We exploit the May–December 2021 Bitcoin crash to study how large, negative wealth shocks in crypto markets propagate through the balance-sheet choices of U.S. households. Linking anonymized brokerage and depository records to a nationally representative panel of 2.3 million households, we show that the median crypto-exposed household lost 48% of its liquid net worth when Bitcoin fell from $63k to $33k.

The crash triggered an immediate and persistent reallocation of portfolio shares that traditional frictionless models fail to explain. Within six weeks, households reduced their aggregate crypto weight from 14% to 6%, but—critically—did not return the proceeds to cash or index funds. Instead, they:

(i) increased margin borrowing against equity positions by 18%;

(ii) shifted toward high-beta tech equities and call-option strategies; and

(iii) doubled allocations to leveraged ETPs that offer synthetic crypto exposure.

Real-side spillovers (difference-in-differences design):

Using a difference-in-differences design that contrasts counties in the top and bottom terciles of pre-crash crypto wealth, we document real-side spillovers: zip codes with high crypto penetration experienced a 9% relative decline in auto-loan origination and a 4% rise in 30-day credit-card delinquencies over the following nine months.

Heterogeneity analysis:

Heterogeneity analyses reveal that the reallocation was strongest among households with low traditional-asset wealth and high leverage, suggesting that crypto losses were treated as disposable risk capital rather than as part of a diversified nest egg.

Survey evidence:

Finally, survey evidence indicates that the crash revised subjective return expectations downward by 30 percentage points, yet left beliefs about crypto’s long-run upside intact—implying that the observed portfolio changes reflect a risk-budgeting response rather than a permanent exit from digital assets.

Our results imply that the microstructure of household portfolios now contains a crypto-feedback channel: negative shocks to the riskiest asset class are amplified by simultaneous increases in leverage and factor-tilt exposures elsewhere in the balance sheet, raising the elasticity of household consumption to future crypto volatility.