Information about Bitcoin’s previous gains can change how much cryptocurrency U.S. households want to own and whether they subsequently buy it, according to a Federal Reserve Bank of Cleveland working paper published on July 14, 2026.
Summary
- Crypto owners expected 22% annual returns, compared with 7% among surveyed people without digital holdings.
- Bitcoin return information raised desired crypto allocations by roughly two percentage points in randomized testing.
- Actual crypto purchases subsequently increased about 2.5 percentage points among households shown Bitcoin performance information.
- Expected returns and perceived risks explained ownership differences better than age, income, gender or wealth.
- Working paper findings represent the authors’ research and do not establish official Federal Reserve policy.
Researchers Michael Weber, Bernardo Candia, Olivier Coibion and Yuriy Gorodnichenko examined recurring surveys containing between 15,000 and 25,000 responses per wave. The surveys drew participants from the Nielsen Homescan Panel and included a randomized information experiment conducted during the second quarter of 2025.
The paper found that households shown Bitcoin’s trailing 12-month return raised their desired crypto allocation by about two percentage points. That represented a 47% increase from the control group’s average desired allocation of 4.3%.
Participants exposed to Bitcoin return information were also about 2.5 percentage points more likely to purchase cryptocurrency in a later survey wave.
Crypto ownership was closely tied to expected returns
The researchers found a wide gap between how owners and nonowners viewed crypto. In the third quarter of 2021, owners who provided forecasts expected an average return of 22% over the following year. Nonowners expected 7%.
Uncertainty was widespread in both groups. About 87% of nonowners selected “don’t know” when asked to estimate crypto’s next-year return. The share among owners was 54%.
The difference remained in 2025, although both groups produced lower forecasts. Owners expected a 13.8% return, while nonowners expected 4.7%, according to the paper’s appendix.
Expected returns also had a stronger statistical relationship with ownership than observable personal characteristics. Each additional percentage point in expected returns was associated with a 0.8-percentage-point increase in the probability of owning cryptocurrency.
The result describes an association rather than proving that optimistic beliefs caused all existing ownership. The randomized experiment provides narrower causal evidence about how specific historical-return information affected later decisions.
Bitcoin performance information changed real purchases
In the 2025 experiment, participants were randomly assigned information about Bitcoin, the S&P 500, GameStop or an inflation forecast. The Bitcoin treatments showed either its 14.3% previous-year return or a chart of its price.
Both Bitcoin treatments raised desired crypto allocations. The increase came partly at the expense of cash, checking and savings accounts. Participants also raised their desired stock allocations, suggesting the information encouraged broader demand for risky assets.
Later survey responses showed an increase of approximately 2.5 percentage points in actual crypto purchases. Before the treatment, roughly 11% of participants held cryptocurrency. The researchers calculated that the treatment raised the unconditional probability of buying crypto by about 23%.
However, the authors noted that relatively few respondents changed their ownership status between survey waves. They pooled the two Bitcoin treatment groups to improve statistical power, producing a reported p-value of 0.017.
The response was strongest among nonowners who previously cited limited knowledge as their reason for avoiding cryptocurrency. Participants who regarded crypto as a poor investment generally showed little response.
Fed study points to a possible price feedback loop
The results support a possible mechanism through which rising prices attract new market participants. Strong historical returns can lift expectations, encourage purchases and potentially create additional demand.
“Positive returns attract new participants, which raises the price further,” the authors wrote. They presented this as a possible bubble mechanism, not a forecast that every Bitcoin rally will become self-reinforcing.
The finding aligns with evidence that investor attention often follows market performance. In related coverage, retail interest began returning after renewed Bitcoin momentum, while periods of falling prices have coincided with lower search activity.
Broader Federal Reserve survey data also show that Americans primarily use cryptocurrency as an investment. As previously reported, crypto activity reached 10% of U.S. adults in 2025, while payment use remained comparatively limited.
Crypto gains affected some household purchases
The paper also examined whether Bitcoin price changes influenced spending. Its estimates indicated that doubling Bitcoin’s price made a household with its entire financial portfolio in crypto 1.4 percentage points more likely to buy a durable good.
With durable purchases occurring among about 20% of surveyed households, that represented a 7% increase relative to the unconditional probability. The estimated response was strongest for products such as computers and refrigerators and weaker for cars and homes.
The researchers found little corresponding change in nondurable spending. They interpreted this pattern as evidence that households may treat crypto gains more like lottery winnings than a permanent increase in wealth.
That comparison remains the authors’ interpretation of their statistical results. It does not establish how every crypto investor views gains or how households would react under different market conditions.
The paper is preliminary research circulated for discussion. The Cleveland Fed states that its working papers may not receive the same formal editorial review as official publications. Its findings represent the authors and do not establish a position of the Cleveland Fed or the Federal Reserve System.
