Monetary tightening initially lowers DeFi stablecoin deposit rates
BDF Paper

Monetary tightening initially lowers DeFi stablecoin deposit rates

Federal Reserve tightening initially lowers stablecoin deposit rates in decentralized finance, Banque de France researchers find. Falling crypto prices reduce borrowing demand faster than arbitrageurs withdraw liquidity toward traditional safe assets.

Deleveraging outweighs interest-rate arbitrage

Andrea Barbon, Jean Barthélemy and Benoit Nguyen study USDC, USDT and DAI in Aave v2 and v3 on Ethereum from January 4, 2021, to January 30, 2026.

Their Bayesian proxy-VAR identifies monetary policy shocks using price changes around Federal Reserve announcements.

A tightening shock raising the two-year Treasury yield by 25 basis points reduces borrowing and supplied liquidity by approximately 10 percent at their four-to-five-month peak response.

Deposit rates initially fall as weaker crypto returns and collateral values depress leverage demand more than liquidity supply contracts.

Arbitrage subsequently gains strength, but the estimated positive long-run rate response is not statistically distinguishable from zero.

Persistent spreads meet costly arbitrage

DeFi deposit rates averaged approximately 100 basis points above the policy rate during January 2021–January 2026, but 160 basis points below it from mid-2022 through end-2023.

Estimated spread half-lives against the secured overnight financing rate are 5 weeks for DAI, 4 for USDC and 2.2 for USDT.

Higher Ethereum transaction fees accompany slower spread decay.

In wallet-level tests, investors proxied as having traditional-market access reduce supplied liquidity by approximately 3–4 percent more after rate increases than other investors.

Disconnect is not immunity

The authors find that monetary policy affects DeFi activity even when deposit rates initially move in reverse.

They caution that the short sample limits conclusions about long-run convergence.

Access to traditional finance is proxied through wallet behavior, not observed directly.

Report an error