Wealthy Tax Avoidance: DeFi Lending Pools Take the Blame

A risk tracing report from a $130 million on-chain default exposure.

Written by: Andjela Radmilac
Compiled by: Saoirse, Foresight News

Imagine an investor buys Ethereum for $1,000, and later the price rises to $4,000. Now, they want to cash out $1,000. If they sell a quarter of their Ethereum, they can get cash, but according to U.S. tax rules on investment digital assets, they will incur a capital gain of $750.

However, DeFi offers an alternative solution. Holders can deposit all their Ethereum into a lending protocol as collateral and borrow a stablecoin pegged to the dollar worth $1,000.

This loan is not considered taxable income, and the Ethereum can still benefit from future price increases; holders can obtain funds that are consumable and convertible to fiat without selling their original assets.

While this method can save holders a significant amount in taxes, it also creates fragile risk vulnerabilities. The initial collateralization ratio for this $1,000 debt is 25% (collateral value of $4,000); if the price of Ethereum drops to $2,000, the loan-to-value ratio will double to 50%, and with ongoing interest on the debt, this ratio will increase further.

When the ratio exceeds the protocol's set limit, the code will trigger collateral liquidation: external traders can repay part of the loan and take some Ethereum at a discount.

Although borrowers delay taxable sales, the lending pool must bear the risks associated with collateral price fluctuations, debt scale, and whether borrowers are willing to proactively manage their positions before liquidation.

Individual tax decisions effectively become part of a shared credit market funded by other users, while most lenders only see a string of wallet addresses.

Researchers Lisa De Simone from the University of Texas at Austin, Peiyi Jin from the National University of Singapore, and Daniel Rabetti from the National University of Singapore have studied this relationship in a working paper on tax planning and DeFi credit risk.

The study focuses on the DeFi lending protocol Venus on the BNB Smart Chain, which executes rules via smart contracts, allowing users to pledge crypto assets and borrow other tokens.

The research sample period spans from November 12, 2020, to July 31, 2022, covering 15 mainstream tokens on the Venus platform. The study includes approximately 13 million transactions, generating 1.36 million daily observations of borrowers (the same active wallet generates one record daily), with about 3% of traders defaulting as defined in the paper.

Defaults in the DeFi context differ from mortgage delinquencies and require clarification. The paper defines a default as a loan-to-value ratio that remains above the 60% threshold of the Venus protocol for at least 7 days, during which the borrower does not make additional deposits or take out new loans; the cumulative default debt exposure totals $133.34 million. A loan at risk will continue to be counted in multiple statistical dates, so this total represents cumulative daily risk exposure, not the actual principal loss in a single event.

Billionaire Strategies Enter Ordinary Wallets

This strategy's traditional version is called "buy, borrow, and die." Investors purchase assets, wait for appreciation, then use the assets as collateral to borrow funds for living expenses, without ever selling the assets or realizing gains.

Continuous borrowing can defer capital gains taxes for years; U.S. estate laws allow heirs to reset the tax basis of inherited assets, wiping out most of the tax burden accumulated during the original holder's ownership.

In the past, this was a strategy exclusive to the wealthy: private banks only served high-net-worth clients with sufficient collateral and strong downside resistance. Banks could review clients' overall financial situations, determine lending limits, negotiate terms, and mitigate risks before collateral was forced to be liquidated.

DeFi has completely transformed this highly manual due diligence process into code. Software does not need to assess the overall qualifications of clients; it simply reads the assets in the wallet and applies the same collateral rules to all users.

The service threshold has been significantly lowered, but the cost is that the entire system is built on over-collateralization: borrowers must pledge assets worth more than the loan from the outset.

According to the parameters described in the paper for Venus: a qualified collateral worth $10,000 can borrow up to $6,000 in debt. Borrowing the full amount leaves almost no buffer against price drops; borrowing only $2,000 provides a thick safety cushion. Both accounts are continuously monitored by code connected to oracle market prices.

When the collateral value drops below the threshold, liquidators can repay part of the debt, acquire collateral at a discount, complete account repairs, and earn rewards. This mechanism is intended to protect the funding pool before the collateral value falls below the debt. However, rapid asset sell-offs and insufficient market liquidity can weaken the liquidation effect, and blockchain network congestion can prevent liquidators from executing operations in a timely manner.

Tax motivations further complicate the risks on the borrower side: to reduce risk, they need to engage in trading operations, repay debts, or sell part of their appreciated holdings.

Borrowers who borrow to delay taxable transactions often procrastinate on position disposal; the higher the paper profits, or the closer the holdings are to meeting the holding period for lower long-term capital gains tax, the more pronounced this procrastination tendency becomes.

Stablecoins make this trading model significantly more attractive. Dollar-pegged tokens convert volatile collateral assets into directly usable dollar purchasing power. Traders can continue to pledge assets like Ethereum and borrow USDT or USDC for other purposes.

Many large holders are using crypto assets as collateral to borrow stablecoins, obtaining funds while retaining exposure to the price of the underlying assets.

At the time the loan is opened, the protocol sees the collateral ratio as normal. However, the protocol cannot know that the borrower initially bought Ethereum at a very low cost, and selling would realize substantial gains, or that holding for a few more months could yield considerable tax benefits. All these factors influence the borrower's behavioral choices when the loan becomes risky.

IRS: Venus Becomes a Natural Experiment

Researchers need to distinguish tax-driven behavior from the market disturbances in the crypto space itself. The Infrastructure Investment and Jobs Act, effective November 15, 2021, provides an opportunity for research.

Section 80603 of the Act expands the reporting obligations of digital asset brokers. Traders thus anticipate that their substantial on-chain activities will be reported to the IRS in the future.

This Act changes traders' expectations regarding third-party tax reporting, constituting an external shock event: potential U.S. taxpayers' behavior will be influenced by it, while international users are unaffected by this policy change.

The implementation of this reporting system took a long time. Custodial brokers will start reporting total income from corresponding buy-sell exchange transactions using Form 1099-DA from January 1, 2025; subsequent IRS rules further require that starting January 1, 2026, certain transactions need to report asset cost basis.

These regulatory rules only constrain institutions that actually hold user assets; non-custodial DeFi services are currently not within the regulatory coverage.

However, the research value of this paper comes from the market participants' expectations at the moment the Act was enacted in November 2021 ------ people anticipated that future transaction records would be reported, rather than waiting for the formal tax forms to be implemented.

Researchers compared data before and after the enactment of the Act (well before formal regulations took effect) to capture the market's reaction to the expectation that "future transactions will be subject to regulation," rather than passive feedback to already implemented tax forms.

The blockchain itself does not disclose users' nationalities or tax residency status, so researchers can only infer which wallets likely belong to U.S. users: wallets that trade predominantly during U.S. working hours, exhibit abnormal behavior on U.S. holidays, and hold U.S.-regulated dollar stablecoins.

Each inference dimension carries the possibility of misjudgment, so the paper provides multiple calculation metrics and employs strict standards with multiple overlapping conditions.

The entire research logic is akin to: the same financial system experiences a policy event shock. Two groups of users face the same token market conditions and the same Venus protocol rules; only the presumed U.S. user group is highly attentive to this new reporting regulation. This allows for the separation of behavioral changes driven by tax motivations from market disturbances.

Main model calculation results: after the Act took effect, presumed U.S.-related borrowers had a 24.5% lower probability of asset trading compared to international users. Borrowers with stablecoin liabilities saw an additional 23% decrease in trading activity. This aligns with the paper's logic: obtaining cash through stablecoins while still keeping appreciated collateral pledged.

In this paper, "liquidity" is defined narrowly at the wallet level: the probability of borrowers engaging in any asset trading daily. The liquidity commonly understood by the public refers to exchange depth, bid-ask spreads, and the cost of large sell-offs; this study measures user group trading activity and whether appreciated assets are "locked up."

Borrowers with higher paper profits and higher loan-to-value ratios exhibit more pronounced behavioral characteristics. Every December (especially the last week), investors generally defer profits to the next tax year, leading to a decline in trading activity; when holdings reach one year, satisfying the conditions for lower long-term capital gains tax in the U.S., trading activity rebounds. These behaviors confirm that the observed changes indeed stem from tax motivations, not merely coincidental phenomena brought about by the Act's timing.

Researchers estimate that U.S. borrowers in the sample defer an average of $3,357.42 in capital gains taxes annually, accounting for about 17% of the portfolio size during the same period. This estimate is based on the assumption that the wallet belongs to a U.S. taxpayer, reconstructed from on-chain data, and applying the corresponding tax rates, representing only a rough magnitude for the overall sample.

Cost Passed to the Funding Pool

The final part of the research analyzes how the decline in trading activity affects loan quality. Borrowers with low trading frequency will maintain high-risk positions longer, miss repayment opportunities, and fail to timely supplement collateral before the ratio hits the threshold. Tax preferences that originally occurred outside of Venus ultimately translate into unpaid debts within the protocol.

There is a risk of causal inversion here: defaults can also lead users to abandon wallets and stop trading.

Researchers used instrumental variable methods to extract the decline in trading activity caused by the Act's shock (this external shock is independent of the protocol itself), eliminating reverse causal interference.

Model calculations show that for every 1% increase in non-liquidity caused by tax factors, the number of default accounts rises by 11.2%, and the scale of default loans increases by 39.6%. A one standard deviation increase in non-liquidity corresponds to an increase of about $350 in default debt per borrower, which is 2.7 times the model's baseline value.

The percentage values may seem high, but this is a statistical limitation: the results only apply to the portion of the sample affected by this policy event and sensitive to tax issues, and cannot be generalized to all DeFi loans; they are only used to explain how motivations influence credit outcomes.

This exposes the inherent blind spots of automated lending. Smart contracts read collateral prices, debt balances, payable interest, and liquidation thresholds, but the borrower's holding costs and tax motivations are entirely excluded from the calculations.

Two wallets, with identical Ethereum collateral and loan conditions, appear equally risky in the eyes of the Venus system; yet one person is willing to sell to reduce risk, while the other is determined to avoid selling at all costs. This creates a borrower selection bias issue.

Over-collateralization can withstand ordinary price fluctuations; however, those borrowers who are extremely reluctant to dispose of appreciated assets will maintain their liabilities and reduce trading even as their safety cushion thins, concentrating risk on a group whose motivations the protocol cannot recognize.

When liquidation is executed smoothly, external participants repay debts and dispose of collateral, and lenders do not incur losses. However, if liquidation fails, losses will be borne according to the protocol's loss distribution mechanism, shared among the protocol reserves, token holders, and funding pool providers. Individual tax preferences ultimately become financial consequences borne by all contributors to the funding pool.

The paper also notes that tax issues are just one of many sources of risk.

This research only targets a single protocol during the 2020-2022 bull and bear cycle; users' U.S. identity is inferred from behavior; defaults are defined in a special way, only counting accounts with long-term high loan-to-value ratios. The research also confirms that extreme fluctuations in collateral prices and defects in liquidation mechanisms can also cause issues with Venus loans.

Other protocols with stronger liquidity and different collateral parameters may yield different results; liquidation bots in more specialized market environments will also perform differently than Venus during the sample period.

Even so, this paper provides a perspective that traditional lending data finds hard to achieve: a complete observation of collateral, debt, user behavior, liquidation, and position abandonment at the single wallet granularity.

DeFi has moved lending strategies that originally belonged to private banks onto public chains, opening them up to the general public. However, the automation replacing credit auditors does not eliminate human subjective motivations. Tax considerations, obsession with appreciated assets, and reluctance to sell will penetrate through the code, ultimately being borne by all those who provide funds to the funding pool.

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.