U.S. Deposit-Rate Model Uses 324 Observations for Stablecoin Estimate

A U.S. model links higher average bank deposit rates to slower stablecoin market-cap growth, with 324 effective observations behind its 10-week estimate in a 339-observation weekly dataset.

The broader dataset contains weekly U.S. observations from January 2019 through June 2025. The 10-week specification uses 324 effective observations. It compares movements in average deposit rates with cumulative market-cap growth across the next 10 weeks, giving the estimate a narrower sample than the full dataset.

In the ordinary least squares model, a one-basis-point increase in average deposit rates was associated with 2.8 percentage points less stablecoin market-cap growth over that period. A basis point equals 0.01 percentage point. The sample’s average stablecoin growth was 15%; applying the estimated difference implies growth of about 12%. That is a slower rate of expansion, not an absolute decline in market capitalization.

An instrumental-variable model gave a larger estimate: a one-basis-point rate increase was associated with 9.1 percentage points less growth across the broader stablecoin market over 10 weeks. The 2.8-point and 9.1-point figures come from different statistical methods, rather than successive revisions to one estimate.

The ordinary least squares model also produced issuer-level estimates. It linked higher average deposit rates with 4.8 percentage points less 10-week growth for Circle’s USD Coin (USDC) and 1.9 percentage points less growth for Tether’s USDT. Those figures are rounded estimates for the same 10-week period.

The market-cap series shows reserve-backed stablecoins below $5 billion in 2019 and above $220 billion by mid-2025. This tracks the market’s size over time; it does not trace transfers from individual bank accounts into stablecoins. The estimates therefore do not establish that aggregate bank deposits or bank lending fell.

The full data window spans six and a half years, while the 10-week result uses the narrower effective sample. The clarified count and model distinction add detail to earlier coverage of the 2.8-point estimate.