Cleveland Fed experiment finds Bitcoin rallies draw new buyers

A Cleveland Fed working paper reports that exposing people to Bitcoin’s recent 12-month gains raised their return expectations and made them more likely to own crypto. The randomized experiment took place in the second quarter of 2025 and appears in the July 2026 paper Do You Even Crypto, Bro? Cryptocurrencies in Household Finance, by Michael Weber, Bernardo Candia, Olivier Coibion and Yuriy Gorodnichenko.

Participants were split into groups and given different pieces of financial information. One group was told Bitcoin returned 14.3% over the prior 12 months. A second group saw a Bitcoin price chart for the same period. Other groups received data on the S&P 500, a GameStop chart, or the Federal Open Market Committee’s inflation outlook. A control group received no extra information.

The Bitcoin treatments changed expectations and stated investment plans. Being told the exact 12-month return raised expected crypto returns for the next year by 3.2 percentage points versus the control group; seeing the price chart raised expected returns by about 1.2 percentage points. The Bitcoin information increased the share respondents said they wanted to allocate to crypto by roughly 2 percentage points from a control-group average allocation of 4.3 percent. Much of that additional allocation would come from checking, savings or cash balances.

When researchers re-surveyed participants later, the information treatments affected reported ownership. Participants who received the return number were 2.41 percentage points more likely to report owning crypto; those who saw the chart were 2.48 percentage points more likely. Combining the two Bitcoin groups produced a pooled increase in ownership probability with a p-value of 0.017. Random assignment of information allowed the authors to link changes in expectations and ownership back to the treatments.

Responses were largest among people with limited knowledge of cryptocurrency. About 40 percent of non-owners said they knew little about crypto and nearly 90 percent of that subgroup declined to give a numerical forecast for expected returns. Those less-informed respondents showed the biggest shifts in expectations and desired allocations after seeing Bitcoin performance. People who had decided crypto was a poor investment changed their allocations much less.

The experiment produced weaker spillovers from other assets. S&P 500 information made some participants more likely to own crypto later. The GameStop chart nudged desired crypto allocations but did not produce a statistically significant change in ownership. The authors identify the Bitcoin treatments as producing the clearest sequence from past performance to higher expected returns and then to greater ownership.

The paper places the experiment in the context of household survey data from the Nielsen Homescan Panel. Reported crypto ownership rose from about 3 percent of respondents in 2021 to roughly 11 percent in 2022 and near 12 percent by mid-2023, fell for a period and returned to around 12 percent as Bitcoin traded above $120,000 in 2025. After controlling for other factors, respondents under 40 were 13 percentage points more likely to own crypto than those over 60; men were about 4 percentage points more likely than women. Higher income, employment and financial wealth correlated with ownership. The authors report that expected returns and perceived risk explain about twice as much of the variation in ownership as detailed demographic characteristics. In 2025, owners expected a 13.8 percent return for crypto over the next year, while non-owners expected about 4.7 percent.

The paper also examines spending. Bitcoin price gains were associated with higher purchases of durable goods among crypto holders. The authors estimate that a household with its entire financial portfolio in crypto would be 1.4 percentage points more likely to buy a durable good in a quarter if Bitcoin doubled in price. The sales increase concentrated in large, occasional items such as computers and refrigerators and faded by the following quarter, in contrast with more persistent consumption links often seen for stock and bond gains.

The document is classified as a Cleveland Fed working paper and the views expressed are those of the authors, not the Federal Reserve Bank of Cleveland or the Federal Reserve System. The authors note that extrapolating recent returns into the future can be a mechanism that contributes to asset-price bubbles. The experiment does not determine whether Bitcoin is overvalued or whether retail buyers are responsible for sustaining rallies.