A tax strategy known as “buy, borrow, die” is creating an unseen source of credit risk in decentralised finance (DeFi), as investors borrow against appreciated cryptocurrency rather than selling it and triggering a taxable gain.
Consider an investor who bought Ethereum for $1,000 and watched its value rise to $4,000. Selling a quarter of the holding would raise $1,000, but under United States tax rules for digital assets held as investments, that transaction would realise a $750 gain.
Instead, the investor could deposit the full Ethereum holding into a DeFi lending protocol and borrow $1,000 in a dollar-linked stablecoin. The loan is not treated as taxable income, while the investor retains exposure to any further rise in the price of Ethereum and gains funds that can be spent or converted into dollars.
The arrangement, however, leaves the lending pool exposed if the value of the collateral falls. The original loan-to-value ratio is 25%, based on $1,000 of debt against $4,000 of Ethereum. If Ethereum drops to $2,000, that ratio rises to 50%, while interest charged on the loan pushes it higher still.
Once the ratio breaches a protocol’s limit, the collateral can be liquidated. An outside trader may repay part of the debt and take possession of some of the Ethereum at a discount.
The borrower has delayed a taxable sale, but the pool has inherited the consequences of the collateral’s price, the size of the loan and the borrower’s willingness to take action before liquidation. For other users, the borrower may be known only through a wallet address made up of letters and numbers.
Lisa De Simone of the University of Texas at Austin, Peiyi Jin of the National University of Singapore and Daniel Rabetti of NUS examined the link between tax planning and DeFi credit risk in a working paper.
They studied Venus, a DeFi lending protocol on BNB Smart Chain that allowed users to pledge cryptocurrency and borrow other tokens under rules enforced by smart contracts. The research covered 12 November 2020 to 31 July 2022 and examined the 15 largest tokens on Venus.
About 13 million transactions produced 1.36 million daily borrower observations. A wallet could therefore appear once for every day on which it was active. Roughly 3% of traders experienced what the researchers defined as a default.
In traditional lending, default generally means a missed payment. In the Venus study, a borrower was classified as being in default when the loan stayed above the protocol’s 60% loan-to-value limit for at least seven days without the borrower subsequently borrowing or depositing funds.
The researchers calculated total defaulted debt at $133.34m. That figure adds the outstanding debt for every day on which a default continued, meaning one troubled loan could appear on several dates. It therefore measures accumulated daily exposure rather than the unique amount of principal lost in a single event.
The traditional “buy, borrow, die” approach involves purchasing an asset, allowing it to appreciate and borrowing against it to fund spending without selling and realising the gain. In the US, borrowing can defer capital-gains tax for years. Estate rules may also reset the tax basis of an asset when it passes to heirs, reducing the gain accumulated during the original owner’s lifetime.
Historically, the strategy was largely available to wealthy investors through private banks. A bank could assess a client’s wider finances, decide how much to lend and negotiate terms, leaving time to respond to a market downturn before collateral had to be sold.
DeFi replaces that personal relationship with code. A protocol assesses the assets in a wallet and applies the same collateral rules to users, regardless of their financial circumstances or tax position.
The model widens access but relies on overcollateralisation. Under the Venus configuration examined in the paper, $10,000 of approved collateral could support as much as $6,000 of debt. A borrower using the full amount had little protection against a fall in the collateral’s value, while a $2,000 loan against the same assets provided a much larger cushion.
Market prices were supplied to the protocol and monitored continuously. If a collateral ratio breached the limit, a liquidator could repay part of the debt and take the collateral at a discount, receiving a reward for restoring the account. The mechanism is intended to protect lenders before collateral falls below the outstanding debt, although a rapid sell-off, a thin market or blockchain congestion can prevent liquidations from happening quickly or efficiently.
Tax incentives can make the problem more difficult. A borrower trying to avoid a taxable sale may be reluctant to trade, repay the loan or sell part of an appreciated holding. That reluctance may be particularly strong when the asset has generated a large paper gain or when waiting longer could move the holding towards the lower US long-term capital-gains rate.
Stablecoins make the strategy especially useful. Dollar-pegged tokens provide spending power while volatile cryptocurrency remains locked as collateral. Investors can borrow USDT or USDC against Ethereum or another token and use the funds elsewhere.
The protocol may judge an account to be healthy when the loan is opened, but it cannot see that the borrower bought Ethereum at a much lower price, has a large gain that would be realised by selling or believes that holding the asset for several more months is financially important.
To distinguish tax-driven behaviour from normal cryptocurrency-market activity, the researchers used the Infrastructure Investment and Jobs Act, enacted on 15 November 2021. Section 80603 expanded information-reporting requirements for brokers handling digital assets, giving traders reason to expect that more of their activity would eventually be reported to the Internal Revenue Service.
The provision changed the expected level of third-party reporting for likely US taxpayers, while international users were not affected in the same way. The final reporting system took years to implement. Custodial brokers began reporting gross proceeds from covered sales and exchanges completed from 1 January 2025 on Form 1099-DA, while Internal Revenue Service broker rules added basis reporting for certain transactions completed from 1 January 2026.
Those rules apply to firms that take possession of customers’ assets. Non-custodial DeFi services are outside their current scope.
For the research, the significance was what traders believed in November 2021, when the law made future reporting appear more certain. The authors compared behaviour around the enactment date, well before the final rules took effect, allowing them to study a reaction to expected visibility rather than to tax forms already being issued.
The blockchain does not identify nationality or tax residence, so the researchers inferred which wallets might be linked to US users. They examined activity concentrated during US business hours, unusual behaviour on holidays observed only in the US and holdings of dollar stablecoins under US oversight. They also used a stricter definition combining several of those indicators, acknowledging that each measure could misclassify users.
The comparison involved two groups using the same protocol under the same market conditions. Both faced identical token prices and Venus rules, but the probable US group had a stronger reason to respond to the reporting provision.
The researchers found that US-linked borrowers became 24.5% less likely than international users to trade assets after the law was enacted. Among borrowers using stablecoin debt, trading fell by a further 23%, consistent with the argument that stablecoins gave investors immediate spending power while appreciated collateral remained pledged.
In the paper, “liquidity” refers specifically to the daily probability that a borrower traded any asset. It does not mean exchange depth, bid-ask spreads or the cost of selling a large position. Instead, it measures how active a borrower’s wallet was and whether appreciated assets remained locked in the protocol.
The pattern was stronger among borrowers with larger gains and higher loan-to-value ratios. Trading declined in December, particularly during the final week, when investors often delay gains until a new tax year. Activity then rose after holdings passed the one-year point associated with lower US long-term capital-gains rates.
The authors estimate that US borrowers in the sample deferred an average of $3,357.42 in capital-gains tax each year, equivalent to about 17% of their trading portfolios during the period. The figure assumes the inferred wallets belonged to US taxpayers, reconstructs portfolios from blockchain activity and applies the relevant tax brackets, so it is intended as a rough estimate of scale across the sample.
The study then examined whether reduced trading affected loan performance. Borrowers who rarely traded could leave risky accounts open for longer, miss opportunities to repay debt or fail to add collateral before the loan exceeded its permitted ratio.
The researchers recognised that causality could operate in the opposite direction: a borrower might default, abandon a wallet and stop trading as a result. They therefore used the law-related reduction in activity among US-linked borrowers to isolate a fall in trading that originated outside the protocol. Their approach used an instrumental-variable design.
They estimate that a 1% increase in tax-induced illiquidity was associated with an 11.2% increase in defaulted accounts and a 39.6% increase in the value of defaulted loans. A one-standard-deviation increase corresponded to about $350 more defaulted debt per borrower, or 2.7 times the baseline value in the model.
The large percentages apply to the tax-sensitive borrowers whose behaviour changed in response to the reporting event, known in economics as compliers. They should not be treated as a universal multiplier for every DeFi loan, but instead show how borrower behaviour can affect credit outcomes.
The findings highlight a limitation of automated lending. Smart contracts can see collateral prices, debt balances, interest owed and liquidation thresholds, but they do not know the purchase price of the collateral or the owner’s tax incentive.
Two wallets with identical Ethereum holdings and identical loans may therefore look exactly the same to Venus, even if one owner is comfortable selling and the other is determined to avoid doing so.
That creates a borrower-selection problem. Overcollateralisation protects a pool against ordinary price movements, but borrowers most attached to appreciated assets may keep loans open for longer and trade less as their safety margin narrows. Risk can become concentrated among users whose motives the protocol cannot measure.
If liquidation succeeds, an outside participant repays debt and removes collateral before lenders suffer a shortfall. If it fails, losses may be absorbed by protocol reserves, token holders or users who supplied assets to the pool, depending on how the system allocates losses. A personal tax decision can therefore become a financial outcome shared across the lending market.
The research was limited to one protocol during the cryptocurrency boom and crash from 2020 to 2022. It inferred US residence from wallet behaviour and used a specialised definition of default based on unresolved high loan-to-value accounts. The researchers also found that volatile collateral and weaknesses in liquidation contributed to troubled Venus loans, meaning tax sensitivity was one factor among several.
A protocol with deeper liquidity, different collateral limits or more professional liquidators could produce different results from Venus during the period studied. Even so, the paper provides a rare public view of collateral, debt, borrower activity, liquidation and wallet abandonment at an individual level.
DeFi has brought a borrowing strategy once associated with private banking onto a public blockchain and made it available to a much broader group of investors. Automating the loan officer, however, does not remove the human motives behind borrowing. Tax bills, attachment to appreciated assets and reluctance to sell can still influence behaviourand ultimately affect everyone funding the pool.
Ethereum was down 2.79% over the previous 24 hours and ranked second by market capitalisation at the time of publication.
Andjela, who has a classical education and previously covered politics, entered the cryptocurrency industry in 2018. Gino Matos is a law-school graduate and journalist with six years of experience in the cryptocurrency industry, with particular expertise in the Brazilian blockchain sector.
CryptoSlate says it may use artificial-intelligence tools to support research, editing and production, while its journalism remains human-led and its editorial team retains responsibility for published material. It directs readers to its full AI-usage disclaimer.
The writers’ views are their own and do not represent CryptoSlate’s position. The publication says its material is not investment advice and that it does not endorse any project mentioned or linked in the article. Cryptocurrency buying and trading are described as high-risk activities, and readers are advised to carry out their own research before acting. CryptoSlate accepts no responsibility for trading losses and directs readers to its company disclaimers.
The publication also promotes a service called “Follow the signal”, offering market-moving updates.
