On July 9, 2020, a criminal complaint was filed against Joshua Thomas Argires, a Houston business owner, for submitting materially false loan applications to obtain and misuse funds from the Paycheck Protection Program (PPP). Around May 2020, Mr. Argires opened an account in the name of Texas Barbecue at Coinbase and began trading crypto, ultimately transferring the entire amount of the $956,250 loan. In another complaint, John Corbin Corona received PPP loans of over $413,000, allegedly transferring $155,000 into a Coinbase account. These are only two of many complaints against small business owners diverting PPP funds into crypto. These prosecuted cases encapsulate the most egregious instances, where, frequently, none of the funds were allocated for essential purposes such as payroll and operational expenses. Instead, the use of shell companies with fabricated financial records, as well as transfers to cryptocurrency platforms, were employed as part of various diversions, including the purchase of additional assets or luxury items.
Paycheck Protection and the Rise of Crypto
Our recent study, featured in this press article, hypothesizes that these extreme cases for which the DOJ collected enough evidence to prosecute may be the tip of a much bigger iceberg, in which small business owners responded to PPP loans by investing in crypto. One should consider concurrent events to PPP disbursements to uncover clues about what really happened.
The US government passed the $2.2 trillion CARES Act in March 2020 to address the unique liquidity challenges created by the pandemic. PPP is the centerpiece of the CARES Act, with the total amount of $800 billion disbursed over three phases between April 2020 and May 2021. Despite being a loan program, the screening standard was lax, and the terms were extremely generous, with an effective interest rate close to zero, since many borrowers did not even expect to pay it back. After the passage of the Act, the prices of cryptocurrencies such as Bitcoin and Ethereum were up by roughly 450% and 1,500%, respectively, in June 2021. The crypto boom, which occurred with no other event that would plausibly affect the value of these assets, may have been driven by the diversion of government funds to purchase cryptocurrency assets, causing a transfer of aid from its intended recipients (i.e., small business employees) to small business owners/prior owners of crypto assets.
Assessing this hypothesis empirically is challenging because we do not observe individual bank accounts wiring cash to crypto wallets, so we adopt an indirect approach starting with parts of the CARES Act most likely to translate into demand for crypto. The PPP was administered as forgivable loans to businesses with fewer than 500 employees. The PPP disbursed $349 billion from April 3 to April 16, 2020, and then an additional $434 billion was disbursed in 2021 as part of the Consolidated Appropriations Act. Most small businesses would qualify for forgivable loans, even if they did not entirely stop their economic activity during the pandemic; they did not need to demonstrate an economic loss equal to the forgivable loan. While local businesses such as restaurants or dental practices temporarily interrupted operations during the pandemic, many other businesses could operate (albeit at lower capacity). Any business whose loss of gross revenue due to the pandemic was less than its operating costs received a windfall gain from the PPP.
Given the size and swiftness of the program, almost no documentation was requested or audited. By October 2, 2022, 97% of loans had been forgiven with an average amount of $72,100. While, in principle, proceeds of the loans are to be used for eligible expenses, it is not illegal for a business to temporarily invest the received funds in assets with the intent of covering operations at a later date. Furthermore, no part of the PPP required proceeds to be kept exclusively in cash. In fact, even if a business owner ultimately “changed their mind” and admitted to not using the PPP for eligible expenses, the PPP could be reimbursed, amounting to a zero-interest loan with almost no condition or verification. Meanwhile, over the same period, crypto financial firms (exchanges and DeFi) charged interest on margin investment in crypto generally over 10% and with strict collateral constraints.
Reimbursing the PPP was typically unnecessary. There was almost no audit or verification of actual bank accounts, and even for businesses that faithfully reported uses of the PPP, the majority of small businesses are family-owned: as money is fungible, a small business owner could retain proceeds of the PPP as traditional assets in the company accounts, while simultaneously investing with personal funds in cryptocurrency, only to transfer back the company balance as dividends or salaries at a later date.
Our theory, which explains the events that ensued, combines the “house money” effect with an extraordinary marketing endeavor by sponsors of crypto assets. This marketing campaign took advantage of the readily available easy money provided by governments and capitalized on the fear of missing out among the middle-class, a demographic that had previously shown little interest in cryptocurrencies. When subject to the house money effect, individuals become more risk-seeking to unexpected gains, as if they were investing someone else’s money. This effect is, in part, a behavioral error, given that money is the same, regardless of whether it has been earned or received as a windfall gain. Via the PPP, many small business owners were “treated” with a windfall gain and responded by seeking more speculative assets.
Of course, crypto is not the only type of risky asset that became popular during the pandemic. Meme stocks (e.g., Gamestop, AMC) and personal trading via sites such as RobinHood are other examples. But crypto is unique because it is one of the only existing investments with no fundamental value, and its value is solely based on the ability to sell later – even national currencies are anchored in their acceptance to pay tax. In line with their appeal in response to the house money effect, they are extremely risky, promised easy gains through “safe” financial products that would yield over 10% through collateralized lending (many of which ultimately defaulted, see FTX, Genesis, or Terra), and were supported by a campaign paid by venture capital to present these assets as good long-term investments.
The campaigns were not solely focused on promoting the largest coins, such as Bitcoin or Ethereum. They encompassed a wide range of other coins and tokens, including scam tokens and me-too cryptos like Dogecoins, Shiba Inu, and Floki Inu, which promised long-term community involvement. Additionally, there were “liquidity” tokens issued to stabilize markets, some of which offered governance voting rights but lacked fundamental cash flows, such as FTT, AAve, or COMP tokens. Moreover, there were numerous Ether and Bitcoin supplements or alternatives that claimed to provide lower costs or faster execution through their blockchains, such as ADA, SOL, or DOT. Furthermore, there were coins like Tether, which were pegged to the dollar and supposedly backed by unaudited and dubious receivables. Lastly, there was a proliferation of non-fungible tokens (NFTs) that were essentially selling pointers to public web addresses featuring automated art projects. Financial firms in the crypto space raised billions from venture capitalists, with as much as $18.7 billion in 2022, a significant portion of which was allocated to marketing and exerting influence.
Quantifying the Diversion of PPP Funds to Crypto
We were able to document clear evidence of the transmission of the PPP to crypto markets using several unique features of PPP loans. First, these loans are observed at the local level, i.e., metropolitan statistical area (MSA), offering the granularity to associate PPP disbursements and interest in crypto. Second, since our interest is the causal effect of an extra $1 of PPP on crypto, this would require an exogenous flow of PPP funds unrelated to MSA events that would drive both interest in crypto and unemployment. For instance, in an MSA where the population is significantly impacted by shutdowns, there might be a higher demand for PPP loans and more time available for individuals to explore the world of crypto. This could create a spurious correlation between PPP, interest in crypto, and unemployment, which is not indicative of PPP driving interest in crypto or influencing unemployment. To address this, we use an instrument approach developed in prior literature, which exploits the unique feature of PPP – although loans are approved by Small Business Administration, they are processed and disbursed through banks. The instrument uses coverage by banks that were able or willing to process PPP funds for PPP disbursement. For instance, as an extreme example, Wells Fargo was exogenously prevented from granting PPP loans due to an ongoing enforcement action so that MSAs with a stronger relative presence of Wells Fargo were less able to submit PPP applications (business owners would have to open an account in other banks) for reasons that are plausibly unrelated to unemployment or interest in crypto. We construct a Bartik-like instrument that reflects the predicted access to PPP funds at the local level.
To capture interest crypto, we would ideally want to observe transfers by individuals to crypto wallets or accounts at major crypto exchanges. However, wallets on the blockchain are anonymous and exchanges do not report the identity of their clients. Instead, we first use Google searches that indicate interest in investing in crypto, based on the following words: “crypto,” “cryptocurrency,” “ether,” “bitcoin,” “token,” “ethereum,” “tether,” “coinbase,” “blockchain,” and “dogecoin,” to compute an average of the interest over each MSA-week period. Robustly, across all specifications, we find persuasive evidence that PPP disbursements cause an immediate interest in crypto.
We corroborate this conjecture further using aggregate evidence on actual activity in crypto markets, which is not at the local level and is affected by worldwide activity rather than just the US. Given this data limitation, we nevertheless note that PPP disbursements were at very specific points in time with no obvious correlated events and further use different metrics of activity to triangulate a pattern of evidence, including data from crypto data providers such as Coin Metrics, Blockchain.com, Coingecko, Glassnode, and First Rate. Our most direct evidence involves the transfer of coins from large to small wallets, under the presumption that many prior owners of crypto had accumulated larger balances (including the large wallet, “whales”) while crypto was trading at much lower prices and sold to new investors starting to accumulate cryptos in smaller wallets. During periods of PPP disbursements, the number of small wallets increases while the number of large wallets decreases. We also find that “new tokens” trading, defined as tokens created and marketed during the pandemic to meet a demand by new investors, increased during periods of PPP disbursements. This suggests – somewhat worryingly – that small businesses may have experimented with exotic cryptos other than mainstream bitcoin or ether cryptos. Using granular trading data, we document that the standard-deviation of trade sizes increases, further indicating entry by a wider set of investors into this market.
Many other measures of activity also boomed during periods of PPP disbursements, with trading volume markedly increasing. This led to significantly more congestion on the blockchain. The higher the congestion, the higher the fee to equalize the network’s limited capacity in recording transactions on the blockchain and the demand for new transactions. The fee is paid to miners who enter and verify transactions by solving encryption problems. We find that the total revenue of miners, which include the fee and the value of new coins minted, increased during PPP disbursements. This also caused an adjustment in the entry of mining efforts, with more minors competing for mining revenue (in fact, for the full anecdote, one of the authors had a friend who made major investments with his dentist neighbor to build a mining operation in the suburbs of a major US city, a backyard mining of sorts). The hash rate, a measure of the computational difficulty involved in solving the encryption problem and serving as a proxy for miners’ costs, demonstrated a significant uptick. This can be attributed to the entry of new miners, who, in response to predetermined miner revenue intended to reduce transaction demand, had to contend with greater difficulty in order to equalize the mining landscape.
A Tale of Economic Consequences: Unemployment and Food Sufficiency
While funds may have been diverted, not all proceeds from a loan would have been immediately used to support employment. Access to cryptos might bolster confidence and decrease business owners’ willingness to reduce their workforce or shutter the business. Consistent with prior studies, we find that the PPP was enormously costly in government money on a per-job basis: the average cost per job saved is $380,873. Diversion toward crypto made this much worse. By comparing the impact on unemployment with and without an interest in crypto, we observed that approximately 14% of the PPP funds, totaling around $110 billion, were diverted towards the crypto market.
In additional tests, we do not find a correlation between crypto returns and PPP disbursements, which, given that these events were largely anticipated, further suggests that savvy investors partly arbitraged these shocks.
We also inspected another government assistance program, the Economic Impact Program (EIP), paid $1,200 per individual making less than $75,000. Unlike the PPP, the amounts received by individuals are much smaller, and certain types of crypto investments (such as buying with a crypto wallet rather than an account at an exchange) would not be of sufficient size to justify the blockchain fee (greater than $100 at its peak). The EIP was distributed to US households rather than business owners, implying that a large portion of the aid went to individuals who do not save significant parts of their income. So, we expect the effects of the EIP to be smaller and find that metrics of activity in the crypto market are less affected. Nevertheless, the EIP aims to meet essential household needs and its effectiveness is evaluated by household surveys. We find a small adverse effect on the percentage of households who claim they have enough wanted food, which worryingly, suggests that in at least some households, funds that should have been used for supplementary food spending were used to speculate on the crypto market, even for income brackets likely lower than small business owners.
This is not direct evidence that identifies who used PPP loans to buy crypto, nor is it clear that such diversion was structured in an illegal mannner and could be prosecuted. But the accumulation of facts clearly delineates a pattern in which cryptos serve purposes contrary to public policy objectives and need to be carefully considered by regulators. These issues are not new, as other studies have shown that cryptos have served to facilitate illegal activities of a more severe nature than using public handouts to make risky investments. Regulation seems to have taken the turn to protect small investors making large losses in crypto exchanges, but deeper issues remain as to their use to support self-interested actions that may be legal but remain ethically dubious.
Many employees may not have kept their job because their employer used PPP funds for crypto, but the business owners did not necessarily do poorly. In the Figure below, we plot the portfolio returns from business owners investing a share of their PPP disbursements. At peak, toward the end of 2021, they made 800% of their initial investment, and even if they were kept until the end of 2022 after most cryptos crashed, most business owners would be up by a whopping 80%. But this also means that nearly $200 billion of investments in crypto (1.8 x 14% x PPP), compared to the $465 billion combined market cap of Bitcoin and Ethereum, may be money from business owners having invested PPP funds. How long will they keep these investments in crypto, and to whom will they sell if there isn’t a generation receiving the next $783 billion government handout?
Jeremy Bertomeu is an Associate Professor of Accounting in the Olin Business School at the Washington University in St. Louis.
Xiumin Martin is a Professor of Accounting in the Olin Business School at the Washington University in St. Louis.
Sheryl Zhang is a PhD candidate in Accounting in the Olin Business School at the Washington University in St. Louis.
This post was adapted from their paper, “Uncle Sam’s Stimulus and Crypto Boom,” available on SSRN.