With commodity risk complete, this article covers the sixth and final individual delta risk class: foreign exchange risk. The bucket, risk weight and correlation structure here is genuinely the simplest of any risk class in this series — but the risk factor definitions turn out to be richer than that simplicity might suggest, thanks to a genuinely important optionality the standard builds in: the choice between a reporting currency and a base currency approach.
Delta FX Risk Factors: Reporting Currency as the Default
By default, FX delta risk factors are all the exchange rates between the currency in which an instrument is denominated and the bank’s reporting currency. For transactions that reference an exchange rate between a pair of currencies, neither of which is the reporting currency, the risk factors instead become the exchange rates between the reporting currency and both of the other currencies the instrument references.
| The Source Document’s Own Worked Example For an FX forward referencing USD/JPY, the relevant risk factors for a CAD-reporting bank to consider are the exchange rates USD/CAD and JPY/CAD — not USD/JPY directly, since neither USD nor JPY is the bank’s reporting currency. The bank picks up two separate delta risk factors, each measured against its own reporting currency, rather than one direct cross-rate factor. |
Sourced from MAR21.14(1)(a), including footnote [7].
The Alternative: Calculating FX Risk Against a Base Currency
Subject to supervisory approval, a bank may instead calculate FX risk relative to a base currency rather than its reporting currency. Choosing this route is not simply a substitution — a bank taking this approach must account for two things at once: the FX risk against the base currency itself, and separately, the FX risk between the reporting currency and the base currency, referred to as translation risk. The resulting FX risk calculated relative to the base currency is then converted into capital requirements in the reporting currency, using the spot reporting/base exchange rate that reflects the FX risk between the two.
This base currency approach is only permitted under specific conditions:
- A bank may only consider a single currency as its base currency — it cannot mix and match base currencies across different parts of its portfolio.
- The bank must demonstrate to its supervisor that calculating FX risk relative to the proposed base currency gives an appropriate risk representation for its portfolio — for example, by showing this approach does not inappropriately reduce capital requirements compared to what would be calculated without it — and that translation risk between the base currency and the reporting currency is properly accounted for.
| Continuing the CAD/USD/JPY Example The source document extends its own worked example to illustrate this alternative directly: if that same CAD-reporting bank instead calculates FX risk relative to a USD base currency, it would consider separate deltas for the JPY/USD exchange rate risk and for CAD/USD FX translation risk, and then translate the resulting capital requirement back into CAD using the USD/CAD spot exchange rate. Notice the structural difference from the reporting-currency default: instead of two CAD-anchored risk factors (USD/CAD and JPY/CAD), the bank now has one USD-anchored risk factor (JPY/USD) plus one translation risk factor (CAD/USD) — genuinely different risk factors, not just a relabelling of the same ones. |
Sourced from MAR21.14(1)(b)–(d).
https://www.bis.org/basel_framework/chapter/MAR/21.htm?inforce=20230101&published=20260323
Vega FX Risk Factors
FX vega risk factors are the implied volatilities of options that reference exchange rates between currency pairs, defined along a single dimension: the maturity of the option, mapped to one or several of five prescribed tenors — 0.5, 1, 3, 5 and 10 years.
Sourced from MAR21.14(2).
Curvature FX Risk Factors
Curvature FX risk factors mirror the delta structure exactly: by default, all the exchange rates between the currency in which an instrument is denominated and the reporting currency, with the same reporting-currency-vs-both-other-currencies rule for pairs of non-reporting currencies. Where supervisory approval for the base currency approach has already been granted for delta risk, FX curvature risk must also be calculated relative to that same base currency, and then converted back to the reporting currency using the spot reporting/base exchange rate — the same mechanism as delta, applied consistently to curvature.
Sourced from MAR21.14(3).
No Onshore/Offshore Distinction
One final rule applies uniformly across delta, vega and curvature: no distinction is required between onshore and offshore variants of a currency for any FX risk factor.
| A Deliberate Contrast With GIRR This is worth pausing on because it directly contradicts what we established for GIRR back in Article 9A. There, an onshore currency curve and its offshore equivalent — onshore Indian rupee versus offshore Indian rupee, for example — must be treated as two distinct curves for interest rate risk purposes. FX risk takes the opposite position: no such distinction is required. The standard evidently judges that the onshore/offshore distinction matters for how a currency’s own yield curve behaves, but not for how its spot exchange rate itself is captured as an FX risk factor. |
Sourced from MAR21.14(4).
The Sensitivity Formula: A Quick Recap
The delta FX sensitivity formula was covered in full in Article 8C: shift the exchange rate by 1 percentage point, and divide the resulting change in market value by 0.01.
Sourced from MAR21.24.
The Simplest Bucket Structure in the Standard
For delta FX risk, an FX risk bucket is set for each exchange rate between the currency in which an instrument is denominated and the reporting currency — one bucket per currency pair, directly mirroring the risk factor definition itself. There is no further sub-division by tenor, sector, or credit quality of any kind.
Sourced from MAR21.86.
One Flat Risk Weight for Everything
A single, uniform relative risk weight of 15% applies to all FX sensitivities, across every currency pair, with no variation by bucket.
Sourced from MAR21.87.
https://www.bis.org/basel_framework/chapter/MAR/21.htm?inforce=20230101&published=20260323
The Square-Root-of-2 Discount for Specified Currency Pairs
For a defined list of specified currency pairs, and for currency pairs forming first-order crosses across those specified pairs, the 15% risk weight may — at the bank’s discretion — be divided by the square root of 2.
| The 19 Specified Currency Pairs |
| USD/EUR |
| USD/JPY |
| USD/GBP |
| USD/AUD |
| USD/CAD |
| USD/CHF |
| USD/MXN |
| USD/CNY |
| USD/NZD |
| USD/RUB |
| USD/HKD |
| USD/SGD |
| USD/TRY |
| USD/KRW |
| USD/SEK |
| USD/ZAR |
| USD/INR |
| USD/NOK |
| USD/BRL |
| What Counts as a “First-Order Cross” The standard’s own example: EUR/AUD is not itself among the 19 specified currency pairs above, but it is a first-order cross of USD/EUR and USD/AUD — both of which are specified pairs. Because of that relationship, EUR/AUD also qualifies for the square-root-of-2 discount, even though it doesn’t appear directly on the list. |
Sourced from MAR21.88, including footnotes [22] and [23].
Correlation Across Buckets
For aggregating delta FX risk positions across buckets, the cross-bucket correlation parameter γbc is uniformly set at 60% — a single flat value applying to every pairing of currency-pair buckets, with no exceptions or bucket-specific carve-outs.
Sourced from MAR21.89.
https://www.bis.org/basel_framework/chapter/MAR/21.htm?inforce=20230101&published=20260323
All Six Delta Risk Classes: A Capstone Comparison
With FX now complete, every individual delta risk class in the Sensitivities-Based Method has been covered across Articles 9 through 15. Here is the complete picture, side by side.
| Risk Class | Buckets | Risk Weight Structure | Cross-Bucket Correlation |
| GIRR | 1 per currency | Tenor-varying (1.1%–1.7%) | Flat 50% |
| CSR non-securitisations | 18 | 18 distinct values (0.5%–12%) | Graduated, Table 5 (0–75%) |
| CSR securitisations (CTP) | 16 | 16 distinct values (2%–16%) | Identical to non-sec |
| CSR securitisations (non-CTP) | 25 | 25 distinct/derived values (0.8%–3.5%) | Flat 0% (100% for bucket 25) |
| Equity | 13 | 13 distinct values (15%–70%) | 4 rules (0%/15%/45%/75%) |
| Commodity | 11 | 11 distinct values (20%–80%) | 2 rules (0%/20%) |
| Foreign exchange | 1 per currency pair | Flat 15% (or ÷√2 for specified pairs) | Flat 60% |
Looking Ahead
With all seven risk classes’ delta structures now built (GIRR, three CSR sub-classes, equity, commodity and FX), the next two articles step back to cover vega and curvature risk buckets, risk weights and correlations — which work across all risk classes together using a shared liquidity-horizon-based structure, rather than being defined separately within each risk class the way delta has been.
Frequently Asked Questions
What is the base currency approach in FX risk?
A supervisor-approved alternative to calculating FX risk against the reporting currency. The bank instead measures FX risk against a single chosen base currency, while separately capturing translation risk between the reporting currency and that base currency, then converts the result back to the reporting currency using the spot exchange rate.
Why is there no onshore/offshore distinction for FX risk, unlike GIRR?
The standard treats these as different types of risk: an onshore and offshore currency curve can behave differently for interest rate purposes (justifying separate GIRR treatment), but the spot exchange rate itself is not distinguished this way for FX risk purposes.
What is the risk weight for FX risk under FRTB?
A flat 15% for all currency pairs, which can optionally be divided by the square root of 2 for a defined list of 19 specified currency pairs and their first-order crosses.
Link to the previous article
Pingback: FRTB Series 16 - MAR21: Vega Risk Factor (All Asset Classes)