Tether on cryptocurrency market
VoxEU Column Finance and Fintech

Leverage and stablecoin pegs

Stablecoins are similar to bank deposits in many respects, but they do not pay interest. This column analyses how stablecoins maintain their peg while compensating holders for run risk. It shows that holders are indirectly compensated through lending stablecoins to levered traders. These lending rates are particularly high when speculative demand is strong. Furthermore, stablecoin issuers can maintain the peg by changing the composition of their reserves as speculative demand changes.

Money is debt that trades with no questions asked, using Holmstrom's (2015) phraseology, but stablecoins are a new form of private money that trades with many questions asked (Gorton and Zhang 2021, Gorton et al. 2022b). Analysing how stablecoins maintain their peg is important to understand their resilience and links to the traditional financial system (Azar et al. 2022).

Cryptocurrencies have seen a tremendous boom in volume and attention in recent years. Despite aspirations to be a store of value and an alternative to fiat money, cryptocurrencies like Bitcoin have extreme price volatility. Stablecoins emerged to solve the volatility problem (Yeyati and Katz 2022). But stablecoins still fall short of fiat currency: Stablecoins are economically equivalent to bank deposits, and stablecoin issuers are economically equivalent to banks. The issuers sell stablecoin tokens and invest the proceeds into reserves, like Treasuries or commercial paper. 1 However, stablecoins do not pay interest, unlike bank deposits. Consequently, stablecoin holders should demand compensation for run risk when there is a chance the stablecoin issuer’s reserves prove too illiquid to keep its peg. How do stablecoins issuers compensate their borrowers for run risk to keep the price fixed at $1?

In a recent paper (Gorton et al. 2022a), we resolve the puzzle of how can stablecoins with nontrivial run risk keep their peg? We start by observing that several cryptocurrency exchanges and decentralised lending platforms allow agents to lend stablecoins to speculators in more volatile cryptocurrencies. We then show that stablecoin holders lend their coins to levered traders, so long as speculative demand is high enough. The lending rates indirectly compensate stablecoin holders for run risk.

We model the leverage-money nexus using a global game within a bank-run model to pin down the probability of a run that depends on the issuer’s reserves and the lending rate that stablecoin holders demand for bearing run risk. Figure 1 shows the schematic of the model, which illustrates the indirect interconnectedness between cryptocurrency speculation and the stablecoin’s underlying reserves.

Figure 1 Model of leverage-money nexus

Figure 1 Model of leverage-money nexus

The model shows that the stablecoin issuer must respond to keep the peg fixed when speculative demand falls. The issuer has two tools. First, it can shift its reserves toward safer assets, like Treasury bills, which provides assurances regarding stability. Second, it can redeem some of its circulating stablecoins, which limits supply and increases lending rates. 

We test the model by combining margin lending rate data for stablecoins with a proxy for speculative cryptocurrency demand. In our sample, the average lending rate for Tether, the largest stablecoin, is roughly 8%. We estimate speculative demand using cryptocurrency derivatives called perpetual futures, which embed leverage and are likely the most liquid type of cryptocurrency derivatives. For a traditional finance future, the future and spot prices converge as the expiration date approaches. This does not happen with perpetual futures. Instead, perpetual futures use a funding rate to keep the spot and future prices linked. If the future trades at a premium to the spot price, the investors long the future must pay a funding rate to investors short the future. Futures funding rates reflect the cost investors face to take leverage and proxy for speculative cryptocurrency demand.

The model predicts a tight relationship between speculative demand and stablecoin lending rates. Intuitively, as demand grows and traders are willing to pay more to borrow stablecoins, the stablecoin lending rate should increase to clear the market. We show that lending rates for stablecoins are tightly linked with our proxy of speculative demand. A one percentage point increase in the future funding rate leads to a 0.2 percentage point increase in the stablecoin lending rate. The relationship is plain in a scatter plot, shown in the left panel of Figure 2.

Figure 2 Speculative demand and stablecoin lending rates

Figure 2 Speculative demand and stablecoin lending rates

Stablecoin issuers can keep their $1 peg by changing the composition of their reserves as speculative demand changes. We confirm this prediction of the model in the right panel of Figure 2, which plots the safe asset share of Tether against the speculative demand proxy. We calculate Tether’s safe asset share using its quarterly disclosures, where we define safe assets as the sum of its reserves held in cash, bank deposits, reverse repurchase agreements, and Treasury bills. When speculative demand is lower, Tether holds more safe assets.

Stablecoin issuers do not provide high-frequency updates about their reserves, and quickly adjusting their safe asset share—say, by shifting from commercial paper to Treasuries—can be difficult. Instead, issuers can buffer demand shocks by issuing redemptions. When the token’s supply falls, lending rates rise, helping to stabilise the coin’s peg. The turmoil in May 2022 following the collapse of TerraUSD follows this playbook. First, the collapse of TerraUSD was a demand shock for levered traders, and Tether’s price fell below its peg. Quickly, however, Tether redeemed billions of its tokens. Lending rates spiked. When the dust settled, Tether returned to its peg.

Although our model implies a causal relationship between speculative demand and lending rates, an obvious concern is that some other unobserved force is driving both variables. We use an instrumental variables approach using Major League Baseball (MLB) data as a robustness test. FTX, a now-defunct but previously large cryptocurrency exchange, announced a sponsorship deal with MLB that placed a prominent FTX advertisement patch on umpire uniforms, among other things. Umpires had not worn advertising patches before; baseball purists were not enthused. While the value of the sponsorship deal is unknown, it was likely hefty, given that FTX’s sponsorship deals with other sports leagues were worth at least $345 million.

We use MLB viewership data for nationally televised games as an instrumental variable for speculative cryptocurrency demand. The identification strategy relies on two assumptions: first, that advertising is effective and that some MLB viewers begin trading more cryptocurrencies after seeing the ads. Second, we assume that cryptocurrency events do not affect the timing or viewership of MLB games — the game schedule is set well in advance of the season. The approach confirms our expectation that cryptocurrency demand drives stablecoin lending rates.


Privately produced money can keep a $1 peg even if it is not no questions asked, but agents will not hold private money unless compensated for its risks. But, still, private money can be a tailwind to boost real economic outcomes, as Yang and Xu (2022) show.

Although the recent crypto winter did not affect the traditional financial system, future turmoil could leave an imprint, as stablecoins remain large and seem poised to grow. We argue a simple channel links the traditional and crypto financial systems: stablecoin reserves. Stablecoin issuers must quickly adjust their reserves to keep their debt trading at par. When speculative demand falls, they can keep their debt trading at par only by moving to a safer portfolio or allowing redemptions. Such reallocation could cause disruptions in the markets they invest in, like the commercial paper market that supplies financing to the real economy.

Authors’ note: The analysis and conclusions set forth are those of the authors and do not indicate concurrency by members of the Board of Governors of the Federal Reserve System, Office of Financial Research, or their staffs.


Azar, P D, G Baughman, F Carapella, J Gerszten, A Lubis, J P Perez-Sangimino, C Scotti, N Swem, A Vardoulakis, A Werman and D E Rappoport (2022), “The Financial Stability Implications of Digital Assets”, Finance and Economics Discussion Series 2022-058.

Gorton, G B and J Zhang (2021), “Taming wildcat stablecoins”, University of Chicago Law Review 90.

Gorton, G B, E C Klee, C P Ross, S Y Ross and A P Vardoulakis (2022a), “Leverage and Stablecoin Pegs”, NBER Working Paper 30796.

Gorton, G B, C P Ross and S Y Ross (2022b), “Making money”, NBER Working Paper 29710.

Holmstrom, B (2015), “Understanding the role of debt in the financial system”, BIS Working Paper No 479.

Uhlig, H (2022), “A Luna-tic Stablecoin Crash”, NBER Working Paper 30256.

Viswanath-Natraj, G and A Chaudhary (2022), “Algorithmic stablecoins and devaluation risk”, VoxEU.org, 13 May.

Yang, H and C Xu (2022), “Historical lessons on the real effects of unstable stablecoins”, VoxEU.org, 3 May.

Yeyati, E L and S Katz (2022), “The Stablecoin Paradox”, VoxEU.org, 15 August.


  1. We focus on so-called collateralised stablecoins that hold reserves, as opposed to algorithmic stablecoins. Collateralised stablecoins constitute the majority of stablecoins by market capitalisation, even before the failure of the largest algorithmic stablecoin TerraUSD in May 2022. Viswanath-Natraj and Chaudhary (2022) and Uhlig (2022) provide detailed studies of TerraUSD’s collapse. The reserves could be either traditional financial assets (commercial paper, reverse repurchase agreements, Treasuries) or crypto-related assets.