A working paper released by the Federal Reserve Bank of Cleveland shows that past price increases in Bitcoin not only influence U.S. households' perceptions of crypto assets but may also directly drive subsequent purchasing behavior. The study found that respondents who saw Bitcoin’s returns over the past 12 months significantly increased their expected allocation to crypto assets, and their likelihood of making actual purchases also rose.
Randomized experiments show an increase in willingness to configure.
This study is based on multiple rounds of surveys, each with a sample size of approximately 15,000 to 25,000 respondents from the Nielsen Homescan Panel. In the second quarter of 2025, the research team introduced a randomized information experiment, presenting different respondents with information on Bitcoin, the S&P 500, GameStop, or inflation forecasts.
Among them, the Bitcoin group saw either a 14.3% return over the past year or a price chart. The results showed that, after being exposed to this information, respondents' expected allocation to crypto assets increased by an average of about 2 percentage points—a 47% increase compared to the control group’s average target allocation of 4.3%.
The study also found that part of this adjustment stemmed from reduced allocations to cash, checking accounts, and savings accounts. At the same time, respondents increased their target allocation to stocks, indicating that this type of information not only boosts interest in crypto assets but also enhances broader appetite for risk assets.
The gap between holders' and non-holders' expectations is significant.
The paper shows a significant gap in expectations of future returns between cryptocurrency asset holders and non-holders. In the third quarter of 2021, respondents who already held cryptocurrency assets expected an average return of 22% over the next year, compared to just 7% for those who did not hold them.
By 2025, expectations for both groups had declined, but the gap remained. The average expected return for holders was 13.8%, compared to 4.7% for non-holders. The study found that the statistical association between return expectations and holding crypto assets is stronger than individual characteristics such as age, income, gender, or wealth.
According to the paper's estimates, for every 1 percentage point increase in respondents' expected return, the probability of holding crypto assets increases by an average of 0.8 percentage points. However, the authors note that this result primarily reflects correlation and cannot be taken to mean that optimistic expectations fully explain the entire process of asset holding.
The actual probability of purchase has also increased.
In subsequent surveys, respondents who saw information about Bitcoin returns were approximately 2.5 percentage points more likely to purchase crypto assets. Prior to the experiment, about 11% of participants held crypto assets. The research team estimated that this information exposure increased the unconditional probability of purchase by approximately 23%.
The author notes that relatively few individuals in the sample actually changed their position across periods; therefore, the two groups of Bitcoin transaction samples were combined to increase statistical power, resulting in a significance level of p=0.017.
Looking at the group results, the most noticeable reaction came from non-holders who previously did not buy cryptocurrency due to insufficient understanding. In contrast, those who already believed cryptocurrency was not a good investment showed smaller changes in behavior.
Price increases may create demand feedback.
The paper argues that this result supports a possible market transmission pathway: price increases raise return expectations, attracting new participants and generating additional demand. The authors describe this as a potential bubble mechanism, but do not conclude that every bitcoin price surge is self-reinforcing.
The study also noted that other Federal Reserve surveys similarly show that the primary purpose of Americans participating in the crypto asset market remains investment, rather than payment.
In addition, the paper examines the impact of Bitcoin price changes on household consumption. The estimates show that if a household’s entire financial assets are allocated to crypto assets, a doubling of Bitcoin’s price increases the probability of purchasing durable goods by 1.4 percentage points, with more pronounced responses for items such as computers and refrigerators, and weaker effects for cars and housing.
The author suggests that this indicates some households may view crypto gains as a one-time windfall rather than a sustained increase in wealth. However, this paper is still a working draft, and its conclusions reflect the author’s personal research findings, not the official policy stance of the Federal Reserve.