Introduction
The paper carries in one dynamical system five features of safe asset status that the literature identifies: several safe assets coexisting at a stable interior configuration, liquidity as a network benefit at low concentration and a congestion cost at high concentration, the slow erosion of institutional fundamentals, gradual and reversible transitions, and frequency-dependent competition, here among U.S. Treasuries (the incumbent, asset 1), a diversified basket of the other safe sovereigns, and gold, with institutional disruption as the mutation rate μ1.
Theoretical Foundations
Congestion turns the coordination game of the safe asset literature into a congestion game with an interior equilibrium that is unique at the calibration, and the replicator-mutator equation carries selection and mutation together: an asset whose fitness exceeds the population mean gains share, and holders leave an asset at its mutation rate.
xi is asset i's share and fj(x) asset j's fitness, Equation (3); Qij, an entry of the mutation matrix, is the probability that a holder of asset j moves to asset i in a period (Qjj = 1 − μj), and φ(x) is the population mean fitness.
The Model
Each asset's fitness is its intrinsic safety plus a concave network benefit less a convex congestion cost, so it rises with the asset's own share and then falls; gold carries the largest congestion cost and the basket the smallest.
Analytical Results
At the calibration the interior equilibrium is unique and lies on the falling branch of the incumbent's fitness, so the superiority σ, the incumbent's fitness relative to the population mean, rises as disruption erodes the share while the incumbent holds the majority, and the equilibrium moves continuously and reversibly toward the basket, the flattest profile; a structural rise in the incumbent's own congestion cost c1 instead lowers σ through unity.
Explorer: the equilibrium against the mutation rate
The controls set the incumbent's mutation rate μ1, intrinsic safety a1, network benefit b1, and congestion cost c1, and the benefit exponent γ, which the three assets share; the panels draw the equilibrium shares and σ along μ1, the mark at the setting of the controls; the regime follows the kernel's rule, Regime I where Treasuries hold at least half, II where they still lead the basket, and III where the basket leads.
Equilibrium shares against the mutation rate
Superiority σ against the mutation rate
The paper's counterpart: Figures 2 and 3.
Numerical Analysis and Simulation
Calibrated, the equilibrium reproduces the composition of the safe asset universe; the incumbent cedes the majority and then the plurality to the basket as the mutation rate rises, and a single basin of attraction covers the simplex; below a mutation rate of about 0.0074 a second equilibrium, with gold rather than the basket second, coexists with the dominant one.
| Parameter | U.S. Treasuries | Basket | Gold |
|---|---|---|---|
| a, intrinsic safety | 0.935 | 0.940 | 0.960 |
| b, network benefit | 1.048 | 0.778 | 0.663 |
| c, congestion cost | 1.300 | 0.980 | 1.600 |
| γ, benefit exponent | 0.60 | 0.60 | 0.60 |
| η, congestion exponent | 2.00 | 2.00 | 2.00 |
| μ, mutation rate | 0.02 | 0.04 | 0.03 |
| Destination | from U.S. Treasuries | from the basket | from gold |
|---|---|---|---|
| to U.S. Treasuries | — | 0.46 | 0.60 |
| to the basket | 0.55 | — | 0.40 |
| to gold | 0.45 | 0.54 | — |
Explorer: trajectories and basins on the simplex
With all three congestion costs scaled by a common multiplier, 1 at the calibration, and the other settings of the first explorer, the first panel draws trajectories from nine starting distributions and the second the basins of attraction, a numbered mark at each attractor.
Trajectories on the simplex
Basins of attraction
The paper's counterpart: Figures 8 and 9.
The Self-Limiting Feedback Loop
On the dominant-equilibrium branch, disruption lowers the incumbent's share, the lower share relieves its congestion and raises σ, and the higher σ opposes further erosion.
The Absence of Critical Slowing Down
The dominant eigenvalue of the linearized system stays bounded away from zero along the equilibrium curve, closest to it near the calibration, so the model predicts no early-warning signature before the transition.
Empirical Calibration
The observed composition of the global safe asset universe places the system at a mutation rate μ1 of about 0.02, inside Regime I, the stable majority; the paper states the calibration's limits, among them the crisis-time flight to safety it does not capture, in Sections 8.5 and 8.6.
Explorer: the observed position on the equilibrium curve
The control sets the observed Treasury share; the panel places it on the equilibrium curve at the calibration and reads the implied mutation rate, σ, and the slope of σ against the mutation rate there.
The observed share on the equilibrium curve at the calibration
The paper's counterpart: Figure 15.
Implications for the United States
The paper reads the projected fiscal path, below, as a rising mutation rate, which erodes the share along a smooth, reversible curve; under elevated disruption it predicts a diversified successor, with gold second, and lowering the incumbent's congestion cost acts on the one channel the restoring force leaves unopposed.
Conclusion
Erosion of the incumbent's status by disruption is self-limiting, and the transition to a basket-dominant configuration is smooth and reversible; the model predicts no critical slowing down before that transition.