When safe reserves become a liquidity problem
This series examines a proposed Synthetic Reserve Deposit (SRD) and the question at its center: can a foreign-reserve claim give a central bank immediate access to foreign-currency cash without forcing it to sell market-risky assets at the worst possible moment? The answer depends on a distinction that is easy to miss. An asset can be readily usable and still lose market value; a facility can postpone liquidation without guaranteeing principal; and a claim can look economically useful without automatically qualifying as an official reserve asset. The aim is to build a practical way to judge those distinctions, rather than treating “liquidity” as a synonym for safety.
What foreign reserves are buying in a crisis
Foreign reserves are often described as a portfolio, but their most important function appears when ordinary financing becomes difficult. A central bank may need foreign currency to slow an exchange-rate depreciation, meet an external payment, hedge foreign liabilities, or reassure markets that it can act. In those circumstances, an asset that can be mobilized quickly has insurance value even if its expected return is modest. The cost of holding it is not a mistake in portfolio management; it is the premium paid for being able to respond without first negotiating for credit or waiting for a market to recover.
The Bank for International Settlements’ account of the 2022 tightening cycle makes that use concrete. As U.S. monetary policy tightened and the dollar strengthened, many central banks drew down reserves to limit the speed and extent of exchange-rate depreciation. The BIS says global foreign-exchange reserves were still about $12.3 trillion after the subsequent rebuilding began, below their peak in early 2022. The important point is not the aggregate number by itself. It is that reserves were used as a policy instrument, not merely held as an investment statement.
That use changes the meaning of risk. For a long-horizon investor, a temporary decline in a bond or equity portfolio may be tolerable if the expected return compensates for it. For a reserve manager facing a sudden external need, the same decline can be expensive because selling then crystallizes the loss and may require selling into a thin or falling market. The reserve asset must therefore be both liquid enough to turn into cash and reliable enough to support action when markets are under stress. This is why central banks continue to hold much of their reserves in traditional assets such as U.S. government bills and bonds: their “flight-to-safety” quality offers a reasonable expectation of rapid execution and capital preservation relative to riskier alternatives.
There is an institutional reason for this caution as well. The wider Global Financial Safety Net includes national reserves, bilateral swap arrangements, regional financing arrangements, and IMF resources, but those layers do not provide identical protection to every country. Access can depend on the borrower, the currency, eligibility, limits, and the state of the world in which support is requested. A country cannot treat every published facility as cash already under its control. Its own reserves remain the most clearly pre-positioned layer of protection, even when they carry an opportunity cost.
This makes reserve management an exercise in matching assets to states of the world. A central bank does not know whether the next call on its foreign currency will be a measured intervention, a sudden payment need, or a broad loss of market confidence. It must hold enough immediately usable assets before the event, not merely enough wealth on average. That requirement explains why a portfolio that looks inefficient from the perspective of a patient investor may be sensible for a monetary authority. The relevant comparison is not simply “Treasury bills versus equities.” It is the value of reliable cash access in a bad state versus the return forgone while waiting for that state to arrive.
The tension becomes sharper when a country’s outside options are uncertain. A swap line may be large but unavailable to that country; an IMF program may provide financing but require activation, conditions, and time; a regional arrangement may have its own eligibility and currency limits. These instruments can still be important, but they do not remove the need for a balance-sheet buffer that the authority controls directly. The resulting reserve demand is therefore defensive rather than opportunistic. It is shaped by the cost of being unable to act, not just by the expected return on the assets held.
That is also why the question is not whether central banks should replace liquid reserves with riskier investments. In practice, reserve managers already divide portfolios by purpose: a working-capital or liquidity tranche is kept close to the safest and most usable assets, while a longer-horizon investment tranche can accept more duration, credit, or other exposure. The unresolved problem is how far that separation can go while the whole reserve position remains credible in a crisis. The SRD is interesting because it proposes to move some of that portfolio-level coordination into a common vehicle rather than asking each country to solve it alone.
Liquidity is not the same as a stable price
The central question becomes clearer if liquidity is split into three ideas. The first is asset usability: whether a monetary authority controls an external foreign-currency asset that can count toward its reserve needs. The second is cash access: how quickly that authority can obtain spendable foreign currency when a shock arrives. The third is value stability: how much the claim can lose while the authority is trying to use it. These properties are related, but they are not interchangeable.
Consider a simple stress sequence. Suppose a reserve portfolio contains assets that can be sold in normal markets but whose prices fall after a sharp interest-rate move. The portfolio may still be liquid in the narrow sense that buyers exist. Yet converting it into cash immediately produces less foreign currency than its pre-shock accounting value. If the reserve manager waits for prices to recover, cash access is delayed; if the manager sells, value stability is sacrificed. A useful design must say which of those costs is being reduced. It cannot describe both as “liquidity” and leave the trade-off invisible.
There is a second distinction between ordinary redemption and emergency funding. Redemption converts a claim into cash and therefore normally realizes the claim’s current market value. Secured borrowing can provide cash while leaving the claim outstanding, but it creates a loan, a collateral requirement, and a future repayment obligation. That may be valuable when a price decline is temporary, yet it does not turn the collateral into a promise that its value will recover. The provider is protected by terms such as a haircut and margining; the holder remains exposed to the asset’s eventual gain or loss.
A Treasury bill may score well on all three in many circumstances, which is why it is such a powerful reserve asset. But “easy to sell” does not mean “immune to a loss.” Longer-duration bonds can lose value when interest rates rise. Corporate bonds and equities can fall when risk appetite disappears. Even a portfolio that remains tradable can be worth less precisely when its owner most wants to preserve wealth. Conversely, a claim may retain exposure to those assets while offering a separate route to short-term cash. That route improves timing, not the underlying market outcome.
The distinction also explains why a cash buffer is not automatically a guarantee. A buffer can settle ordinary withdrawals and reduce the need for small, costly trades. It cannot promise that every holder will receive yesterday’s price after the pool has had to sell assets today. If the claim is valued at the price that can actually be realized, market losses are recognized rather than transferred to whoever remains in the pool or to an unspecified guarantor.
The proposal examined here is the Synthetic Reserve Deposit. It would be a floating-value claim on a pooled vehicle holding a settlement buffer and a diversified investment portfolio. In some versions, a participant could also borrow foreign currency for a short period against the market value of their SRD units. The pool would centralize portfolio selection, settlement, custody, and—where the facility exists—collateral management. It would not make the underlying portfolio riskless. The intended separation is narrower and more precise: the participant might obtain cash without immediately liquidating the investment portfolio, while the participant still bears the portfolio’s market risk.
This is why the word “synthetic” matters. The claim would seek to satisfy reserve-use requirements at the level of the pooled portfolio, even though not every security inside it would necessarily be an individually recognized reserve asset. A central bank managing its own balance sheet may prefer to hold only securities that are plainly eligible, because that makes the reserve status of the portfolio easy to communicate as its composition changes. A monitored pool could instead publish and maintain portfolio rules designed to make the claim’s combined liquidity and safety properties credible. That is a design possibility, not an automatic legal result.
The Synthetic Reserve Deposit as a testable proposal
The SRD should therefore be judged by several questions at once. Does pooling improve the return available for a reserve-compatible claim without taking on an unacceptable downside? Does the redemption rule expose both departing and remaining holders to the same contemporaneous value, rather than hiding losses through stale pricing? If a borrowing window is used, are haircuts, maturity, margining, and access limits strong enough to protect the liquidity provider without shifting ordinary market losses away from SRD holders? And can the claim satisfy the legal and operational conditions that make a foreign-currency asset freely usable by a monetary authority?
Those questions also explain why the proposal matters beyond the return earned by one central bank. When many sovereigns insure themselves independently, they concentrate demand in a narrow class of short-duration safe assets. That concentration can depress the return on those assets and leave countries paying a high collective price for the same liquidity service. Pooling could broaden the set of assets backing a reserve-usable claim and reduce the need for each country to solve diversification, settlement, and contingent funding alone. But the benefit would depend on credible governance, market execution, and participation rules. A pooled claim that cannot be trusted in a stress episode would not solve the problem; a claim that offers cash only by hiding losses would merely move the problem elsewhere.
The proposal also changes the location of decisions. Instead of every reserve manager independently deciding how much duration, credit, or equity exposure can coexist with immediate usability, a pooled vehicle would publish a portfolio rule and apply it across participants. That could make a diversified reserve claim easier to monitor, but it would also make governance consequential: someone must define executable value, maintain the settlement buffer, enforce collateral terms, and decide what happens when markets become hard to price. Centralization can reduce duplicated trading and custody work, yet it can also create a common point of operational or institutional failure.
A fair assessment must therefore examine both sides of the balance sheet. On the asset side, the pool may earn more by holding a broader mix than a purely short-duration Treasury portfolio. On the liability side, participants still expect cash when they need it, and the pool must meet that expectation without granting one group a hidden priority over another. The answer will depend less on the label “deposit” than on the valuation rule, the funding route, and the authority that stands behind any contingent liquidity. Those are the features that later evidence must test.
The first conclusion is consequently modest but important. Foreign reserves cannot be separated from market risk in the sense of becoming immune to changes in bond, credit, or equity prices. An SRD could separate the moment at which a holder needs cash from the moment at which the pool must sell risky assets. Whether that is valuable depends on who bears the loss, how current value is measured, and whether the claim remains legally and operationally usable. The next question is why the existing safety net leaves so many countries relying on concentrated self-insurance—and what system-level problem a pooled claim is meant to address.