as:

V_economic = V_clean − CVA + DVA − FVA − MVA − KVA + other governed adjustments.

This sign convention is illustrative, not universal. Institutions use different definitions and booking conventions, especially for FVA and KVA. The important point is architectural: the clean market-risk price and the non-clean adjustments should be identifiable separately enough that the bank can explain why value changed.

2. CVA turns counterparty default risk into today’s price

Credit Valuation Adjustment (CVA) is the fair-value adjustment for the possibility that the counterparty defaults when the derivative has positive value to the bank. Federal Reserve supervisory guidance describes CVA as the market value of counterparty credit risk and emphasises that the calculation should reflect current market information where possible.

See Interagency Supervisory Guidance on Counterparty Credit Risk Management.

A teaching version of unilateral CVA is:

CVA ≈ (1 − Recovery) × Σ DiscountFactor(t) × EE(t) × DefaultProbabilityIncrement(t).

EE(t), expected exposure, is the average positive derivative exposure at future time t under the model. The default-probability increment measures the chance that default occurs in that interval, conditional on survival to that point.

3. Why expected exposure is not the current mark-to-market

A swap worth S$2 million to the bank today may be worth S$10 million, zero or negative in two years depending on rates. CVA therefore needs a future exposure distribution rather than today’s value repeated through time.

For each simulation path ω:

Exposure(t,ω) = max[V(t,ω) − collateral(t,ω), 0].

Then:

EE(t) = average_ω[Exposure(t,ω)].

This means CVA inherits market-risk modelling assumptions from the derivative itself, plus collateral assumptions and credit assumptions. A CVA engine is therefore a model-of-models.

4. Netting can reduce CVA because positive and negative values offset

Suppose two derivatives with the same counterparty have future values of +S$8m and −S$6m at a particular state. Without enforceable netting, gross positive exposure can be S$8m. With legally enforceable close-out netting, the net exposure can be only S$2m before collateral.

This is why CVA is normally calculated at a legally defined netting-set level rather than transaction by transaction and then blindly summed.

The legal assumption is part of the mathematics: if netting is not enforceable in a jurisdiction or agreement, the lower exposure is not economically available.

5. Collateral reduces exposure but introduces timing and margin mechanics

Variation margin can reduce current counterparty exposure by transferring collateral as mark-to-market changes. But the reduction depends on threshold, minimum transfer amount, margin frequency, settlement delay and dispute behaviour.

A perfectly collateralised mathematical model can therefore understate real exposure if it assumes zero gap between valuation and collateral receipt.

For the underlying collateral mechanics, see How Central Counterparties Calculate Margin and How Banks Optimise Collateral.

6. Wrong-way risk makes CVA larger exactly when exposure and credit quality deteriorate together

Many simplified CVA calculations assume future exposure and counterparty default are independent. That can fail badly.

Suppose a bank has a derivative that becomes strongly positive when oil prices fall, and the counterparty is an oil producer whose credit quality also deteriorates when oil prices fall. The same scenario creates:

• higher derivative exposure;
• higher default probability.

This is wrong-way risk. The correct joint expectation is:

E[Exposure × DefaultIndicator]

not necessarily:

E[Exposure] × E[DefaultIndicator].

7. DVA reflects the bank’s own non-performance risk

Debit Valuation Adjustment (DVA) is the valuation effect of the bank’s own credit risk on derivative liabilities in fair-value frameworks. If the bank itself becomes less creditworthy, the fair value of some liabilities can fall because there is a greater chance the bank will not pay them in full.

This creates an uncomfortable result: a deterioration in the bank’s own credit quality can create an accounting valuation gain. That does not mean becoming less creditworthy creates economic wealth. Prudential capital rules therefore treat own-credit effects carefully; the Basel Committee has stated that banks must derecognise certain own-credit valuation effects from CET1 rather than allow them to strengthen capital.

See Basel Committee FAQ on own-credit and funding valuation adjustments.

8. CVA and DVA are bilateral mirror images only in a simplified world

It is tempting to say CVA equals the counterparty’s DVA and vice versa. In practice, the two parties can use different curves, funding assumptions, collateral models, recovery assumptions and legal interpretations. Accounting and regulatory treatment also differ.

Therefore:

economic symmetry does not guarantee operational valuation symmetry.

9. FVA asks how funding economics alter derivative value

Funding Valuation Adjustment (FVA) is used by many institutions to represent the effect of funding costs or benefits associated with derivative positions after collateral and other funding arrangements are considered.

A simplified funding-cost representation is:

FVA ≈ Σ DiscountFactor(t) × ExpectedFundingRequirement(t) × FundingSpread(t) × Δt.

But FVA is not a single universally standardised object. Whether, how and with what sign an institution recognises funding effects depends on valuation policy, accounting interpretation, collateral terms and internal methodology.

This is a key boundary: FVA is a market-practice valuation concept, not simply a Basel capital charge.

10. MVA prices the funding cost of initial margin

Initial margin is collateral posted to protect against potential future exposure during the close-out period after a default. Unlike variation margin, initial margin is generally segregated or restricted and cannot simply be reused as ordinary funding by the posting party.

If the bank expects to post IM(t) over the life of the trade, a stylised Margin Valuation Adjustment is:

MVA ≈ Σ DiscountFactor(t) × E[IM(t)] × margin funding spread(t) × Δt.

For uncleared derivatives, industry participants often use ISDA SIMM or approved schedule approaches to calculate regulatory initial margin. ISDA’s SIMM framework is recalibrated and backtested under industry governance. See ISDA SIMM.

11. MVA depends on market volatility even if the contractual derivative cash flows do not change

If volatility rises, initial margin requirements can increase because potential future movement over the margin period becomes larger. That increases the funding amount even before any default occurs.

This creates a nonlinear connection:

higher volatility → higher IM → higher funding requirement → larger MVA.

A derivative therefore acquires an economic sensitivity to the margin model itself.

12. KVA is an internal capital-cost concept, not a single accounting rule

Capital Valuation Adjustment (KVA) is used by some banks and practitioners to represent the economic cost of holding capital against a derivative over time. A stylised form is:

KVA ≈ Σ DiscountFactor(t) × Economic/RegulatoryCapital(t) × required capital return × Δt.

But KVA is not a universally required IFRS or US GAAP line item and definitions vary materially. It is best treated as an internal economic-pricing or management concept unless a particular institution’s accounting policy says otherwise.

This connects to How Banks Allocate Economic Capital.

13. XVA is not additive without care

A naive implementation calculates CVA, FVA, MVA and KVA independently from the same clean exposure and adds them. But the adjustments can change one another’s inputs.

Examples:

• collateral reduces CVA but creates funding/margin requirements;
• initial margin reduces counterparty exposure but increases MVA;
• funding policy changes discounting and cash requirements;
• capital can depend on counterparty exposure and CVA risk;
• hedging CVA creates market-risk and funding positions.

Some XVA systems therefore solve coupled equations or iterate until adjustments and exposures are internally consistent rather than treating every component as independent.

14. The CVA capital charge is not the same as accounting CVA

Accounting CVA adjusts fair value for counterparty credit risk. Basel’s CVA risk capital framework instead asks how much regulatory capital is needed for changes in CVA caused by movements in counterparty credit spreads and market risk factors.

The European Banking Authority summarises CVA risk as the risk of losses from changing CVA values in response to counterparty credit spreads and market factors. See EBA — Market, counterparty and CVA risk.

Current Basel rules contain Basic Approach CVA (BA-CVA) and Standardised Approach CVA (SA-CVA) structures, plus an alternative treatment for certain banks below a materiality threshold subject to supervisory conditions. See the current Basel Framework.

15. CVA hedging can create a second-order optimisation problem

If CVA rises when a counterparty’s credit spread widens, the bank can use eligible credit or market hedges to reduce sensitivity. But hedges have their own:

• basis risk;
• liquidity risk;
• counterparty risk;
• funding cost;
• market-risk capital;
• hedge-accounting or valuation effects.

The objective is therefore not “make CVA zero.” It is to reduce total risk-adjusted cost subject to the fact that each hedge changes several XVA and capital components.

16. Proxy credit spreads are a hidden model-risk source

Many derivative counterparties do not have a liquid traded CDS curve. The bank then needs a proxy based on rating, sector, region or comparable issuers.

Federal Reserve guidance warns against over-reliance on broad non-market estimates and says proxy spreads should reasonably capture counterparty-specific and liquidity characteristics.

A proxy can be mathematically smooth and economically wrong. Validation should therefore ask whether proxy CVA moves with the actual credit deterioration of the counterparty rather than simply with a generic rating bucket.

17. A miniature CVA calculation

Suppose a one-year netting set has expected exposure of S$10m over the year, a one-year default probability of 2%, recovery of 40%, and ignore discounting for illustration.

CVA ≈ 10m × 2% × (1 − 40%) = S$120,000.

Now suppose collateral reduces expected exposure to S$3m:

CVA ≈ 3m × 2% × 60% = S$36,000.

The CVA improvement is S$84,000 in this toy example. But if the collateral arrangement creates substantial initial-margin funding costs, the total XVA improvement can be smaller. This is why component optimisation can mislead.

18. Scenario decomposition makes XVA explainable

A useful XVA explain can attribute a daily change to:

• market moves affecting exposure;
• counterparty spread moves;
• own-credit spread moves;
• funding-curve changes;
• initial-margin changes;
• new trades and unwinds;
• collateral movement;
• model or data changes.

If the desk cannot explain why CVA moved, it cannot distinguish real risk from a broken feed or model recalibration.

19. Creative-work lens: the “true cost” of a journey

A train ticket price is not the same thing as the full cost of a journey if the traveller must also pay for transfers, insurance, luggage, time and contingency. The contractual derivative price is similar: the clean value prices the market cash flows, while XVA attempts to price additional institutional costs created by who the counterparties are and how the trade is collateralised, funded and capitalised.

The analogy has limits. Unlike a travel budget, XVA components can interact mathematically and may have different accounting, regulatory and internal-management status.

20. The XVA algorithmic pipeline

1. Value the clean derivative under the approved market model.
2. Group trades into legally valid netting sets.
3. Simulate future market states and derivative values.
4. Apply variation margin, thresholds, timing and collateral terms.
5. Calculate expected positive and negative exposure profiles.
6. Build counterparty and own-credit default/spread curves.
7. Calculate CVA and DVA under the governed framework.
8. Project funding requirements and calculate FVA where policy requires.
9. Project initial margin and calculate MVA where applicable.
10. Estimate capital cost/KVA where used internally.
11. Model wrong-way risk and proxy-spread uncertainty.
12. Calculate sensitivities and hedging effects.
13. Reconcile XVA changes to market, credit, funding, margin and trade events.
14. Backtest and independently validate the full stack.

21. Failure modes

• Today’s-MTM exposure. Current derivative value is repeated into the future instead of simulating exposure.
• Netting fantasy. Offsets are assumed despite weak legal enforceability.
• Perfect-collateral assumption. Margin timing and disputes disappear.
• Independence assumption. Exposure and default probability are modelled independently despite wrong-way risk.
• XVA double count. Funding, margin or capital effects are charged twice under overlapping definitions.
• Proxy-spread complacency. A smooth sector curve substitutes for real counterparty credit information.
• Accounting/regulatory confusion. Internal KVA or FVA policy is presented as though it were a universal accounting rule.
• Static IM assumption. Initial margin is frozen despite changing volatility and portfolio composition.

22. Diagnostics and falsifiers

• How much CVA is reduced by netting versus collateral?
• Which counterparties contribute the largest CVA after exposure size is controlled?
• What happens to CVA if recovery falls or spreads widen?
• Does wrong-way risk materially change the independent-exposure result?
• How much MVA rises under a volatility shock?
• Which XVA component changes when the funding curve moves but market curves do not?
• Can every valuation adjustment be mapped to a distinct economic mechanism?
• What observation would show that the proxy credit curve is misrepresenting the counterparty?

Suppose someone claims, “The derivative is fully collateralised, so CVA is zero.” A falsifier is any realistic margin-period, threshold, dispute or wrong-way-risk scenario that leaves positive exposure at default. Collateral can reduce CVA dramatically without making counterparty risk mathematically impossible.

23. Verification and update triggers

• reconcile clean values to independent pricing benchmarks;
• compare simulated exposure with realised collateralised exposures;
• validate netting and collateral legal assumptions;
• backtest proxy spreads against counterparty market deterioration;
• stress wrong-way risk explicitly;
• recalculate MVA after margin-model recalibrations;
• separate model changes from market P&L in daily XVA explain;
• update capital and regulatory treatment when Basel or local implementation rules change.

Connections across the finance-and-banking algorithms lane

• Counterparty credit risk — provides the exposure profiles that feed CVA.
• Yield-curve algorithms — price and discount the underlying derivative cash flows.
• CCP margin — connects volatility and exposure to margin requirements.
• Economic-capital allocation — the internal capital-cost layer behind KVA-like concepts.

Research anchors

• Federal Reserve — Counterparty Credit Risk Management.
• European Banking Authority — Market, Counterparty and CVA Risk.
• Basel Framework — current CVA risk and counterparty-credit standards.
• Basel Committee — own-credit and FVA capital treatment FAQ.
• ISDA — SIMM initial margin methodology.

The deeper lesson

XVA is the mathematics of putting the institution back into the price. Clean derivative valuation asks what contractual cash flows are worth under market curves and volatility. CVA adds the counterparty. DVA adds the bank itself. FVA adds funding. MVA adds margin. KVA-like frameworks add capital economics. The strongest XVA system therefore does not hide complexity behind one number; it makes each adjustment traceable to the mechanism that created it and keeps accounting, prudential and internal-economic meanings separate.

Educational note: This article explains public derivatives and banking mathematics. It is not trading advice, derivative-pricing advice for any transaction, accounting advice or regulatory advice.