Why countries self-insure in low-yield assets
This chapter focuses on a specific puzzle in foreign-reserve management: if the Global Financial Safety Net contains international facilities beyond a country’s own reserves, why do governments still hold so many immediately usable, relatively low-yield assets? The answer is not simply that reserve managers dislike risk. It is that external protection is conditional and uneven, while the cost of being unable to obtain foreign currency in a crisis can be much larger than the return forgone in normal times. Understanding that diagnosis shows what a Synthetic Reserve Deposit is trying to change—and what it cannot replace.
The safety net is layered, not interchangeable
The Global Financial Safety Net is commonly described as having four layers: a country’s own international reserves, bilateral swap arrangements, regional financing arrangements, and the resources of the International Monetary Fund. That list is useful, but it can create a misleading picture if it is read as one large account from which any country can withdraw when needed. The layers differ in their providers, currencies, eligibility rules, ceilings, activation procedures, and speed. Some create additional international liquidity when activated; others redistribute liquidity that already exists among participating members.
For a reserve manager, those differences are not administrative footnotes. They determine whether protection is available in the state of the world that actually matters. A bilateral swap line may offer reliable access to one central bank while offering none to another. An arrangement denominated in a currency that cannot be used for the relevant external payment may be less useful than its headline amount suggests. A regional facility may require an eligible shock or a formal request. IMF resources can be important and sizable, but access is bounded by ordinary limits and institutional procedures. These instruments may be valuable parts of a country’s contingency plan without being equivalent to foreign currency already controlled on its own balance sheet.
The four layers also differ in what they ask the user to surrender or wait for. National reserves are costly to hold, but the authority chooses their composition and can deploy them directly. A swap arrangement can provide powerful access to a reserve currency, yet it depends on the existence and terms of a relationship between institutions. A regional arrangement pools insurance across members, which can spread risk, but it also requires rules for deciding when and how members draw on the pool. IMF resources supply a multilateral financing channel, but a country still enters an institutional process with limits and conditions. These are not criticisms of the facilities; they are descriptions of different kinds of insurance.
Timing compounds the difference. A reserve manager preparing for a sudden external payment cannot value a facility only by its maximum announced amount. They must ask whether the facility can be activated before the payment is due, whether the required currency will be available, and whether the instrument’s eligibility and procedural conditions fit the shock. Even a facility that is ultimately adequate may not substitute one-for-one for assets already on hand. The possibility of access is useful insurance; pre-positioned cash is immediate insurance. A country may rationally hold both.
This is the difference between global capacity and country-level protection. Global capacity adds together facilities that do not have the same users, timing, or conditions. Country-level protection asks a narrower question: when this country faces this shock, which resources can it plausibly mobilize, in which currency, and within the time required to prevent a liquidity problem from becoming a broader crisis? The IMF paper emphasizes that coordination across layers is limited and activation may be sequential. A published aggregate can therefore look adequate while individual countries still face shortfalls, particularly among emerging-market and low-income economies.
That unevenness helps explain the persistence of self-insurance. A country’s own reserves are pre-positioned, under the monetary authority’s control, and generally usable without waiting for another institution to assess eligibility or assemble a package. This does not make reserves costless or make external facilities unimportant. It means that reserves perform a different function: they insure against the timing and uncertainty of access itself.
The evidence summarized in the IMF paper is consistent with that insurance interpretation. Higher foreign-exchange reserves have been associated in the cited literature with a lower probability of subsequent financial crises. In a stress-event comparison reported by the paper, countries with greater reserve accumulation showed lower sovereign spreads and smaller output losses than countries with less accumulation; the underlying estimates report a trough GDP-growth difference of roughly two percentage points. Those results are reported evidence from cited work, not a new causal estimate produced here, but they clarify why a country may rationally pay to hold a liquid buffer. The reserve is valuable precisely because the external backstop may arrive later, under different terms, or not at all for that country.
The finding should not be read as proof that reserves cause every favorable outcome. Countries that accumulate more reserves may differ from other countries in fiscal capacity, exchange-rate policy, external debt, or the shocks they face. The paper uses the evidence to establish the insurance value of reserves, not to settle every question about optimal reserve adequacy. For the present diagnosis, the robust lesson is narrower: if a country’s own liquid buffer reduces the consequences of a funding shock, then replacing that buffer requires a credible alternative with comparable timing and control. A higher expected return is not comparable protection unless it can be converted into usable foreign currency under the same conditions.
The diagnosis should still be kept provisional. Reserve accumulation can reflect many motives: exchange-rate intervention, foreign-liability hedging, precaution against sudden stops, or a desire to reassure investors. The sources do not establish that every country holds too many reserves or that every external facility is unreliable. They establish a narrower and more decision-relevant point: the GFSN’s layers cannot be treated as fungible substitutes, so national reserve holdings remain the most direct form of self-insurance for many sovereigns.
From private opportunity cost to a system-wide externality
Self-insurance protects an individual country, but the same choice can create costs that no single reserve manager fully considers. The simplest cost is the return differential. A country holds short-duration, high-quality foreign assets because it wants them to be available in stress. Those assets may earn less than the country’s own sovereign funding cost or less than a longer-horizon portfolio. The gap is a quasi-fiscal cost of insurance: the country exchanges some return for immediate usability.
The cost is not only the difference between two yields. If many countries respond to uncertainty by demanding the same narrow set of short-duration safe assets, their combined demand can push down the return on those assets. The IMF paper describes this as a safe-asset-scarcity externality. Individual central banks may not internalize the effect of their own reserve demand on global safe rates, nor the way those rates influence private-sector borrowing and risk-taking elsewhere. A decision that is prudent on one balance sheet can therefore contribute to a less efficient allocation across the system.
The mechanism works through an interaction between precaution and pricing. When the safest claims are especially valuable in bad states, investors accept a lower yield to hold them. More reserve demand can deepen that convenience premium. The lower safe rate then changes the relative attractiveness of longer-dated bonds, equities, and other assets, while also changing the incentives of borrowers that rely on cheap short-term funding. The paper relates this channel to the possibility of excessive leverage and foreign-currency mismatches elsewhere in the system. It is not saying that a lower safe rate is always harmful; it is saying that the private decision to insure can have equilibrium consequences that are not visible in the reserve manager’s own accounting return.
This diagnosis also distinguishes a pecuniary externality from a direct failure of a market to clear. The assets are still traded, and a reserve manager may be acting exactly as their mandate requires. The concern is that the aggregate price response to many individually reasonable choices can alter borrowing, risk allocation, and the cost of insurance for everyone. A remedy therefore has to preserve the insurance service while reducing the need for every country to demand the same narrow assets. That is why the IMF paper considers expanding the effective supply of reserve claims rather than simply imposing a penalty on reserve holdings.
That does not mean the solution is to tell every reserve manager to reach for yield. The insurance need remains real, and the sources explicitly leave room for reserve accumulation to be globally efficient in models without the relevant externalities. The more precise issue is whether the same liquidity service can be produced with a broader and better-diversified set of underlying assets. If it can, then the supply of claims that satisfy reserve-use requirements becomes more elastic, and the system need not concentrate as much precautionary demand in the narrowest class of safe assets.
Reserve managers already have a partial answer to this problem: they divide their portfolios by purpose. The BIS describes a working-capital or liquidity tranche made up of the safest assets and an investment tranche containing assets that are less liquid or more exposed to market risk but may produce higher returns over a longer horizon. Its estimate is that around two-thirds of reserve managers use some form of tranching. The IMF paper, drawing on the World Bank’s 2023 reserve-management survey, reports the same institutional logic in more operational terms: central banks use several adequacy measures, and investment tranches have longer average horizons than liquidity tranches.
Tranching separates jobs, but it does not fully solve the coordination problem. Each country must decide how much can move into its investment tranche, maintain the rules and operations needed to manage it, and continue to communicate that its overall reserve position can be mobilized when required. The BIS reports that roughly 20–25 percent of reserve managers have added equities, but that figure itself illustrates the boundary: a majority have not, and even those that do generally keep the exposure limited. Diversification can improve long-horizon return, yet it can also add duration, credit, equity, valuation, and operational risks at the moment when the reserve manager may need cash.
The time horizon is part of the design, not a cosmetic label. The IMF paper reports that investment tranches in the World Bank’s 2023 reserve-management survey had an average horizon of about 44 months, compared with about 17 months for liquidity tranches. A manager can accept a wider distribution of outcomes over the longer horizon only if the shorter-horizon tranche remains adequate for expected interventions and external payments. Moving assets from one tranche to another is therefore a choice about the probability and cost of forced sales, not merely a search for a higher average yield.
There are also limits to what the practice figures can tell us. The BIS percentages are estimates reported in a speech, not a complete, independently reproduced survey in this analysis. They show that tranching and diversification are established practices, but they do not identify a single optimal equity share or prove that the same allocation would be suitable for every reserve issuer. The important institutional evidence is the direction of the trade-off: reserve managers are already willing to separate liquidity needs from longer-horizon return, but they do so cautiously because the two jobs can collide in a crisis.
The Synthetic Reserve Deposit is designed as a supply-side response to that boundary. Rather than asking every central bank to build a more complicated portfolio while preserving reserve usability on its own balance sheet, a pooled vehicle would combine participants’ claims against a common settlement buffer and diversified investment portfolio. The pool could centralize trading, custody, collateral management, and the monitoring of a published portfolio rule. In principle, a claim on the whole portfolio could satisfy safety and liquidity requirements even if every security inside it would not qualify individually.
That last sentence contains the proposal’s institutional bet. A central bank may find it easier to defend a portfolio as reserve-usable when every security it owns is plainly a reserve asset. If a portfolio includes assets outside that category, changes in prices or trades can change its composition and make its status harder to communicate. A pooled arrangement could take responsibility for maintaining a portfolio-level rule and making that rule credible to participants and their authorities. It could also reduce duplicated transaction costs when participants’ liquidity needs are not perfectly synchronized: one country’s withdrawal and another’s deposit may be partly netted inside the pool rather than turned into separate trades with private intermediaries.
But pooling does not erase the cause of the original trade-off. It changes who performs the diversification and liquidity-management work, not the fact that diversified assets can lose value. It may reduce the amount of forced selling required to meet temporary cash needs, but only if the pool has a credible settlement buffer or an additional collateralized funding route. It may broaden the supply of reserve-usable claims, but only if legal recognition, governance, valuation, and operational control are accepted by participating authorities. The proposed remedy is therefore conditional on the diagnosis: the problem is not a universal shortage of investment ideas; it is the difficulty of coordinating return, safety, and immediate usability under uneven external protection.
This leaves the SRD with a clearly bounded role in the GFSN. It would improve the self-insurance layer by pooling part of the reserve-management task. It would not turn a bilateral swap line into a universal facility, replace IMF financing, or guarantee that every country can draw foreign currency in every crisis. To assess whether it can deliver even that narrower improvement, the next step is to inspect the pooled claim itself: how it is valued, who bears a market loss, and what happens when a participant needs cash.