Hedging a Credit Portfolio: Cash Bonds versus Index CDS
A UK credit manager runs a £900m sterling investment-grade corporate bond portfolio benchmarked to a broad index. Modified duration is 6.4 years, spread duration 7.2 years, yield 5.4% and average rating A-. The manager expects a recession within 12 months, with spreads widening by 100bp and default rates rising to 1.5% from 0.3%. Two hedging routes are proposed. First, sell £300m of BBB-rated cyclical holdings, which are quoted with a 15bp bid-offer spread and would take several days to liquidate, and hold the proceeds in cash yielding 4.9%. Second, buy five-year CDS protection on the iTraxx Main index at 85bp, with a notional of £300m. The portfolio includes £120m of bonds held at amortised cost and used as collateral in an open repo facility; the remainder is fair-valued. The risk team notes that the index is dominated by financials, which are only 8% of the portfolio.
1. Compare the two hedging routes on cost, effectiveness and reversibility.
Read the worked answer
Selling £300m of cash bonds. This is an immediate and clean reduction in credit risk, removing exposure to the names most likely to default. It costs 15bp bid-offer on £300m, about £450k, plus several days of market impact in precisely the conditions where liquidity is worst. There is a carry drag: 5.4% yield replaced by 4.9% cash is about £1.5m a year, and the manager gives up any spread tightening. On accounting, selling amortised-cost bonds crystallises a loss that would otherwise not be recognised; selling fair-valued or FVOCI bonds moves the loss through OCI. The £120m of repo-encumbered bonds cannot easily be sold without unwinding the facility.
Buying £300m iTraxx Main protection at 85bp. This costs about £255k a year, has no upfront, is reversible, and does not disturb the cash portfolio or the repo. But the hedge is against the index, not the portfolio. The index is financial-heavy while the portfolio's risk is BBB cyclicals; if cyclical spreads widen 150bp and the index only 100bp, the hedge underdelivers. Index spread duration of about 4.5 is also much lower than the portfolio's 7.2. Sizing: £300m x 4.5 equals £1.35bn of spread-duration protection against £900m x 7.2, or £6.48bn of portfolio spread duration, so the hedge covers only about 21% of the risk. Hedging 50% would need roughly £720m notional. Counterparty and margin risk are low because the trade is cleared, but the position needs collateral and daily management.
Conclusion. The CDS is the better instrument for a tactical, reversible hedge, but as proposed it is badly mis-sized and carries significant basis risk. Selling cash bonds is better where default risk is concentrated in specific names.
2. Recommend a course of action, including how to handle the accounting and repo constraints.
Read the worked answer
Recommendation: a combined approach.
- Sell £120-150m of the weakest BBB cyclicals where default risk is highest and where a bid is still available, accepting the 15bp cost. Prefer fair-valued or FVOCI holdings over amortised-cost holdings so the accounting loss is not a surprise, and do not disturb the repo collateral unless the facility can be re-collateralised at short notice.
- Buy £600-700m of iTraxx protection rather than £300m, bringing the hedge to about 45-50% of spread duration. Accept the index basis as the price of liquidity.
- Monitor the basis. If the portfolio-index spread differential widens beyond a pre-agreed level, switch to a bespoke CDS basket or single-name protection on the largest holdings.
- Match tenor better. A five-year CDS has less spread duration than a 7.2-year cash portfolio; use seven to ten year CDS, or a larger notional, and document the residual duration gap.
- Accounting. Flag to the audit committee that the economic hedge will not qualify for hedge accounting, so the CDS mark-to-market runs through profit and loss while the amortised-cost bonds do not, creating reported volatility even though the economics are hedged. This affects communication to stakeholders, not the hedging decision itself.
- Repo. Because £120m of bonds are encumbered, the sale must come from the unencumbered portion; the CDS overlay avoids that constraint entirely, which is a further argument for combining the two routes rather than relying on either alone.