We no longer support this browser. Using a supported browser will provide a better experience.

Please update your browser.

Close browser message

RESEARCH The Dynamics and Demographics of U.S. Household Crypto-Asset Use

U.S. households’ involvement in crypto-assets (hereafter, crypto) rose sharply during the COVID-19 pandemic alongside a substantial increase in the overall personal savings rate. Five years ago, only a tiny fraction of individuals held crypto. As of mid-2022, almost 15 percent of individuals had conducted transfers into crypto accounts, according to our data. The trend has potential implications for the health of household balance sheets, given market volatility and uncertainty of how use of crypto-assets may evolve.

This report uses de-identified data covering a sample of nearly 5 million active checking account customers, over 600 thousand of which have conducted transfers to crypto accounts. Importantly, we link the dynamics of such transfers with demographic indicators, enabling analysis of heterogeneity across income, gender, and racial groups.

Our findings can be used to assess differential effects of the rise of crypto investing to-date and further extend our understanding of how financial trend-chasing behavior plays out in the real world. In one of the report’s main conclusions, we estimate that lower-income individuals have fared worse—buying later and at higher prices on average—than those with higher incomes. While our data only show transfers into the crypto ecosystem and not the direct purchases of crypto-assets we estimate that the median investor in crypto has probably experienced substantially negative investment returns in percentage terms. Notably, the dollar values involved have been quite small for most.

In our investigation into how crypto fits into households’ financial health, we organize the analysis around following questions and findings:

Research questions

  1. How has crypto engagement evolved over time?
  2. What are the demographics of individuals that are using crypto-assets?
  3. How much financial risk have retail market participants taken in crypto?

Findings

  1. Most crypto users made their first transactions during spikes in crypto-asset prices.
  2. Usage of crypto is broader and deeper for men, Asian individuals, and younger individuals with higher incomes.
  3. Crypto holdings for most individuals are relatively small—as median flows equal less than one week’s worth of take-home pay—but almost 15 percent of users have net transfers of over one month’s worth of pay to crypto accounts.
  4. Most individuals who transferred money to crypto accounts did so when crypto-asset prices were significantly higher than recent levels, and those with lower incomes likely made purchases at elevated prices relative to higher earners.

Finding One: Most crypto users made their first transactions during spikes in crypto-asset prices.

The share of the population that have ever transferred funds into a crypto-related account tripled during the COVID-19 pandemic, rising from a cumulative 3 percent prior to 2020 to 13 percent as of June 2022. See box below for background on how we track crypto-related flows. The adoption of crypto accounts (defined by the first observed crypto transaction) and the volume of transfers have come in concentrated episodes that coincide with sharp increases in the price of bitcoin. The majority of new crypto users in our sample, from 2015 to 2022, made their first transactions in a set of days spanning less than five months, all of which coincide with a trailing monthly price change exceeding 25 percent.  

Data and Methodology

This research relies on de-identified data on Chase checking account transactions vis-à-vis dozens of retail crypto platforms. We count transfers to crypto brokers as “flows” into crypto, and vice versa for transfers from those platforms into Chase accounts. Our main analytical sample for this report considers a universe of active checking account customers for which we have a significant and sustained stream of activity, characterized by at least 5 transactions per month with total inflows and outflows each over $1,000 per month. This helps center our analytics on a set of individuals for which we see a representative amount of their overall financial activity. Measures of the extent of crypto involvement in this report are relative to this population.


Figure 1: Surges in the crypto user base coincide with significant spikes in bitcoin prices.

The intensity of transfer activity at certain points in time, correlated with price movements, suggests herd-like behavior driving a notable share of individuals’ overall transactions with crypto accounts. Investors using traditional investment accounts exhibit similar dynamics—inflows are stronger just after significant increases in U.S. stock prices—but crypto transactions have been more concentrated.i

Consistent with crypto-assets being a new entrant to the retail investment landscape, in aggregate U.S. households have been net purchasers of crypto-assets over the past several years. In our sample, the ratio of transfer to crypto accounts to money flowing back into traditional checking accounts was 2 to 1 from 2017 to mid-2022. However, the relative flows had shifted to nearly balanced after the declines in bitcoin prices in May and June of 2022 (Figure 2), as transfers to crypto assets fell and outflows remained elevated. We view the rise and fall of crypto use since the onset of COVID as consistent with the joint relationship between retail flows and market prices seen in prior research. Additionally, the trend in crypto flows also tracks dynamics of household savings, which spiked to historic highs early in the pandemic but has begun to reverse.ii

Figure 2: For most of the past several years, retail investor outflows to crypto accounts far exceeded inflows from those accounts, but the net direction of transfers became more balanced alongside crypto price declines in the first half of 2022.

Finding Two: Usage of crypto is broader and deeper for men, Asian individuals, and younger individuals with higher incomes.

Prior analyses using information from surveys has shed light on the demographics of crypto usage. Banking transactions data provide a check that individuals’ survey responses are consistent with financial activity. Overall, our results on the gender and race makeup of crypto users are broadly consistent with data from the Federal Reserve’s Survey of Household Economic Decisionmaking and separate studies from Pew Research Center.iii

Age and gender

In our sample, crypto usage is more prominent among younger individuals—20 percent for millennials, 11 percent for Generation X, and 4 percent for baby boomers (see Figure 3). Men are more actively engaged in each age group. Across the full sample, men are about twice as likely to have transferred money into or out of crypto accounts.

Figure 3: Use of crypto accounts is most common for millennials and men.

Among crypto users, men are also more deeply involved in crypto than women, as measured by the gross amount of transfers into and out of crypto accounts vis-à-vis traditional checking accounts. The median total gross transfers for men is approximately $1,000 for men and $400 for women. We investigate the depth of financial risk taken by demographic groups in more detail in Finding 3. 

Race

We find additional variation in the demographic makeup of crypto users by race. To isolate differences in race, we focus on millennials—which make up the plurality of crypto users—and look at engagement within income brackets. For millennials around the median income in our sample, Asian individuals have the highest involvement rate at 27 percent.iv The crypto user shares for Black and Hispanic individuals in our data were approximately equal at 21 percent, and White individuals came in at 20 percent.

Crypto involvement increases moderately with higher incomes across race groups, as shown in Figure 4. For reference, we plot these figures relative to the share of individuals that have traditional investment brokerage accounts. The share with each type of account is increasing in income, although the differences between those with higher- and lower-incomes is much smaller in percentage point terms for crypto.

Figure 4: Use of both crypto accounts and traditional investments generally increases with income across racial groups.

Finding Three: Crypto holdings for most individuals are relatively small—as median flows equal less than one week’s worth of take-home pay—but almost 15 percent of users have net transfers of over one month’s worth of pay to crypto accounts.

Most investors in crypto have only small holdings. Cumulating transfers at the individual level, the median gross amount transferred to crypto accounts over the period 2015 through the first half of 2022 was approximately $620. The level of crypto engagement—as measured by raw dollar value of transfers and when scaled by income—is generally higher for higher income individuals (see Figure 5).v The distribution of crypto risk exposure is consistent with the view that higher income individuals were more willing and able to bear crypto price risk.vi

Figure 5: Higher income individuals have transferred more money into crypto accounts, measured in dollar values and when scaled by weeks’ worth of take-home-pay.

Moving beyond median involvement, we look to the share of individuals in different groups at different levels of (scaled) financial risk. This perspective can be useful to understand the extent of possible financial distress related to losses in crypto markets. As depicted in Figure 6 below, 11 percent of individuals in the first income quartile and 15 percent of the fourth quartile have transferred over one month’s worth of income into crypto accounts. Raising the bar to higher degree of exposure, three months of income, decreases the shares to 4 and 6 percent, respectively.

Figure 6: About 15 percent of individuals have transferred over 1 month’s worth of income into crypto accounts, with a somewhat higher share for high-income individuals.

Additional context on households’ finances—namely, other indicators of wealth—may reduce spillover risks that could emanate from crypto market volatility, all else equal. Our data show that liquid balances and investment brokerage usage are higher for individuals with elevated crypto exposure. Median liquidity, measured by weeks’ worth of pay held in checking and savings accounts, is 50 percent higher for individuals with more than one month’s worth of pay in crypto in the lowest income quartile. Similarly, those individuals were also more likely to have additional financial wealth in brokerage accounts. This suggests that individuals with high levels of crypto exposure, on balance, may also have greater ability to bear market risk.

Finding Four: Most individuals who transferred money to crypto accounts did so when crypto-asset prices were significantly higher than recent levels, and those with lower incomes likely made purchases at elevated prices relative to higher earners.

With crypto asset prices showing substantial volatility over the previous five years, there is potential for a wide range of average transaction prices and therefore investment returns. We estimate the impact of differential investment timing using the methodology described in the box below.

Looking at the distribution of average purchase prices—measured at the individual level—across our sample, we see substantially different median implied purchase prices to crypto by income group. The typical individual transferred money to crypto accounts when bitcoin was trading at a transaction-amount-weighted average level of purchase price of $43,900. The figure is $45,400 for individuals from the lowest income quartile and $42,400 for the highest, among millennials. Since some of these transfers may have been destined for investment in less volatile cryptocurrencies—or remained denominated in USD—our estimated purchase price exercise may inflate implied investment losses.

Figure 7: Individuals with lower incomes transferred money into crypto accounts when prices were higher, suggesting lower investment returns.

Estimated crypto purchase prices also exhibit a modest premium paid by non-college graduates. These individuals bought at purchase prices about 2 to 4 percent higher than those with college and graduate degrees, respectively. Controlling for race, generation, gender, observed income, there is still about an $1,100 discount (approximately 2.5 percent) for grad school versus high school educated individuals. These are fairly muted differences compared with the overall income group disparities.

Figure 8: Individuals with lower education had somewhat higher implied crypto purchase prices.

Purchase price by education level

 

Median purchase price

Discount (to No college price)

No college

44,500

0.0%

College degree

43,700

1.8%

Graduate degree

42,800

3.8%


Note: Purchase prices are computed for each individual as the transfer size-weighted average price of bitcoin on the date of the transfer.

Measuring Crypto Investment Purchase Prices

Since we do not directly observe the price of crypto trades, or the actual cryptocurrency involved, we rely on the timing of a transaction and the predominance of bitcoin in the market to estimate transaction prices. We use the closing price of bitcoin on the date of the transfer as an estimate of the purchase price. For each individual then, we can arrive at a flow-weighted average purchase price estimate via the following equation, where p̅is the average purchase price of person i, Fi is the total amount of the person’s gross transfers to crypto, fit is the dollar value of each transfer at time t, and BTCt is the associated daily bitcoin closing price.

equation showing that p̅i, the average purchase price of person i, equals one over Fi, which is the total amount of the person’s gross transfers to crypto, multiplied by the sum of that person's transfers at time t, f subscript it, the dollar value of each transfer multiplied by BTCt, the associated daily bitcoin closing price

 

With prices trading near $20,000 over the past several months, our average transaction price measurements imply that the majority of U.S. individuals have faced losses on their crypto investments. While a number of individuals have transferred money out of crypto, especially around the price declines in May and June 2022, only 13 percent of individuals have transferred out as much money as they transferred into crypto accounts. Figure 9 shows the share of individuals in our sample with implied average crypto purchase prices across the range of bitcoin prices. Less than 20 percent of individuals that transferred money into crypto accounts did so when bitcoin was below the recent trading range sub-$20,000 as of November 2022. Over half of individuals made their average crypto transfers when prices were above $40,000, suggesting significant investment losses for that group.

Figure 9: The majority of U.S. crypto users made transfers to crypto accounts when prices were at substantially elevated values relative to current market levels.

Finding Five: Implications

Crypto use has increased quickly among U.S. households, but the amounts invested for most people are relatively small. Despite the relatively modest overall exposure, a small subset of the population may have substantial financial risk to further crypto market declines. About 15 percent of crypto users have transferred on net over one month’s worth of take-home pay into crypto accounts, making them more vulnerable to the greater than 50 percent decline in crypto-asset prices observed since their peaks in late 2021.vii

The timing of transfers around significant price spikes is characteristic of herd-like behavior. A wide range of U.S. households transferred money into crypto accounts when those assets were trading near their highest levels. Using bitcoin prices around the time of transfers to crypto accounts as a proxy for investment price, we find that lower income households bought crypto at substantially higher prices. The majority of U.S. households were likely facing significant losses in percentage terms at cryptocurrency prices prevailing in late-2022.

Our analysis suggests that policymakers concerned with investor protection should consider the wide range of financial sophistication across the user base. The extent of crypto usage—in terms of the number of individuals involved and the depth of involvement—is higher for those with higher incomes. However, compared with investors in traditional investment accounts, the median crypto user is more likely to come from lower rungs of the income ladder and is more likely to be young and male. Crypto-assets may therefore merit a differentiated policy approach—compared with the existing architecture for traditional markets (e.g., stocks and bonds)—to effectively protect investors and the economy.





Authors

Chris Wheat

President

George Eckerd

Financial Markets Research Lead