If the watchdogs cannot see the risk, why are investors so sure they have transferred it. The Basel Committee says a lack of data is hindering its assessment of banks use of synthetic risk transfers. That is not an information glitch. It is the risk. A trade that supposedly removes credit exposure is being scaled up in the dark, leaving supervisors and counterparties with crude proxies instead of hard telemetry. When the dashboard is blank, the engine is running on hope.
Synthetic risk transfers let banks keep loans while buying credit protection on a defined slice of expected losses, often via credit default swaps. The reward is capital relief. The loans stay. The bank keeps origination economics and operational exposure. The protection seller harvests yield for absorbing a tranche of losses. This looks like a neat division of labor until you recall a basic engineering point: risk is conserved under stress. It migrates. The illusion is that a spreadsheet can convert a concentrated pool of loans into a diffused set of obligations that will behave in a crisis. Without reliable, comparable disclosure, there is no way to verify whether transfer is real, conditional, or illusory. When conditions tighten, synthetic hedges can meet legal gray zones, dispute mechanics, and counterparty fragility. The hedge that looks perfect ex ante can be path dependent ex post.
Uncertainty is not just about the distribution of losses. It is about the uncertainty of the distribution itself. When supervisors say their assessment is hindered by missing data, they are signaling Knightian uncertainty. That is where models break first. Thin disclosure around pool composition, seasoning, loss vintages, correlation assumptions, and trigger mechanics turns the market into a lemons problem. Sellers have more information than buyers and regulators. The highest risk portfolios have the strongest incentive to seek capital relief; the best portfolios can self insure. Adverse selection becomes the baseline. Basel sets significance thresholds for risk transfer, but if the inputs are opaque and heterogeneous, the test can be satisfied in letter while failing in substance. Definitions of credit events, cap on payouts, commutation rights, and performance covenants matter as much as expected loss numbers. In a data desert, fine print becomes the fulcrum.
Who is selling the protection. Hedge funds, insurers, reinsurers, credit funds. In a benign cycle, they want the spread. In a stress, they face funding calls, margin squeezes, and correlated mark to market losses. That is wrong way risk. The very moment a loan book deteriorates is the moment the protection seller is least able to pay or most likely to litigate. We have seen this movie. Monolines and AIG wrote protection on structured credit to harvest carry, only to discover that correlation is not a parameter. It is a regime. SRTs replicate the vulnerability in quieter form. The derivative is only as good as the margin, the collateral agreement, and the operational will to perform under pressure. If the hedge evaporates or is locked in dispute when the bank most needs it, capital relief becomes a cliff. The asset has not moved. Only the illusion of solvency has.
Regulatory capital relief is a levee. It controls floods in fair weather and encourages more building in the floodplain. Banks expand lending and risk taking on the confidence that tranches are transferred. That is rational under static assumptions. Under dynamic stress, relief disappears, premiums jump, and counterparties retrench. The system becomes more brittle as leverage climbs behind the levee. Supervisors in Europe have already warned against excessive reliance on securitization for capital planning because risk can remain embedded in the system. The logic is simple. Offloading expected losses via a derivative does not change the macro shock that drives correlations across the pool. When that shock arrives, multiple synthetic hedges call for liquidity, and the same players are on both sides. Market depth is thinner than it looks. A capital tool becomes a procyclical amplifier.
The incentives are not subtle. Banks know more about their loan pools than any outsider. They have the tools to optimize tranching, reference pool selection, and term sheets to pass significance tests while bottling tail risk in places that do not trigger immediate capital. That is a textbook principal agent problem. Goodhart s law applies: once a measure becomes a target, it stops being a good measure. The more precisely Basel defines SRT criteria, the more precisely structures will be engineered to meet them. Regulators are not blind to this. The Basel framework has evolved after each crisis, from Basel II s underestimation of securitization risk to Basel III s capital and liquidity fixes. But complexity spawns complexity. Each new rule adds state. Each state creates a surface to arbitrage. The result is a system that looks robust in normal weather and reveals hidden couplings in storms.
If supervisors want to see the risk, they should demand clarity where the fragility hides. Portfolio level data on obligor concentration, geography, industry, vintage, and loan grade migration. Clear mapping from internal ratings to default and loss given default, with back tests. Full visibility on tranche thickness, attachment detachment points, amortization profiles, and replenishment rights. Counterparty specifics matter: identity, diversification, collateral terms, margin frequency, thresholds, eligible collateral, and any triggers that change collateralization in stress. Dispute and credit event definitions, cure periods, and commutation options belong in the sunlight. Historical performance through past mini cycles is worth more than one perfect model. Supervisory reporting must be standardized enough to prevent gamesmanship. Public transparency should be sufficient to let markets price the true cost of protection. Data will not eliminate risk, but it can relocate it from rumor to analysis.
Another blind spot is basis. The reference portfolio in an SRT may not track the bank s broader credit risk. Losses depend on servicing, restructuring, and workout behavior that can diverge between the hedge and the bank s incentives. Macroe shocks, like a regional real estate bust or a sector specific downturn, can hit the bank s retained book and the reference portfolio differently. The hedge loses potency precisely when needed. Add in correlation misspecification. In credit, correlations are not stationary. They compress for years and then leap. The math assumes smoothness; the world supplies thresholds. Structural breaks turn safe tranches into cliff edges. Without long run data and stress scenarios that assume changing correlation regimes, risk managers are modeling peacetime on wartime assumptions. That is not prudence. It is fragility in a suit.
The answer is not a pamphlet of new ratios. It is incentive design that assumes models will be gamed and counterparties will suffer stress. Require durable skin in the game on both sides of the transfer so that originators and protection sellers bear losses that matter. Treat SRT capital relief as a function of counterparty strength, collateral quality, and correlation to the bank s own funding, with conservative haircuts that expand in booms. Consider floors so that leverage does not inflate on the back of bespoke hedges. Simpler backstops like a binding leverage ratio and robust liquidity buffers are harder to optimize away. Clearing can standardize margining, but do not mistake centralization for safety; it can create single points of failure. The aim is not to ban innovation. It is to ensure that when the surface is smooth, the structure beneath gains from volatility instead of shattering. A transfer you can survive without is a hedge. A transfer you need to stand upright is leverage by another name.
