What the portfolio simulations really show
The IMF paper’s Synthetic Reserve Deposit is not supported by the word “diversified” alone. This chapter examines the particular dollar portfolio, benchmark, constraints, and simulation method used to test the idea, then interprets the reported returns and downside measures without turning them into a forecast. The practical value of the exercise is learning what a reserve-compatible portfolio can show under explicit assumptions—and where the evidence stops.
A constrained portfolio, not an unconstrained search for yield
The quantitative illustration uses U.S.-dollar-denominated assets observed from January 2003 through December 2024. The menu includes U.S. Treasury indices, sovereign and sub-sovereign bond indices rated A, AA, or AAA, investment-grade corporate bond indices, and two equity indices: the MSCI World and the S&P 500. Bonds are separated by maturity buckets, so the menu contains both short and longer exposures. The exercise is deliberately limited to dollar assets. It keeps the reserve-liquidity question separate from the additional problem of deciding how a reserve portfolio should be divided among currencies.
That scope matters. A dollar portfolio is not a universal foreign-reserve portfolio, and a historical dollar return is not a prediction for a central bank whose liabilities, interventions, or reserve currency composition are different. The paper says the same balance-sheet logic could be applied to a multicurrency portfolio or one benchmarked to the Special Drawing Rights basket, but that would be another design choice. The reported numbers should therefore be read as evidence about one calibrated environment, not as a recommended allocation for all reserve managers.
The selection process also imposes rules before it looks for a winner. Every candidate receives a fixed 10 percent allocation to the 1–3-year U.S. Treasury index. The remaining weight is distributed across the broader asset menu. The combined weight of the MSCI World and S&P 500 cannot exceed 20 percent. A candidate must then pass two comparisons with a Treasury-only benchmark: its fifth percentile of annualized rolling 36-month returns must be higher than the benchmark’s median, and its mean annualized rolling return must also be higher. Among candidates that pass, the selected portfolio is the one with the highest mean return.
Those conditions encode a particular interpretation of reserve compatibility. The aim is not to maximize an unconstrained Sharpe ratio or to put as much as possible into equities. It is to look for a portfolio that can earn more than a short-duration Treasury strategy while meeting a lower-tail requirement over a medium-term horizon and keeping equity exposure limited. The constraint is doing substantive work: it rules out portfolios whose average return looks attractive only because they accept a downside distribution that would be difficult to reconcile with reserve use.
The selected allocation illustrates how the result is generated.