Why One-Way CSAs Are an Inefficient Approach to Credit Risk Management

One-way Credit Support Annexes (CSAs), most commonly used by supranationals, agencies, and certain public-sector entities, are often justified on the basis of counterparty strength. In modern derivatives markets, however, they represent a structurally inefficient and value-destructive approach to managing credit risk, funding, pricing, and balance-sheet usage.

They embed persistent costs, distort competition, and transfer economic value to dealers without delivering commensurate risk reduction for the client.

Credit Risk That Could Be Eliminated — But Is Not

Under a properly calibrated two-way CSA, credit risk can be reduced to a negligible level. Variation margin eliminates ongoing mark-to-market exposure, while initial margin addresses gap risk and close-out uncertainty. With sufficient initial margin, bilateral collateralisation does not create any material credit risk for either party.

This is most clearly demonstrated in the case of supranationals with long-standing derivatives portfolios that are materially out-of-the-money—often by an amount far exceeding any plausible market move. In such cases, posting collateral against a portion of that exposure would release substantial dealer XVA and regulatory capital without introducing any realistic risk of loss for the supranational. The collateral posted would sit well inside the distance to default, rendering the probability of a margin shortfall immaterial.

A one-way CSA prevents this economically efficient outcome. Instead, the dealer is forced to carry a permanently uncollateralised exposure to the client—an exposure that must be funded, capitalised, and priced. Credit risk that could be neutralised at low cost is instead warehoused on dealer balance sheets at high and recurring cost.

Balance-Sheet Costs and XVA Inflation

The consequences of this asymmetry are immediate and measurable. Uncollateralised exposure generates CVA, consumes regulatory capital, absorbs leverage capacity, and creates funding costs. These costs do not disappear; they are embedded directly into the prices quoted to the supranational.

Crucially, under a one-way CSA the dealer has no structural mechanism to mitigate these costs, regardless of the client’s true economic risk profile. Funding valuation adjustment (FVA) is driven by the dealer’s own funding curve, meaning the supranational is effectively charged at dealer funding rates—even though the supranational could fund the same exposure materially more cheaply through collateralisation.

Distorted Pricing and the Failure of Competition

In a seasoned, uncollateralised portfolio, it is the marginal XVA impact of a new trade—not the economics of the underlying market risk—that dominates pricing. These marginal XVA effects differ substantially across dealers and across trades. For any given transaction:

Some dealers will face incremental XVA costs;

Others will experience material XVA rebates driven by portfolio offsets, netting effects, or capital relief.

This heterogeneity fundamentally distorts competitive behaviour.

When a trade generates a large marginal XVA cost, a dealer recognises that there is a meaningful probability a competitor will benefit from an XVA rebate and therefore be able to price more aggressively. In such cases, the dealer rationally reduces pricing effort, knowing that even tight market-risk pricing may be uncompetitive once balance-sheet effects are included.

Conversely, when a trade generates a significant marginal XVA rebate, the dealer will price tightly—but will not voluntarily pass that rebate to the client. Instead, the dealer will attempt to retain the benefit, hoping that competitive pressure does not force it to be given up. Because XVA is opaque, dealer-specific, and not observable by the client, there is no reliable mechanism that ensures these rebates are transferred to the supranational.

The Supranational Is Short an Option on Pricing

The result is that the supranational is effectively short an option on pricing.

• When marginal XVAs are positive (i.e. costs), those costs are systematically and reliably passed through in pricing.
• When marginal XVAs are negative (i.e. rebates), the upside is rarely realised.

This asymmetry exists for two structural reasons:

• Dealers have a strong incentive to retain XVA rebates unless competition compels them to surrender them; and
• Supranationals lack visibility into the size or even the existence of those rebates, and therefore cannot negotiate their capture.

The payoff profile is therefore asymmetric and unfavourable: downside is transferred with certainty, while upside is contingent and typically retained by dealers. This is not a failure of execution or negotiation; it is a structural consequence of one-way CSAs.

Inferior Portability and Stress Performance

One-way CSAs also perform poorly in stress and default scenarios. Asymmetric collateralisation complicates trade portability, weakens resolution mechanics, and increases friction precisely when markets require balance-sheet flexibility.

While the supranational may appear protected in isolation, the broader market impact—through constrained dealer balance sheets and reduced intermediation capacity—is negative. By contrast, bilateral collateralisation supports portability, simplifies default management, and aligns with the design assumptions of modern clearing and resolution frameworks.

A Superior Alternative

A two-way CSA combined with a fair and proportionate level of initial margin provides a strictly superior solution.

• Variation margin eliminates ongoing mark-to-market credit exposure for both parties.
• Initial margin addresses gap risk in a transparent and quantifiable manner.

This structure materially reduces systemic balance-sheet usage, compresses XVAs to levels consistent with true economic risk, and restores clean, competitive pricing. Most importantly, it removes dealer-specific balance-sheet effects from price formation.

For the supranational, the central challenge is to maximise the recovery of XVAs already embedded in the existing portfolio—net of carry, but inclusive of XVA progression as markets evolve. This is an economic optimisation exercise, not a credit-risk decision.

Conclusion

One-way CSAs are a legacy construct that create the illusion of safety while embedding structural inefficiencies into pricing, balance-sheet usage, and market liquidity. They prevent credit risk from being neutralised where it can be eliminated most cheaply and instead force it to be warehoused by dealers at high and recurring cost—a cost ultimately borne by the supranational.

In modern derivatives markets, two-way collateralisation with appropriately calibrated initial margin is not merely fairer; it is economically and systemically superior. Overall costs can be materially reduced without assuming any meaningful additional credit risk, and the operational burden is modest. Specialist XVA expertise can support this transition and ensure that the benefits are fully identified, quantified, and realised.