Cleveland Fed Experiment Finds Bitcoin Gains Can Pull New Buyers Into Crypto

Bitcoin gains first changed return expectations

Participants who received Bitcoin’s numerical return increased their expected cryptocurrency return for the following year by 3.2 percentage points relative to the control group. Respondents shown the Bitcoin price chart raised their expected return by about 1.2 percentage points.

Those changes then affected intended portfolios. The Bitcoin treatments increased respondents’ desired crypto allocation by about 2 percentage points from a 4.3% average in the control group, representing a substantial increase relative to the allocation respondents otherwise wanted to hold.

Respondents largely made room for the additional crypto exposure by reducing desired allocations to cash, checking accounts and savings accounts. They also increased planned stock holdings.

Researchers later surveyed participants again to test whether changed expectations and portfolio preferences were followed by changes in actual reported ownership.

About 11% of respondents owned crypto before the experiment, making the treatment-related increase roughly 23% relative to the starting ownership rate.

The ownership analysis covered 5,352 respondents across the second through fourth quarters of 2025 and controlled for whether participants owned cryptocurrency before receiving the information. Ownership was measured through respondents’ own survey answers rather than transaction records.

Because relatively few respondents changed ownership status between survey waves, researchers combined the two Bitcoin treatment groups for part of the analysis. The pooled result was statistically significant, with a p-value of 0.017.

The randomized design distinguishes the findings from a simple observation that existing crypto holders tend to expect higher returns. By assigning information treatments randomly, the researchers could connect changes in expectations and later ownership to the information participants received.

The experiment does not establish that every Bitcoin rally will produce the same amount of new demand or quantify how much resulting purchases could move prices. It also does not determine whether Bitcoin is overvalued or prove that new retail buyers sustain market rallies.

Less-informed investors reacted most strongly

The Bitcoin treatments had their strongest effect among participants who said they avoided cryptocurrency because they did not know enough about it. People who already regarded crypto as a bad investment changed their desired allocations much less, and the combined Bitcoin treatments produced no statistically significant ownership effect for that group.

Broader household survey data showed a similar divide based on familiarity. About 40% of non-owners said they knew little about cryptocurrency, and close to 90% of that group would not provide a numerical forecast for expected crypto returns.

Those respondents with weaker prior knowledge were among the most responsive after receiving Bitcoin information, indicating that recent positive performance had a stronger effect on people who had not yet formed firm views than on respondents with established negative opinions.

The researchers also found some spillover from other risk assets. Participants shown S&P 500 information became more likely to own cryptocurrency later, though stock-return information did not change desired portfolio allocations. Participants shown a GameStop chart increased their desired crypto allocation, but that treatment did not produce a statistically significant difference in later ownership.

Bitcoin produced the clearest sequence from new information to higher expected crypto returns, higher desired exposure and greater subsequent ownership.

“Positive returns attract new participants, which raises the price further,” the authors wrote. “The experience of high returns in the past does not seem to lead individuals to expect any mean reversion; extrapolation of past returns into future returns seems to be the rule.”

The authors said that pattern represents one mechanism through which speculative bubbles can develop, with recent gains increasing expected future returns and attracting additional participants.

Beliefs explained ownership better than demographics

The research also used Nielsen Homescan Panel data tracking tens of thousands of U.S. households and their spending behavior to examine broader differences between crypto owners and non-owners over time.

Reported cryptocurrency ownership rose from roughly 3% of respondents in 2021 to about 11% in 2022 and around 12% by mid-2023. Ownership later fell by several percentage points before recovering to roughly 12% as Bitcoin traded above $120,000 in 2025.

Demographic differences were substantial but did not explain participation as strongly as investors’ beliefs about returns and risk.

Higher income, employment and greater financial wealth were also associated with cryptocurrency ownership. However, researchers found that expected returns and perceived risk had around twice the explanatory power of detailed demographic characteristics when accounting for who owned crypto.

The difference between owners and non-owners was particularly visible in their return expectations.

Researchers said the gap in beliefs between owners and non-owners was much less pronounced for stocks, bonds and gold.

Bitcoin wealth gains also affected household purchases

The same research linked Bitcoin gains to changes in spending among households already exposed to cryptocurrency, particularly purchases of durable goods.

Researchers estimated that a hypothetical household holding its entire financial portfolio in crypto would be 1.4 percentage points more likely to buy a durable good during the quarter if Bitcoin’s price doubled.

The strongest effects appeared in large, occasional purchases such as computers and refrigerators. Researchers found less evidence of a sustained increase in recurring spending on food, utilities and other nondurable goods, and the spending effect faded by the following quarter.

That pattern differed from traditional financial wealth, where gains in stocks and bonds showed a stronger relationship with routine consumption. The authors compared the cryptocurrency response with temporary windfalls, including gambling proceeds, where gains are more likely to finance one-time purchases than permanently increase household spending.

The household findings identify two separate behavioral channels tied to rising Bitcoin prices: existing holders can experience higher paper wealth that affects durable-goods purchases, while potential investors can revise return expectations, change desired portfolios and become more likely to enter the cryptocurrency market.