The dollar as a safety machine

Market participants use safe assets to settle payments, store value, and post collateral. Because these assets provide services beyond their cash flows, they carry a ‘convenience yield’: investors accept a lower return to hold them (Krishnamurthy and Vissing-Jorgensen 2012). Dollar-denominated safe assets – Treasuries, agency debt, and high-grade private claims – command the highest convenience yield in the world, and the dollar exchange rate is the price that clears the global market for the safety and liquidity services they provide.

One way to get at how much foreign investors value this service is the covered-interest-parity (CIP) wedge: the gap between the yield on a US Treasury bill and the yield on a synthetic dollar bill built from foreign government bonds swapped into dollars. Since both yields are denominated in US dollars, this comparison reveals the premium of the US Treasury relative to a dollar safe asset constructed from a foreign government bond.

The wedge is positive on average, widens when volatility spikes – as it did during the Global Financial Crisis – and co-moves with a stronger dollar. It is the visible tip of a much larger premium. Jiang et al. (2021) estimate that the uncovered-interest-parity component (i.e. the unhedged return on holding a safe/liquid dollar bond relative to foreign currency bond) of the dollar’s convenience premium is about nine times the CIP component, so a 20-basis-point CIP basis signals a dollar premium on the order of 2%. When volatility spikes and investors increase their demand for holding dollar safe assets, the convenience yield rises and the dollar appreciates relative to its long-run level. Jiang et al. (2021) also estimate that transitory shifts in the demand for dollar safe assets from its long-run level explain between 16% and 28% of the quarterly fluctuations in the dollar exchange rate.

But what happens if the demand for dollar safe assets permanently falls? Then the long-run level of the exchange rate will also change.  Our question is what happens to the long-run dollar exchange rate, and to US interest rates, if the world stops demanding US safe assets as international liquidity?

Three signs the regime may be shifting

The question is not purely hypothetical. Three patterns in recent data are consistent with an erosion of the dollar’s special status.

First, the convenience yield on public dollar safe assets has fallen. Measured relative to currency-hedged euro funding, the premium on Treasuries and agency debt has compressed from its 2022 peak towards zero (Figure 1). Because the UIP wedge magnifies the measured CIP basis several-fold, the underlying decline in the value foreign investors place on dollar safety is larger than the raw series suggests.

Figure 1 Convenience yield on public dollar safe assets has declined

Second, the erosion is concentrated in public debt. Dollar safe assets include public claims (Treasuries, agency debt) and private claims (commercial paper, investment-grade corporate bonds, repo, short-term bank liabilities).  Figure 2 shows that the premium on public dollar assets (10-year CDS-hedged corporate yields vs 10-year Treasury, in green) has fallen, while the premium on private dollar assets (10-year CDS-hedged corporate yields relative to the 10 year repo-indexed swap rate or SOFR, in blue) has held broadly steady.

Figure 2 Public debt has lost more convenience yield than private debt

Third, foreign ownership has retreated from public debt specifically. The share of public dollar bonds held by foreign investors has slid from around 45% to about 30% over the past decade, while the foreign share of private dollar bonds has stayed near 15% (Figure 3). Together, these patterns – a secular convenience-yield decline, sharper for public than private assets, alongside foreign outflows from public debt – are what one would expect to see at the start of a loss of reserve currency status.

Figure 3 Foreign ownership of US public debt has fallen

What is the end-game?

To pin down the magnitudes, we build a two-country general equilibrium model of the US and the rest of the world. Each country produces a distinct good, and aggregate consumption is a constant-elasticity-of-substitution basket of home and foreign goods, so home bias matters. There are three bonds: a dollar Treasury bond, a dollar private bond, and a foreign bond. The two dollar bonds deliver convenience benefits, modelled through a bond-in-utility specification in which investors derive direct utility from holding dollar safe claims. Governments can levy tariffs, rebated to households, and goods and bond markets clear.

The mechanism runs through wealth and relative prices. Because foreigners value the convenience of dollar safe assets, the US borrows from the rest of the world at rates below the foreign rate. This interest saving is seigniorage – the ‘exorbitant privilege’ – and it funds a permanent US trade deficit. The extra income makes US households richer; combined with home bias, their spending bids up the price of home goods, leaving the real dollar permanently stronger. As a result, the reserve currency status causes an overvalued dollar in this model, an insight that builds on Jiang et al. (2024) and Jiang (2024). The loss of that status reverses the channel: seigniorage disappears, US households are poorer, demand for home goods falls, and the dollar depreciates in real terms.

How large is the effect?

We calibrate the model to US data. The bond-demand elasticity follows Krishnamurthy and Vissing-Jorgensen (2012); bond quantities come from the 2016 Flow of Funds, with public debt at 94% of GDP (39% held abroad) and private debt at 111% of GDP (14% held abroad). Convenience-yield weights are chosen to match foreign holding shares and a 2% convenience yield from Jiang et al. (2021). The trade elasticity is set to 1.5, following Itskhoki and Mukhin (2021).

Consider the cleanest experiment: foreign demand for dollar safety goes to zero. As foreign investors no longer hold the dollar safe asset at a premium, US seigniorage of about 1% of GDP vanishes per annum. The real dollar depreciates permanently by 8.8%, and US real interest rates rise – by 87 basis points on government debt and 72 basis points on private debt – because once foreign investors no longer demand dollar safe assets, roughly 50% of GDP in debt must be absorbed by US investors, who require a higher yield to hold it.

The exchange rate and the interest rate numbers come from different elasticities. The exchange-rate move is governed by the consumption home bias and the trade elasticity: a wealth loss of 1% of GDP requires an 8.8% depreciation to clear the goods market. The interest rate move is governed by the elasticity of substitution in the bond market: the less substitutable dollar bonds are, the larger the rate increase needed to clear the bond market.

It is worth stressing how different this is from a transitory shock. Take the same 0.87% convenience-yield decline but treat it as temporary, with a quarterly persistence of 0.85 – a half-life of about a year. The present-value formula then implies essentially no change in today’s dollar if interest rates fully adjust to this shock, or at most an appreciation of around 1.5% if interest rates are held fixed. The 8.8% depreciation arises precisely because the loss of reserve status is permanent: it works through the long-run exchange rate expectation, not the transitory variations. Over the last one and a half years, the moves in the dollar have likely reflected movements in the transitory component plus a probability-weighted change in the permanent value of the dollar.

Sizing the cost in terms of a wealth effect

The exchange rate number is the one that draws attention, but it is not the largest cost. The recurring loss of seigniorage – foreigners hold close to 52% of GDP in dollar safe assets earning a 2% convenience yield – is worth about 1.04% of GDP every year. Capitalising that flow at a discount rate net of growth (a risk-free rate of 1.7%, a GDP risk premium of roughly 1.1%, and growth of 1.8% per annum) gives a present value near 107% of GDP, or roughly $33 trillion.

Implications

The policy debate, framed sharply by Miran (2024), holds that reserve currency status props up the dollar and damages US manufacturing, so shedding that status would reindustrialise the economy. Our calibrated exercise qualifies that view on two counts. The exchange-rate effect is present but modest: reserve status buys a dollar appreciation of about 9%, so giving it up would not dramatically transform US trade competitiveness. The far larger consequences fall elsewhere – roughly a 90-basis-point rise in US borrowing costs and a national wealth loss on the order of $29 trillion. Whatever one thinks of an overvalued dollar, the cost of losing reserve currency status would be paid mainly in higher fiscal burden and lost wealth, not in a manufacturing renaissance.

References

Itskhoki, O and D Mukhin (2021), “Exchange rate disconnect in general equilibrium”, Journal of Political Economy 129(8): 2183–2232.

Jiang, Z, A Krishnamurthy, and H Lustig (2021), “Foreign safe asset demand and the dollar exchange rate”, Journal of Finance 76(3): 1049–1089.

Jiang, Z, A Krishnamurthy, and H Lustig (2024), “Dollar safety and the global financial cycle”, Review of Economic Studies 91(5): 2878-2915.

Jiang, Z (2024), “Exorbitant privilege: A safe-asset view”, NBER Working Paper No. w32454. 

Krishnamurthy, A and A Vissing-Jorgensen (2012), “The aggregate demand for Treasury debt”, Journal of Political Economy 120(2): 233–267.

Miran, S (2024), A user’s guide to restructuring the global trading system, Hudson Bay Capital.