Defi Protocol Risks

DeFi Protocol Risks: the Paradox of DeFi 1
By Nic Carter 2 and Linda Jeng 3
Abstract
Decentralized Finance (or “DeFi”) is growing in volume and in importance. DeFi promises
cheaper and more open access to financial services by reducing the costs and risks of using
centralized intermediaries. DeFi also holds the promise of interoperability across blockchains
that could help tear down financial sector silos, greatly promoting innovation and building
vibrant financial ecosystems. However, DeFi is not without its challenges, which are
understudied. This article does not seek to provide a comprehensive list of DeFi but to help
readers conceptually understand the drivers behind the risks inherent in DeFi. Many of the risks
described above stem from the decentralized nature of blockchains. The goal of automating the
delivery of financial services and reducing human dependencies also has the congruent effect of
reducing oversight and control. Disintermediating traditional intermediaries reduces high fees
and entry friction, but also creates new opportunities for new types of intermediaries. This article
discusses some of the new types of risks introduced by DeFi that are inherent to blockchain
systems along with traditional types of financial risks in DeFi that manifest in new ways: (i)
interconnections with the traditional financial system, (ii) operational risks stemming from
underlying blockchains, (iii) smart contract-based vulnerabilities, (iv) other governance and
regulatory risks, and (v) scalability challenges. In an effort to remove humans and automate as
much as possible through smart contracts, DeFi has introduced or amplified these risks. The
growth of DeFi will depend on its ability to navigate and build compatibility with traditional
finance and on how laws and regulations respond. Perhaps the biggest challenge of all is that the
DeFi ecosystem continues to grow while its underlying base layer (public infrastructure such as
Bitcoin or Ethereum) faces growing pains. As DeFi grows in importance and becomes more
mainstream, policymakers and industry representatives need to better understand the economic
and policy consequences of these new types of risks in order to build regulatory approaches and
risk management practices that can support and facilitate a healthy and robust DeFi ecosystem
and, ultimately, the financial stability of the greater financial system and real economy.
1
This article will be included as part of a forthcoming book edited by Bill Coen and Diane Maurice
“Regtech, Suptech and Beyond: Innovation and Technology in Financial Services” RiskBooks –
forthcoming 3Q 2021. The authors wish to thank Jon Frost and Greg Xethalis for reviewing this chapter
and for their invaluable feedback.
2
Nic Carter is a General Partner at Castle Island Ventures (CIV), a seed-stage venture firm investing in
startups building on public blockchains, and the co-founder of Coin Metrics, a blockchain analytics firm.
Disclaimer: CIV holds active liquid and illiquid positions in several decentralized finance tokens and
startups. Disclaimer: All mentions of protocols, tokens, and digital assets in this chapter are merely
exemplary and do not constitute endorsements
3
Linda Jeng is Visiting Scholar on Financial Technology at Georgetown University Law Center’s Institute
of International Economic Law. She is also the Global Head of Policy at Transparent Financial Systems.
She has held various regulatory roles, including at the Federal Reserve Board, the Financial Stability
Board, and the US Treasury Department.
Electronic copy available at: https://ssrn.com/abstract=3866699

1. Introduction
On February 16, 2021, the price of Bitcoin crossed $50,000 for the first time, doubling its value
in less than two months. 4 Earlier in the year, a string of announcements by a number of Wall
Street banks and traditional financial firms, including Bank of New York Mellon, Mastercard,
and Blackrock, proclaimed that they would begin working with bitcoin. The companies Square
and Tesla made splashes by investing a combined total of nearly $2 billion USD in bitcoin. 5
Meanwhile, Square’s and PayPal’s retail customers now buy an amount equivalent to a majority
of the new supply of bitcoin entering the market each day. 6 Visa also unveiled a bitcoin and
crypto plan to be launched later in 2021. 7 Crypto is becoming mainstream and is here to stay.
Decentralized finance (or “DeFi'') is typically understood by crypto users and enthusiasts as
platforms and protocols that seek to replicate existing financial services by using
crypto/blockchain technology with limited centralization. CoinDesk defines DeFi as: “an
umbrella term for a variety of financial applications in cryptocurrency or blockchain geared
toward disrupting financial intermediaries.” Fabian Schär defines it more specifically as “an
open, permissionless, and highly interoperable protocol stack built on public smart contract
platforms, such as the Ethereum blockchain.” 8
Central banks and financial regulators presently do not view the crypto market as being large
enough to pose a significant threat to global financial stability. 9 However, this assessment does
not discount the need for regulators, industry and academics to understand (1) what are the new
emerging risks of DeFi and (2) how DeFi may be impacting the transmission of traditional
financial risks. Crypto markets are not insignificant and can no longer be discounted as small.
For example, at the time of writing, DeFi projects on Ethereum hold collateral having the value
4
Vigna, Paul & Ostroff, Caitlin. “Bitcoin Trades Above $50,000 for First Time” Wall Street Journal (Feb.
16, 2021). https://www.wsj.com/articles/bitcoin-trades-above-50-000-for-first-time-11613479752
5
Son, Hugh. “Feeling the heat from employees, Wall Street banks get closer to adopting bitcoin” CNBC
(Feb. 12, 2021). https://www.cnbc.com/2021/02/12/bitcoin-banks-closer-accepting-cryptocurrency-asset-
class.html
6
Rooney, Kate. “Square and PayPal may be the new whales in the crypto market as clients flock to buy
bitcoin” CNBC (Nov. 24, 2020). https://www.cnbc.com/2020/11/24/square-and-paypal-emerge-as-whales-
in-the-crypto-market-.html
7
Bambrough, Billy. “Visa Reveals Bitcoin and Crypto Banking Roadmap Amid Race to Reach Network of
70 Million” Forbes (Feb. 3, 2021). https://www.forbes.com/sites/billybambrough/2021/02/03/visa-reveals-
bitcoin-and-crypto-banking-roadmap-amid-race-to-reach-network-of-70-million/?sh=7cc0664b401c
8
Fabian Schär, "Decentralized Finance: On Blockchain- and Smart Contract-Based Financial Markets,"
Federal Reserve Bank of St. Louis Review, Second Quarter 2021, pp. 153-74.
https://doi.org/10.20955/r.103.153-74 . See also, Buterin, Vitalik. "A Next-Generation Smart Contract and
Decentralized Application Platform." 2013; https://blockchainlab.com/pdf/Ethereum_white_paper-
a_next_generation_smart_contract_and_decentralized_application_platform-vitalik-buterin.pdf.
9
Financial Stability Board. “Crypto-asset markets: Potential channels for future financial stability
implications” (Oct. 10, 2018). https://www.fsb.org/wp-content/uploads/P101018.pdf

of around $50 billion. 10 As we will discuss below, crypto markets are becoming highly
interconnected with the traditional financial sector. In time, DeFi could become a significant, if
not the predominant, type of financial system, platformizing to varying extents the traditional
financial sector. In the meantime, we need to take on the challenge of identifying and assessing
the unique features of DeFi and what risks DeFi pose to the financial system.
i. Evolution of DeFi Movement to DeFi
DeFi comprises several components and continues to evolve quickly: (1) the public base layer
with the digitally native token, (2) software protocols that codify agreed rules, (3) smart contracts
that implement financial logic (i.e., execute transactions once specific conditions are met), and
(4) stablecoins backed by reserves held at banks. In this chapter, we look at the various
components of the DeFi universe with a particular focus on software protocols (aka. DeFi/DeFi
protocols). DeFi protocols are automated systems deployed on a public blockchain, typically
Ethereum, whereby users can take advantage of liquidity supplied by many counterparties in
order to engage in asset swaps or acquire leverage, without dealing with a centralized financial
counterparty.
After a call for crypto regulation by France and Germany, 11 the G20 Ministers of Finance and
Central Bank Governors instructed the Financial Stability Board (FSB) to assess its work and the
work of standard-setting bodies on crypto-assets. The FSB concluded that crypto-assets “did not
pose a material threat to global financial stability” at the time of assessment but that crypto-
assets would require “vigilant” monitoring. 1213 However, the FSB’s approach focused on
potential transmissions of risk to traditional financial sectors. We argue that, as DeFi becomes
mainstream, regulators and industry will need to quickly get up to speed on how DeFi operates
and what are its inherent risks for users and the real economy.
After the introduction in Section 1, Section 2 provides historical background on the evolution of
DeFi. The main body of this chapter, Section 3, identifies and attempts to categorize DeFi risks
into five main buckets. As we explore these crypto-centric risks, we keep in mind how these new
risks compare to the traditional credit, liquidity, counterparty, market and operational risks, and
how our understanding of these traditional risks could be applied to DeFi. How these traditional
financial risks manifest themselves in DeFi may differ somewhat from traditional financial
10
“Defi: Value Locked by category” The Block. Accessed April 12, 2021.
https://www.theblockcrypto.com/data/decentralized-finance/total-value-locked-tvl
11
https://www.telegraph.co.uk/technology/2018/02/09/france-germany-demand-bitcoin-clampdown/
12
Financial Stability Board. “Crypto-asset markets: Potential channels for future financial stability
implications” (Oct. 10, 2018). Accessed April 11, 2021. https://www.fsb.org/wp-
content/uploads/P101018.pdf
13
Financial Stability Board. “Crypto-assets: Report to the G20 on work by the FSB and standard-setting
bodies” (16 July 2018). Accessed April 11, 2021. https://www.fsb.org/wp-content/uploads/P160718-1.pdf

sectors. Section 4 concludes and provides a preliminary analysis of what these crypto-based risks
and vulnerabilities could mean to the global financial system.
2. Definitions
DeFi blockchain projects include decentralized exchanges (or “DEXs”), lending platforms where
central intermediaries are not needed to hold funds and transactions occur on a peer-to-peer basis
through automated processes, 14 and decentralized applications (or “dApps”). 15 One definition of
DeFi is “the movement that leverages decentralized networks to transform old financial products
into trustless and transparent protocols that run without intermediaries.” 16 Another defines DeFi
to mean where it “expands the use of blockchain from simple value transfer to more complex
financial use cases.” 17 And as mentioned earlier, another more specific definition is “an open,
permissionless, and highly interoperable protocol stack built on public smart contract platforms,
such as the Ethereum blockchain.” 18 Many argue that DeFi is a form of finance that uses
blockchain and does not rely on traditional central intermediaries, such as banks, stock
exchanges or broker/dealers.
DeFi has been rapidly evolving since the introduction of first generation bitcoin to the emergence
of second generation stablecoins and the use of initial coin offerings (ICOs) to fundraise. These
DeFi projects in theory can become active ecosystems, even alternatives to traditional financial
systems, by leveraging smart contracts and decentralized asset custody to replace costly,
traditional intermediaries. 19 Most DeFi projects are built on Ethereum, and many credit
Ethereum’s easy-to-program platform for enabling the explosion in DeFi projects. As of March
2021, 87% of 5,727 ICO-funded DeFi projects have been built on Ethereum. 20
Researchers Chen and Bellavitis have identified four main categories of DeFi projects: (i)
decentralized exchanges (DEXs), (ii) decentralized lending and borrowing, (iii) programmable
decentralized derivatives, and (iv) automated financial processes. 21 Each of these categories
14
Supra.
15
See, https://ethereum.org/en/developers/docs/dapps/
16
https://defiprime.com/
17
Hertig, Alyssa. “What is DeFi?” Coindesk (Sept. 18, 2020, updated Dec. 17, 2020).
https://www.coindesk.com/what-is-defi
18
Schär (2021), supra.
19
Qian, DJ. “Defi’s Rise is Inevitable, and Fusion is Driving this Evolution of Conventional Finance”
Bitcoin.com (Aug. 10, 2020). https://news.bitcoin.com/defis-riseDefi's Rise Is Inevitable, and Fusion Is
Driving This Evolution of Conventional Finance – Sponsored Bitcoin News-is-inevitable-and-fusion-is-
driving-this-evolution-of-conventional-finance/
20
www.icobench.com. Accessed March 1, 2021.
21
Chen, Yan & Bellavitis, Cristiano. (2020). “Blockchain Disruption and Decentralized Finance: The Rise
of Decentralized Business Models.” Journal of Business Venturing Insights 13 (June 2020).
10.1016/j.jbvi.2019.e00151.

possesses a set of risks, but they share some common features. They all leverage decentralized
infrastructure and smart contracts. Smart contracts, however, are not legal contracts. They are
software protocols that live “on chain” to automatically implement a procedure, legal contract or
business practice. 22
Why use smart contracts? Why use DeFi at all? The benefits of automated delivery of financial
services by smart contracts are attractive. The transparency offered by blockchain technology
provides efficient auditing of solvency and proof of reserve. Decentralization and the process of
unbundling financial services can remove expensive traditional intermediaries – making finance
more equitable. The use of smart contracts can also reduce execution risk. DeFi could allow for
more open and cheaper access to financial services, reducing costs and risks from using
centralized intermediaries. DeFi also holds the promise of interoperability across blockchains.
This borderlessness of DeFi can help tear down financial sector silos, greatly promoting
innovation and building vibrant financial ecosystems.
DeFi is not without its challenges, though. It introduces new types of risks, discussed below in
Section 3. The promise of interoperability offered by DeFi has led to a concentration of nearly all
DeFi projects on the blockchain Ethereum - a new form of concentration risk. Ironically, in the
mission to remove humans and automate as much as possible, other risks have been either
introduced or amplified, including the challenge to maintain code security. The growth of DeFi
will also depend on its ability to navigate and build compatibility with traditional finance. It will
also depend on how national and state laws and regulations evolve. Perhaps the biggest challenge
of all is that the DeFi ecosystem continues to grow while its underlying base layer (public
infrastructure such as Bitcoin or Ethereum) faces growing pains, manifesting in high fees.
3. Risk Factors in DeFi
The DeFi system is predicated on the notion of extreme transparency in which anyone can
effectively see everyone else’s transactions (although larger entities have found ways to be
anonymous by using popular analytics tools, such as pseudonymity and privacy enhancing
features). Extreme transparency offers tremendous potential for disintermediating traditional
financial intermediaries and automating delivery of financial services. But extreme transparency
also provides ample opportunities for exploitation. At its core, DeFi depends on shared, public
databases with public read access and unfettered write access – provided the entity adding an
https://www.researchgate.net/publication/337111343_Blockchain_Disruption_and_Decentralized_Finance
_The_Rise_of_Decentralized_Business_Models
22
See definition: “‘Smart contracts’ is a term used to describe computer code that automatically executes
all or parts of an agreement and is stored on a blockchain-based platform.”
https://corpgov.law.harvard.edu/2018/05/26/an-introduction-to-smart-contracts-and-their-potential-and-
inherent-limitations/

entry in the blockchain pays a sufficient fee. Anyone with knowledge of these systems, an
internet connection, and sufficient tokens to pay for fees can deploy a smart contract that any
other user can subsequently engage with in a permissionless manner. Smart contracts are
software protocols that live “on-chain” – they are publicly available for anyone to engage with,
audit, or scrutinize. This open access to smart contracts vastly increases the scope for financial
innovation, as developers, for instance, are not limited by financial institutions requiring
permission to engage with their APIs. Inevitably, this also introduces new forms of risk, as there
are no required professional or licensing qualifications restricting who can deploy, manage, or
engage with smart contracts.
A general objective shared by DeFi practitioners is stripping human discretion from financial
contracts, and encoding the rules for behavior into highly automated, publicly available systems.
In practice, however, human discretion remains. DeFi systems must be deployed, governed, and
upgraded, and face occasional bugs or exploitative interactions with other protocols. They also
run on public blockchains, which face similar issues – and occasionally require human
intervention, too. As such, the core DeFi protocols tend to retain some level of human
involvement from controlling entities. This is a means to mitigate risks when they emerge, but it
also poses a potential threat to these systems if the administrators themselves are compromised,
malicious, or somehow co-opted.
Some risk factors and exploits are analogous to those evident in existing financial products, like
market risk, the manipulation of an underlying price to interfere with a derivative – one of the
most frequent forms of attack against DeFi protocols, and frontrunning transactions through fee
upping and quant models. Others are completely novel and idiosyncratic to the asset class, like
protocol-level reorganizations to invalidate prior transactions, validators reordering transactions
to extract value from on-chain marketplaces, or ‘flash loans’ giving attackers unlimited free
leverage. 23
We divide our discussion of DeFi risk factors into five general buckets:
(i) interconnections with the traditional financial system,
(ii) operational risks stemming from underlying blockchains,
(iii) smart contract-based vulnerabilities,
(iv) other governance and regulatory risks, and
(v) scalability challenges.
23
There has been a series of flash loan attacks in the past year. Most recently as of time of writing,
PancakeSwap, a yield-farming aggregate (value lostunknown) and bEarn.Fi, a cross-chain farming
protocol (loss of almost $11 million) suffered flash loan attacks. See, Crawley, Jamie “Flash Loan Attack
Causes DeFi Token Bunny to Crash Over 95%” CoinDesk (May 20, 2021).
https://www.coindesk.com/flash-loan-attack-bunny-token

The list of risks identified in this chapter is by no means exhaustive, but we attempt to outline the
primary categories.
Figure 1. DeFi Protocols: Map of Interconnected Risks
Note: DeFi applications (e.g., MakerDao) rely on public blockchains (e.g., Ethereum, Bitcoin), which in turn rely
on miners/validators to validate blocks of transactions as well as human oversight and governance. DeFi protocols
are subject to governance, administration and maintenance. They rely on liquidity from stablecoins backed by
reserves held at banks. (Green: decentralized, blockchain-based. Blue: centralized)
i. Interconnections with traditional financial system
a. Banks holding reserves backing stablecoins
While DeFi aspires to create a parallel and independent financial system based on code rather
than legal enforcement, key components of the DeFi system rely in practice on traditional
financial market infrastructure. The most critical nexus between the two systems can be found in
stablecoins. These consist of dollar-denominated tokens circulating on public blockchains and, in
principle, are backed by commercial bank dollars immobilized at financial institutions.

Stablecoins are useful for transactions in DeFi as they introduce fiat-denominated collateral into
the open transactional context 24 of public blockchains. However, the vast majority of stablecoins
derive their value from underlying dollar instruments and thus introduce a dependency on an
issuer of the underlying instruments and the financial institution where the dollars are parked. At
the time of writing, at least $65 billion worth of stablecoins circulate on public blockchains, but
only around $3.1 billion consists of non-redeemable stablecoins issued against crypto-native
collateral. 25 The remainder is fully dependent on an ongoing bank relationship and the promise
of redeemability for the underlying instruments to be upheld.
Even some of the most purportedly decentralized stablecoins have introduced points of
compromise. The MakerDAO system is a set of tools for users to issue dollar-denominated
tokens (named “Dai”, which is “soft-pegged” to the US dollar 26) in an automated way against an
overcollateralized basket of other assets. Issuing dollar-denominated assets against crypto
collateral within a smart contract is intended to insulate the token from the traditional financial
system and potential points of compromise.
i. Market risk in stablecoins’ underlying reserves
In November 2019, MakerDAO introduced “non-native” forms of collateral backing an upgraded

form of Dai, dubbed “multi-collateral Dai”, to manage market volatility of ether (ETH). 27
Initially, all Dai were issued in an overcollateralized manner against the digitally-native
cryptocurrency ether. Collateralizing against ether made the MakerDAO system more insulated
from third-party liabilities, less interdependent with traditional finance and, thus, arguably more
robust and resilient. Since ether is no one’s liability and its value is solely market-determined, it
is arguably more suitable to back assets like Dai as long as its downside volatility is managed.
However, in November 2019, MakerDAO began to permit users to broaden the portfolio of
crypto-assets backing Dai in order to obtain a less volatile collateral, including the USD Coin
(USDC), Tether (USDT), Wrapped Bitcoin (WBTC), and Basic Attention Token (BAT). 28 This
collateral diversification introduced new risks. These new collateral types were not “liability-
free” like ether, but in some cases the liability of a single issuer. As of the time of writing, $1.06
24
‘Open transactional context’ means public blockchain-based assets can be exchanged on a peer-to-
peer basis worldwide with virtually no oversight. In practice, most stablecoin transactions happen on the
internals of the transactional graph and do not involve the issuer (and are hence not exposed to
KYC/AML). Keep in mind that most stablecoins refer to the USD as their unit of account, but others target
alternative sovereign currencies.
25
Coinmetrics, Dai and sUSD as the crypto-native stablecoins in question. Data current as of Apr. 12,
2021.
26
“Busting MakerDAO Myths: Seven Misconceptions about Dai” (Nov. 11, 2020).
https://blog.makerdao.com/busting-makerdao-myths-seven-misconceptions-about-dai/
27
Id.
28
https://daistats.com/#/

billion worth, or 16 percent, of the $6.5 billion collateral in the MakerDAO system represents the
liability of a third party. 29 All of the assets in question can be frozen by entities administering
these stablecoin systems, obviating the trustlessness of a portion of the MakerDAO system. For
example, if the USDC governing consortium Centre were to freeze the $332 million worth of
USDC 30 held in the MakerDAO reserve, MakerDAO’s ability to maintain the dollar peg of Dai
could be compromised. Furthermore, while Centre’s USDC has largely coexisted with DeFi, this
status quo could be tested should a regulator apply pressure to Centre 31 (primarily the founding
members Circle or Coinbase) or the regulated financial institutions issuing USDC. Centre’s
blacklisting policy indicates that they would blacklist blockchain addresses in order “[t]o comply
with a law, regulation, or legal order from a duly recognized US authorized authority, US court
of competent jurisdiction, or other governmental authority with jurisdiction over Centre.” 32
Additionally, the banks holding reserves backing USDC could withdraw their support for the
token issuer, as happened repeatedly with the stablecoin Tether. 33 34 So the presence of liability-
laden collateral in purportedly purely crypto-economic systems like Maker/Dai injects the
potential for interference through regulatory oversight, commercial bank policy, or direct action
from the stablecoin issuer itself.
ii. Sources of market illiquidity
As for the standard fiat-backed stablecoins, they now account for a significant share of liquidity
for the major DeFi protocols. The top five DeFi protocols by USD-equivalent amount of
collateral supplied – MakerDAO, Curve, Uniswap, Aave, and Compound – collectively host
29
The assets in question included ‘wrapped BTC’, and the stablecoins USDC, USDT, GUSD, Paxos,
TUSD. Data current as of Apr. 12, 2021.
30
https://duneanalytics.com/hagaetc/maker-dao---mcd. Figures current as of Mar. 15, 2021
31
For example, the FATF may consider centralized stablecoin issuers to be Virtual Asset Service
Providers (VASPs) and has suggested in their draft March 2021 guidance that member states impose
additional disclosure burdens on VASPs facilitating ‘unhosted’ transactions (possibly capturing how
stablecoin issuers operate). See p. 71, “Draft updated Guidance for a risk-based approach to virtual
assets and VASPs” FATF (March 2021). https://www.fatf-
gafi.org/media/fatf/documents/recommendations/March%202021%20-
%20VA%20Guidance%20update%20-%20Sixth%20draft%20-%20Public%20consultation.pdf
32
https://f.hubspotusercontent30.net/hubfs/9304636/PDF/Centre_Blacklisting_Policy_20200512.pdf
33
See, Attorney General of the State of New York. “Settlement agreement with Tether and Bitfinance”
(July 2020). https://ag.ny.gov/sites/default/files/2021.02.17_-_settlement_agreement_-
_execution_version.b-t_signed-c2_oag_signed.pdf
34
Tether is a controversial so-called stablecoin that was unable to substantiate its dollar reserves.
Despite the unclear backing of its dollar reserves. Nonetheless, Tether is an important source of liquidity
in crypto-finance. See, Attorney General’s settlement with Tether (2020), supra. To comply with this
settlement, Tether released for the first time in May 2021 a breakdown of its reserves composition. As of
March 31, 2021, Tether’s reserves were composed of 75.85% cash and equivalents, 12.55% secured
loans, 9.96% in corporate bonds and precious metals and 1.64% in other investments, including digital
currencies. Interestingly, 49% is backed by unspecified commercial paper. See, De, Nikhilesh. “Tether’s
First Reserve Breakdown Shows Token 49% Backed By Unspecified Commercial Paper” CoinDesk (May
13, 2021). https://www.coindesk.com/tether-first-reserve-composition-report-usdt

$3.818 billion in USDC and $1.06 billion in Tether (USDT) in deposits. 35 These figures
represent 42 percent of outstanding USDC and 5.2 percent of outstanding USDT circulating on
Ethereum. 36 These two stablecoins represent critical sources of liquidity for these various DeFi
protocols. USDC represents 19.5 percent of collateral on the lending protocol Compound, and
the USDC-ETH pair is the second-most liquid pair on the decentralized exchange Uniswap.
These stablecoins are naturally exposed to the failure of the banks holding collateral reserves
backing these two stablecoins. Historically, banking support for certain stablecoin issuers can be
questionable, as evidenced by the disclosures found in a settlement agreement between Tether
and the New York Attorney General’s office. 37 A bank insolvency, regulatory action, or issuer
failure – likely causing the stablecoins in question to trade at a discount to par, as happened
historically during confidence crises 38 – would impair the collateral and liquidity that powers
these DeFi systems.
b. High interconnectedness: banking relationships with crypto trading

firms
Aside from stablecoin banking, a handful of banks provide critical services to cryptocurrency
firms. Historically, only a small number of U.S. banks, including Silvergate Bank, Signature
Bank, and Metropolitan Community Bank, have actively pursued clients in the DeFi space.
These banks represent critical points of centralization for the crypto industry. A disruption or an
insolvency among any one of these banks would adversely affect whole swathes of the
cryptocurrency industry.
Perhaps the bank with the greatest concentration of the crypto industry, Silvergate Bank is a
California state-chartered bank based in San Diego that turned its focus to the cryptocurrency
industry in 2013 and now provides banking services to firms active in this space. Their flagship
product is the Silvergate Exchange Network (SEN), which enables real-time USD transfer
between its clients, which are largely centralized crypto exchanges and institutional investors. 39
Acquiring banking services has been so challenging for crypto exchanges and firms that
35
Maker: https://duneanalytics.com/hagaetc/maker-dao---mcd, Curve https://www.curve.fi/totaldeposits,
Uniswap https://info.uniswap.org/home, Aave https://aavewatch.com/ , Compound
https://compound.finance/markets (As of March 15, 2021)
36
Based on figures: 9 billion USDC circulating and 20.41 billion USDT_eth circulating (Source: Coin
Metrics). (As of March 15, 2021)
37
NY Attorney General Letitia James. Press Release: “Attorney General James Ends Virtual Currency
Trading Platform Bitfinex’s Illegal Activities in New York” (Feb. 23, 2021). https://ag.ny.gov/press-
release/2021/attorney-general-james-ends-virtual-currency-trading-platform-bitfinexs-illegal
38
For instance, when Wells Fargo withdrew its support for Tether in spring 2017 and convertibility was
temporarily suspended, (https://bitcoinist.com/bitcoin-trading-at-a-premium-on-bitfinex-and-poloniex/)
Tether traded as low as 92 cents on the dollar (Coin Metrics data)
39
Silvergate Capital Corporation Investor Presentation (January 2021).
https://s23.q4cdn.com/615058218/files/doc_presentations/2021/01/Silvergate-Capital-Investor-
Presentation-January-2021.pdf
10

Silvergate has become a key nexus connecting traditional banking and the digital currency
industry. As of 4Q2020, Silvergate boasted $5.5 billion in total assets on their balance sheet and
$5.03 billion in cryptocurrency deposits. 40 Their SEN transfer network processed $59.2 billion in
intra-bank transfer volume in the fourth quarter, 41 providing an alternative settlement means for
crypto firms looking to settle the USD fiat leg of crypto-fiat trades. While a small number of
more mature crypto firms are able to obtain banking relationships with the largest banks in the
U.S., 42 most firms active in the virtual currency industry rely on Silvergate and its peers, which
are relatively small community banks, to settle the USD fiat leg of crypto-fiat trades and for
banking services. Any instability or cessation of banking in this cohort could cripple the crypto
industry, as crypto exchanges, brokerages, and OTC desks would have to scramble to find
alternative sources of USD fiat liquidity. More recently, Facebook-backed Diem announced that
Silvergate will be the exclusive issuer of the Diem USD stablecoin in a sudden about-face from a
cross-border payments strategy to a US-centric approach. 43 This partnership with Facebook’s
Diem only further augments the U.S. crypto industry’s exposure to Silvergate.
c. Retail exposure: consumer fintech apps
DeFi has begun to cross the threshold to mainstream consumer fintech apps, thus moving beyond
an audience of high-tech early adopters. A number of retail crypto exchanges have begun serving
as interfaces for DeFi protocols, effectively reducing the frictions involved in getting access to
DeFi – and exposing retail users to their benefits and risks. Now there are publicly traded firms
that depend on the functionality of smart contracts and may well have user funds deposited with
them.
Consumer fintech apps now make crypto highly accessible to retail investors who may not fully
understand what they are trading. The popular retail-facing brokerage Coinbase, which boasts 56
million verified users as of their Q1 2021 quarterly filing, 44 has begun to embrace DeFi,
positioning themselves among other things as an interface to these blockchain protocols. For
instance, Coinbase details their growing proximity to and engagement with the decentralized
interest rate swap protocol Compound in its Form S-1:
40
Silvergate Capital Corporation 4Q20 Earnings Presentation (Jan. 20, 2021).
https://s23.q4cdn.com/615058218/files/doc_financials/2020/q4/Ex.-99.2-SI-4Q20-Earnings-Presentation-
1.20.2021.pdf
41
Silverage SEN Network Transfer Volume (Quarterly) Q42018 - Q4 2020.
https://www.theblockcrypto.com/data/crypto-markets/public-companies/sen-transfer-volume
42
Palmer, Daniel. “JPMorgan Bank Takes on Coinbase, Gemini as Its First Crypto Exchange Customers”
Coindesk (May 12, 2020). https://www.coindesk.com/coinbase-gemini-first-crypto-exchange-customers-
jpmorgan-bank-report
43
Diem Association. “Partnership with Silvergate and Strategic Shift to the United States” (May
12, 2021). https://www.diem.com/en-us/updates/diem-silvergate-partnership/
44
Coinbase First Quarter 2021 Announcement. https://investor.coinbase.com/news/news-
details/2021/Coinbase-Announces-First-Quarter-2021-Estimated-Results-and-Full-Year-2021-
Outlook/default.aspx
11

Our relationship with Compound began in 2018 when Coinbase Ventures invested
in Compound Labs, Inc., the DeFi pioneer behind the Compound protocol.
Coinbase was also an early adopter of Compound, supplying USDC liquidity to
the protocol in 2019 and allowing Coinbase Wallet users to access Compound
directly starting in early 2020. 45
A number of other cryptocurrency brokers, custodians, and lenders have begun to see themselves
as interfaces to DeFi protocols, in addition to their core businesses.
Binance, one of the largest spot and derivatives exchanges for cryptocurrencies, has reported a
24 hour trading volume of $80 billion on January 4, 2021 46 and has over 350,000 BTC and 3.6
million ETH held on deposit on behalf of clients. 47 This large cryptocurrency exchange has now
openly embraced DeFi, providing not only a centralized brokerage and exchange experience, but
a number of passthrough products enabling users to participate in decentralized protocols
through its Binance Earn 48 suite.
Additionally, the Swiss fintech firm Taurus Group has integrated the lending and borrowing
Aave protocol 49 into its infrastructure, permitting institutional clients to access liquidity on the
DeFi protocol. 50 This presages a possible scenario where fintechs or financial institutions start to
put client assets in DeFi protocols in order to take advantage of attractive interest rates, 51 which
are generally higher than returns on cash held at banks (although they offer fundamentally
different risk profiles).
d. Corporate exposure: corporate treasuries
Lastly, some corporations are obtaining direct exposures to native cryptocurrencies on their
balance sheet, either as an alternative treasury asset (as with Microstrategy, Square, or Tesla) or
45
Coinbase Global, Inc. Form S1 Registration Statement. (filed with the SEC on Feb. 25, 2021).
https://www.sec.gov/Archives/edgar/data/1679788/000162828021003168/coinbaseglobalincs-1.htm
46
Binance reported daily trading volume of $80 billion in 24-hour trade activity on Jan. 4, 2021. See,
Haig, Samuel. “Binance hits record high of $80B in daily volume as crypto markets surge” Cointelegraph
(January 5, 2021). https://cointelegraph.com/news/binance-hits-record-high-of-80b-in-daily-volume-as-
crypto-markets-surge
47
CoinMetrics data. Data retrieved Mar. 15, 2021
48
See, https://www.binance.com/en/earn#flex-item
49
Aave is a “decentralised non-custodial liquidity market protocol” in which users can provide liquidity to
earn an interest rate, or borrow against their assets in either an overcollateralized manner or
undercollateralized with a flash loan. https://docs.aave.com/faq/
50
Akhtar, Tanzeel. “Digital Assets Firm Taurus Integrates Aave Protocol to Improve Banking Access to
DeFi” Coindesk (Mar. 8, 2021). https://www.coindesk.com/digital-assets-firm-taurus-banking-access-defi-
aave-partnership
51
See, https://defirate.com/lend/
12

in preparation to actually use the tokens to transact on the protocol directly. This presages more
engagement from public corporations with these shared infrastructures. The Chinese smartphone
firm Meitu Inc. acquired 15,000 ETH (worth $22m at the time of purchase), citing its potential
utility in future transactions on the Ethereum network:
[T]he Ether purchased would become the gas reserve for the Group’s potential
dAPP(s) to consume in the future, as well as being used as consideration for
investing in blockchain-based projects that take Ether as consideration. 52
Meitu indicated in their disclosure that they were considering launching Ethereum-based dApps,
and would thus require a reserve of ether in order to transact on the Ethereum network.
While interactions between traditional firms and DeFi systems have been historically sparse,
growing evidence suggests that integration is taking place. The earliest adopters were crypto
exchanges exchanging crypto-assets with traditional assets and providing passthrough services to
DeFi protocols. More interactions are emerging between banks servicing crypto businesses,
transacting on these networks directly and increasingly with other firms looking to benefit from
the assurances of public blockchains. More recently, Visa announced their intention to engage
with DeFi directly, enabling the settlement of transactions with USDC on the Ethereum
network. 53 As DeFi comes to offer more modes of transactions, firms like Meitu may come to
have an interest in using these DeFi networks directly. Such corporate firms will need to assess
their risk exposures to a protocol’s smart contracts and underlying cryptocurrency and
blockchain (discussed below). They will also need to assess how they may even pass these risks
on to their customers and business partners.
ii. Operational risks stemming from underlying blockchains
DeFi applications ultimately rely on public blockchains for settlement and contract resolution.
The most popular base layer, as measured by liquidity for such applications, is Ethereum with
around $46 billion worth of collateral (composed of various crypto-assets and stablecoins) being
employed in Ethereum-based smart contracts. 54 A number of other blockchains now host DeFi
applications and are eyeing Ethereum’s lead.
52
Meitu, Inc. Voluntary Announcement: Purchase of Cryptocurrencies (Ether and Bitcoin) (Mar. 7, 2021).
(https://corp-static.meitu.com/corp-new/92016878a68bac4ad8121e906eae6687_1615115628.pdf
53
Visa. “Digital currency comes to Visa’s settlement platform.” (Mar. 29, 2021). https://usa.visa.com/visa-
everywhere/blog/bdp/2021/03/26/digital-currency-comes-1616782388876.html
54
Gross Value Locked and Net Value Locked (Ethereum DeFi). Accessed April 12, 2021.
https://www.theblockcrypto.com/data/decentralized-finance/total-value-locked-tvl/true-value-locked-and-
total-value-locked
13

The orderly operation of these applications relies critically on these base layer blockchains
functioning, which cannot always be guaranteed. Transacting parties internalize novel risks,
which may have no analogues in traditional finance where messaging and settlement systems are
governed by single entities or bodies (like SWIFT, The Clearing House, or the Federal Reserve
with Fedwire). Instead, public blockchains are largely decentralized settings where validators are
compensated for assembling transactions into blocks and are expected to do so honestly based on
economic incentives. 55 As there are no central administrators in these systems, the responsibility
for evaluating the risk of relying on these infrastructures effectively falls on the end users,
applications, and new types of intermediaries involved in the DeFi systems.
a. Consensus failures
Consensus – the construction, approval and distribution of blocks of transaction across

distributed ledgers – on these blockchains is, however, not a given. While the largest and most
robust blockchains such as Bitcoin and Ethereum experience virtually no outages, outages are
not completely unheard of. Bitcoin infamously had two major “rollbacks” in 2010 56 and 2013 57
when a significant number of blocks, and hence transactions, were unrecorded or essentially
reversed. Collectively, around 15 hours’ worth of transactions were removed over the course of
those two events.
Ethereum is arguably more fragile to outages since most users do not run nodes but instead rely
on service providers like Infura to query and index the blockchain and broadcast transactions.
When these service providers experience downtime, as was the case with Infura during an
unplanned chain split in 2020, 58 intermediated transactions ground to a halt.
b. Underlying protocol interventions
Blockchains are not immune to politics, as they are, after all, governed by the humans that
establish their rules. Most infamously, in 2016, Ethereum leadership coordinated the selective
removal of balances from the blockchain after a particularly large DeFi application called “The
DAO” was hacked and exploited. 59 Ethereum leadership deemed it necessary to intervene on the
55
For more reading about economic incentives, see, Auer, Raphael, Cyril Monnet and Hyun Song Shin.
“Permissioned distributed ledgers and the governance of money” BIS Working Papers No 924 (January
2021). https://www.bis.org/publ/work924.pdf
56
Coopahtroopa, Cooper. “YFI Minting Ownership” (Jul. 2020). https://gov.yearn.finance/t/yfi-minting-
ownership/155
57
Andresen, Gavin. “March 2013 Chain Fork Post-Mortem” (Mar. 3, 2020).
https://github.com/bitcoin/bips/blob/master/bip-0050.mediawiki
58
Khalili, Joel. “Massive Ethereum outage forces crypto exchanges to block withdrawals” techradar.pro
(Nov. 11, 2020). https://www.techradar.com/news/massive-ethereum-outage-forces-crypto-exchanges-to-
block-withdrawals
59
Siegel, David. “Understanding the Dao Attack” Coindesk (Jun. 25, 2016, updated Dec. 17, 2020).
(https://www.coindesk.com/understanding-dao-hack-journalists
14

Ethereum blockchain due to the large fraction of outstanding ether locked in the faulty DAO
contract. Some Ethereum community members rebelled against the arbitrary changes and
supported the original Ethereum chain. The intervention caused a hard fork in the blockchain, as
two versions of Ethereum came to exist in tandem (with the original, but less widely adopted,
version ultimately being called “Ethereum Classic”). This is an example of a contract failure
ultimately affecting the underlying protocol itself and demonstrating that certain critically large
applications can take on a systemic nature within protocol politics. At the time of the exploit,
The DAO contract accounted for 15 percent of ether outstanding at the time. While Ethereum
leadership have not intervened to remediate subsequent hacks and failures, one might imagine
that if a popular contract with a similar threshold of ether was breached, it might call them into
action at the blockchain level once again. In the case of the DAO, Ethereum’s future switch to
Proof of Stake was cited as justification for rolling back the exploit (which would have granted a
presumably hostile actor a large share of the outstanding ether, and hence a significant role in the
future of the network under a Proof of Stake regime). Other less critical bugs or exploits have not
met the seriousness threshold to merit a rollback, even when those affected lobbied Ethereum
leadership. 60
While the post-DAO hard fork of Ethereum is generally seen as a prudent move, it constituted,
on strict terms, a violation of property rights and brought into question the settlement assurances
of the blockchain. On Ethereum, there is no legal adjudication – knowledge of a private key is
tantamount to ownership. Thus, under the protocol rules, the entity that exploited the DAO was
the rightful owner of the ether in question, and those rules were overridden to “bail out”
depositors in the DAO contract. 61 Such interventions could be helpful for obtaining recourse
when catastrophic failures or bugs occur, but they also introduce subjectivity and arbitrariness
into the settlement process. 62
c. Proof of Work (PoW) consensus failures
60
In the case of the Parity hack, Parity asked for (and were denied) a hard fork to undo the loss of 513k
Ethere. See https://www.businessinsider.com/ethereum-price-parity-hack-bug-fork-2017-11
61
Please note that holders of Ethereum Classic on the original blockchain kept their property rights but
not the funds stolen in the DAO attack. Arruñada, Benito, Prospects of Blockchain in Contract and
Property (February 23, 2020). See, Pompeu Fabra University, Economics and Business Working Paper
1696, 2020, Available at SSRN: https://ssrn.com/abstract=3543137 or
http://dx.doi.org/10.2139/ssrn.3543137
62
Settlement in the blockchain is a complex issue for blockchain systems and involves a combination of
both operational and legal risks. In traditional finance, a number of operational and legal frictions are
baked into the settlement process in both payments and trading,such as time, central intermediaries,
and/or contractual agreements. These frictions, which also act as risk mitigation, do not often have
analogous features on blockchains. Note also that the settlement process differs significantly between
UTXO- vs account-based blockchains.
15

More straightforwardly, smaller blockchains can be exploited when miners believe they are not
sufficiently compensated. When miners gain sufficient hashpower, 63 they can coordinate
consensus attacks, of which a subset is known as reorganization attacks or “51 percent attacks”.
These attacks consist of exploits in which validators employ their privileged access to transaction
ordering to extract some value from the blockchain. These consensus attacks generally take place
on Proof of Work (PoW) blockchains because such blockchains provide relatively low
compensation thresholds to miners, making validator attacks more economically plausible.
Often, these attacks occur in the presence of general-purpose computing hardware, which can be
borrowed or rented. 64
As an example, in early 2021, validators on the Verge blockchain rolled back 200 days of data,
effectively invalidating months of transactions. 65 These reorganizations of blocks can be used to
omit certain transactions that were presumed settled, including deposits credited by an exchange.
Thus, reorganizations are often tools for the fraudulently misleading merchants or crypto
exchanges into believing that there is a valid deposit, which is then ultimately excluded from the
ledger. Indeed, both the Ethereum Classic 66 and the Bitcoin Gold 67 blockchains have suffered
multiple such protocol-level attacks, some of which were used to successfully defraud crypto
exchanges. DeFi applications rely on the base layer blockchains to settle and clear transactions,
so the application stack is compromised when the underlying blockchain malfunctions.
Both Ethereum and Bitcoin currently rely on PoW, so they are theoretically exposed to these
kinds of attacks. However, Ethereum and Bitcoin offer incredibly large security budgets, 68
making an attack extremely expensive and likely impractical. Additionally, Bitcoin is mined with
bitcoin-focused hardware (known as “Application-Specific Integrated Circuits” or ASICs) that
cannot be repurposed for use in general computing or for most other crypto networks, so miners
would have less incentive to attack the Bitcoin blockchain and destroy what gives their Bitcoin
ASICs value. Moreover, as Bitcoin ASICs are essentially the physical embodiment of future cash
63
“Hashpower” of “hashrate” refers to “the total combined computational power that is being used to mine
and process transactions on a Proof-of-Work blockchain, such as Bitcoin and Ethereum (prior to the 2.0
upgrade).” See, https://www.coindesk.com/what-does-hashrate-mean
64
Note: a 51% attack allows malicious actors to unrecord or prevent the recording of transactions, but not
to fraudulently generate new transactions that they cannot otherwise digitally sign. Of course, ordering
and recording of blocks can be powerful tools nonetheless, particularly for crypto-finance.
65
Mapperson, Joshua. “Verge of disaster: 200 days transactions wiped from blockchain,” Cointelegraph
(Feb. 16, 2021). https://cointelegraph.com/news/verge-of-disaster-200-days-transactions-wiped-from-
blockchain
66
Shen, Muyao. “Crypto Investors Have Ignored Three Straight 51% Attacks on ETC,” Coindesk (Sept. 8,
2020). https://www.coindesk.com/crypto-51-attacks-etc
67
Nelson, Danny. “Attempted 51% Attack on Bitcoin Gold Was Thwarted, Developers Say” Coindesk (Jul.
10, 2020). https://www.coindesk.com/attempted-51-attack-on-bitcoin-gold-was-thwarted-developers-say
68
On a trailing seven day basis, Bitcoin offers miners an average of $60m/day, and Ethereum is offering
miners $48m/day (Source: Coin Metrics, as of Apr. 12, 2021).
16

flows in bitcoin over the ASIC’s useful lifetime, miners are strongly incentivized to support the
long term value of the Bitcoin blockchain. 69
Similarly, a large share of the value of the high-end graphics processing units (GPUs) used to
mine ether derives from the value of ether itself, 70 so miners attacking the Ethereum blockchain
would be depreciating their own equipment in doing so. For blockchains such as Bitcoin with a
capped issuance, questions remain over the long run viability of PoW when Bitcoin becomes a
network based solely on transaction fees. Various studies have identified the potential instability
or insufficiency of a fee-based PoW market environment. 71
d. Miner extractable value (MEV)
Nevertheless, reorganizations of blocks (or 51 percent attacks) are only one class of a broader set
of validator-based exploitations known as “miner extractable value” (or MEV). 72 First
introduced by researchers Daian et al (2019), 73 the term MEV refers to the value that validators
(the entities assembling transactions into blocks) or third parties can extract from transacting
users by frontrunning them and selectively reordering transactions. 74 MEV is made possible due
to the innate transparency of Ethereum transactions, their utility in on-chain exchanges, and the
possibility of gaining priority by outbidding other users (or simply reordering transactions if you
are the miner). MEV can be thought of as somewhat analogous to a hedge fund paying for order
flow in order to trade against uninformed or retail flow.
69
Note: there could be a risk of state action. For example, Chinese authorities could take control of a
significant chunk of bitcoin mining. https://blog.lopp.net/are-chinese-miners-threat-bitcoin/
70
It is helpful to think of the hardware used to mine blockchains as a physically-instantiated bundle of call
options for the underlying token gradually unlocking over the useful lifetime of the hardware.
71
See, Carlsten, Miles & Harry Kalodner, Matt Weinberg, and Arvind Narayanan. Working Paper: “On the
instability of Bitcoin without the block reward.” PrincetonEconomics (Oct. 2016).
https://economics.princeton.edu/working-papers/on-the-instability-of-bitcoin-without-the-block-reward/.
See also, Budish, Eric. “The Economic Limits of Bitcoin and the Blockchain” University of Chicago Booth
School of Business,(Jun. 5, 2018) https://faculty.chicagobooth.edu/eric.budish/research/Economic-
Limits-Bitcoin-Blockchain.pdf, and Auer, Raphael. “The doomsday economics of ‘proof-of-work’ in
cryptocurrencies” VOXeu/CEPR (Mar. 8, 2019). https://voxeu.org/article/doomsday-economics-proof-
work-cryptocurrencies
72
Because not all entities extracting value in this manner are miners, MEV is sometimes styled as
“Maximal Extracted Value”
73
Daian, P., Goldfeder, S., Kell, T., Li, Y., Zhao, X., Bentov, I., Breidenbach, L., Juels, A. “Flash Boys 2.0:
Frontrunning, Transaction Reordering, and Consensus Instability in Decentralized Exchanges” (Apr. 10,
2019). https://arxiv.org/abs/1904.05234
74
Flashbots defines MEV as “the total value that can be extracted permissionlessly (i.e. without any
special rights) from the re-ordering, insertion or censorship of transactions within a block being produced.
As miners currently have the ultimate say on transaction ordering and inclusion in Ethereum, they can be
seen as the most powerful player in this game [...] MEV exists on any blockchain and layers where there
is a party responsible for transaction ordering (eg. validators, rollup providers).” Flashbots.net, “FAQ”.
Accessed Apr. 13, 2021. https://explore.flashbots.net/faq
17

As the complexity of transactions increases, more frontrunning and risk-free arbitrage
opportunities emerge. Thus, the vast majority of observed MEV takes place on Ethereum, and
largely relates to transactions occurring on automated market maker (AMM) exchanges – where
users can frictionlessly swap assets by engaging with pools of liquidity. AMMs offer users
guaranteed liquidity on exchanges, albeit at the potential cost of efficient execution. According
to Flashbots.net, a lower bound of $369 million worth of MEV has been harvested by validators
(or arbitrage bots) since January 2020. 75 This represents a net drag on users, which end up
financing the MEV through slippage on their trades. Effectively, MEV can be understood as
similar to a rake at a casino.
Most researchers consider MEV endemic to blockchains – like Ethereum – where transactions on
decentralized exchanges or DEXs (including DEX platforms employing Automated Market
Makers, such as Uniswap) are transparent. If the parties engaged in frontrunning materially
degrade users’ transactional experience, the logic of transparent DeFi could be called into
question. While some analysts contend that MEV represents an alternative subsidy to miners or
validators, 76 permitting blockchains to function at a lower level of issuance or fees, researchers
Qin, Zhou and Gervais have highlighted how aggressive MEV poses a threat to consensus. In
their estimate, “[the] biggest danger lies in the willingness of miners to extract and compete over
MEV, which would increase the stale block rate and consequently aggravate the risks of double-
spending and selfish mining.” 77 Stale blocks and double spending reduce the predictability of the
base layer and introduce uncertainty into settlement finality, impairing the assurances of crypto-
economic protocols.
In an attempt to mitigate the protocol-level harms of MEV, Ethereum developers have proposed
an Ethereum node client that codifies MEV and allows miners to auction off their rights to
reorder transactions within a block, delegating the process of finding risk-free arbitrage to
specialized third parties. 78 This would turn MEV into an explicit part of the compensation
structure for miners, reduce the protocol instability caused by the current adversarial state of
MEV, and increase fairness in mining by allowing less sophisticated miners to cheaply monetize
MEV. Other researchers, unconvinced by the codification of MEV, have proposed alternative
solutions, such as fair transaction ordering, or the encryption of transactions between the
75
https://explore.flashbots.net/. Accessed Apr. 12, 2021.
76
As of Apr. 12, 2021, 58 percent of Ethereum hashrate is associated with pools auctioning off the rights
to reorder transactions via the Flashbots protocol. Effectively, these miners are selling the rights to
specialists who extract economic rent by engaging in frontrunning trades. See, Flashbots Transparency
Report - March 2021 (Apr 12). https://medium.com/flashbots/flashbots-transparency-report-march-2021-
d3930b4b98a9
77
Qin, K., Zhou, L., and Gervais, A. “Quantifying Blockchain Extractable Value: How dark is the forest?”
(Jan. 22, 2021). arXiv.2101.055==v3 [cs.CR] 22 Jan 2021 https://arxiv.org/pdf/2101.05511.pdf
78
See, Floersch, Karl. “MEV Auction: Auctioning transaction ordering rights as a solution to Miner
Extractable Value” ethresear.ch (Jan. 2020). https://ethresear.ch/t/mev-auction-auctioning-transaction-
ordering-rights-as-a-solution-to-miner-extractable-value/6788 and https://ethresear.ch/t/flashbots-
frontrunning-the-mev-crisis/8251/1
18

broadcast and execution stage. 79 Due to the downsides associated with broadcasting transactions
to a global mempool, 80 privately-mined transactions are becoming more popular. Effectively,
this involves routing transactional data to miners directly in a manner reminiscent of dark pools.
At the time of writing, over 75,000 Ethereum transactions have been sent directly to miners 81
rather than being broadcast to the network in the conventional manner.
At present, MEV appears to be a fundamental feature of data-rich blockchains that facilitate

transparent on-chain exchange. Transparent queuing systems for pending transactions combined
with the ability to outbid and displace a transaction inevitably yields exploitation opportunities.
Unlike a retail brokerage like Robinhood selling customer order flow, 82 MEV extractors are not
obliged to the individuals they are arbitraging. Thus, there are no natural limits to the
exploitation of end users through MEV.
e. Validator Cartels
Non-PoW blockchains are not immune to protocol interventions at the validator level. One
popular alternative to PoW is known as Proof of Stake, where the power to assemble transactions
into blocks (and, in some cases, exert political power over the network) is a function of one’s
share of all protocol tokens held. In certain network arrangements, the number of validator slots
is fixed, creating strong incentives to consolidate power and cartelize. Because validators are
typically rewarded with fees or new issuance, the consolidation of power through vote-buying
has been observed 83 in Proof of Stake blockchains, such as EOS (which maintains 21 slots for
validators). Such measures allow validators to consolidate power, granting them eventual control
over which transactions can be included in the final ledger. If validators are fixed and free
market competition for blockspace is snuffed out, the censor-resistance of the protocol properties
would be at risk. Since DeFi is built on the assumption that the underlying financial
infrastructure is neutral and unstoppable, such concentration of power in validators is a
significant threat. An instance of validator collusion can be found on the STEEM network, where
STEEM coins owned by blockchain entrepreneur Justin Sun were frozen after validators
suspected his intentions to co-opt the network:
79
Juels, Eyal, and Kelkar. “Miners, Front-Running-as-a-Service Is Theft,” Coindesk (Apr. 7, 2021).
https://www.coindesk.com/miners-front-running-service-theft
80
The memory pool or mempool is a node’s holding area for broadcasted but un-mined transactions.
Transactions present in the mempool are transparent to anyone participating in consensus.
81
Etherscan, “Private Transactions.” Data current as of Apr. 14, 2021.
https://etherscan.io/txs/label/private-transaction
82
See, Roberts, Jeff John and Morris, David. Robinhood makes millions selling your stock trades … is
that so wrong?” Fortune (July 8, 2020).
https://fortune.com/2020/07/08/robinhood-makes-millions-selling-your-stock-trades-is-that-so-wrong/
83
Dale, Brady. “EOS investors can’t say they weren’t warned.” Coindesk (Oct. 3, 2018).
https://www.coindesk.com/vitalik-called-it-vote-buying-scandal-stokes-fears-of-eos-failure
19

Specifically, the witnesses were able to unilaterally lock out Sun after a simple
majority vote passed 19 to 1. They had orchestrated the plan in a private Slack
group, ran a software upgrade on the blockchain and froze the Tron Foundation
CEO’s funds. 84
In this case, Sun fought back by enlisting custodial exchanges – which held large fractions of the
supply of STEEM on behalf of users – to employ user deposits to vote in his favor and overrule
the actions of the validators. 85 This illustrates how large cryptocurrency custodians and
cryptocurrency deposit-taking institutions can take on vital roles as kingmakers in Proof of Stake
systems. With Ethereum, the largest DeFi platform slated to transition to Proof of Stake,
custodians holding large quantities of ether will have outsize control over the network and may
be able to materially influence network outcomes. Thus far, crypto exchanges have generally not
recused themselves from protocol interventions. Effectively, they act as principals rather than as
agents when deploying client funds for on-chain votes.
f. Inflation bugs
Other more catastrophic protocol vulnerabilities abound, which can affect the
DeFi applications built on top of them. One such risk is posed by inflation bugs, which inflate
the supply of coins ahead of a pre-agreed or expected schedule. 86 As coins (minted in excess of
the defined schedule) are issued and begin to circulate, recipients of these new coins have a
strong disincentive to roll back the chain and undo the unexpected inflation. Inflation bugs are
frequent and have affected many of the largest blockchain protocols – and in some cases, were
not fully remediated. Blockchains that have witnessed material inflation bugs that were exploited
include Bitcoin, 87 Bitcoin Private, 88 and Stellar, 89 as well as many other less notable cases.
84
Copeland, Tim. “Steem vs Tron: The rebellion against a cryptocurrency empire” Decrypt (Aug. 18,
2020). https://decrypt.co/38050/steem-steemit-tron-justin-sun-cryptocurrency-war
85
Id.
86
Avan-Nomayo, Osato. “Inflation Bug Still a Danger to More than Half of All Bitcoin Full Nodes”
Cointelegraph (May 19, 2019). https://cointelegraph.com/news/inflation-bug-still-a-danger-to-more-than-
half-of-all-bitcoin-full-nodes
87
See https://en.bitcoin.it/wiki/Value_overflow_incident
88
CoinMetrics. “Don’t Trust, Verify: a Bitcoin Private Case Study” Dec. 23, 2018.
https://coinmetrics.io/bitcoin-private/
89
Zmudzinski, Adrian. “Stellar Patched an Inflation Bug and Burned the Resulting 2.25 Billion XLM:
Research” Cointelegraph (Mar. 27, 2019). https://cointelegraph.com/news/stellar-patched-an-inflation-
bug-and-burned-the-resulting-225-billion-xlm-research
20

Other networks that have faced potential inflation bugs but were not known to have been
exploited include Zcash 90 and Monero 91 – a particularly insidious threat when privacy-focused
chains are concerned, as the inflation is harder to detect on more opaque blockchains. Another
Bitcoin vulnerability patched in 2018 92 could have been used to create unexpected inflation but
was not exploited. Since DeFi protocols are highly automated, run continuously, and operate
with minimal (or in some cases, no) human oversight, inflation bugs on the underlying native
protocols can significantly destabilize DeFi applications. Inflation bugs are among the most
severe threats that blockchains face, and remediation often requires halting or rolling back the
blockchain, which would impair the assurances of any smart contracts relying on the underlying
blockchain. Recently, the DeFi-focused blockchain Kava was halted 93 to address a bug which
was significantly overpaying planned distributions (known as ‘yield farming’).
iii. Smart contract-based risks
a. Technical vulnerabilities of smart contracts
Smart contracts as described earlier are not legal contracts. Instead, they are code that automates
actions. These actions could be parts of native cryptocurrencies on public blockchains, such as
bitcoin and ether, which can be understood as synthetic commodity money 94 – they are not
guaranteed or backed by any third party and are not redeemable for anything, including fiat
currency. Instead, they serve as ‘access’ tokens to the Bitcoin and Ethereum networks,
respectively, and as a form of collateral and transactional medium within these networks.
For these native cryptocurrencies, it is fully possible to destroy, permanently immobilize, or

render them unspendable. While some exploits (like the June 2016 DAO Hack on Ethereum 95 or
the August 2010 Value Overflow Incident on Bitcoin 96) represent such an existential threat to the
90
Hackett, Rober. “Zcash Discloses Vulnerability That Could Have Allowed ‘Infinite Counterfeit’
Cryptocurrency” Fortune (Feb. 5, 2019). https://fortune.com/2019/02/05/zcash-vulnerability-
cryptocurrency/
91
Luigi1111 and Riccardo “fluffypony” Spagni. “Disclosure of a Major Bug in CryptoNote Based
Currencies” Monero (May 17, 2017). https://www.getmonero.org/2017/05/17/disclosure-of-a-major-bug-in-
cryptonote-based-currencies.html
92
Hertig, Alyssa. “The Latest Bitcoin Bug Was So Bad, Developers Kept Its Full Details a Secret”
Coindesk (Sept. 21, 2018). https://www.coindesk.com/the-latest-bitcoin-bug-was-so-bad-developers-kept-
its-full-details-a-secret
93
Harper, Colin. “Kava Halted After Yield Farming Bug Discovered in Latest Release” Coindesk (Mar. 4,
2021). https://www.coindesk.com/kava-halted-yield-farming-bug
94
See, Selgin, George, Synthetic Commodity Money (April 10, 2013). Available at SSRN:
https://ssrn.com/abstract=2000118 or http://dx.doi.org/10.2139/ssrn.2000118
95
Siegel, Dan. “Understanding the Dao Attack” Coindesk (Jun 25, 2016).
https://www.coindesk.com/understanding-dao-hack-journalists
96
See, Value overflow incident. Accessed April 13, 2021. https://en.bitcoin.it/wiki/Value_overflow_incident
21

network that they have been remediated with recourse to social consensus (overriding the
technical reality of the blockchain), the vast majority of exploits do not reach a critical threshold
of importance. Therefore, users who interact with faulty smart contracts could risk losing all of
their coins and are generally unable to obtain bailouts or recourse.
Perhaps the largest unremediated failure of a smart contract was the immobilization of 513,774
ether held in “multi-signature” (“multi-sig”) wallets written by Parity, 97 an Ethereum
development organization. Common multi-sig setups involve 2-of-3 or 3-of-5 schemes; the
former would allow outputs to be spent if valid signatures from any two of three predetermined
keys were provided.
The multi-sig wallets were exploited by an anonymous user who triggered a function in a smart
contract, effectively causing each wallet to self-destruct, irredeemably immobilizing the ether
contained within. The newly-locked ether – equivalent to 0.52% of all the ether in circulation at
the time – was worth $174 million at the time of the hack and $1.175 billion at the time of
writing. 98 Because Ethereum and other smart contract-enabled blockchains are more expressive 99
and permit more complex transactional logic, such failures are unavoidable. In this case, the
faulty multi-sig wallets were produced by an organization run by Gavin Wood, one of the
cofounders of Ethereum and the inventor of Solidity (Ethereum’s dedicated programming
language). The difficulty of writing a truly safe multi-sig contract illustrates the inherent risk
involved in transacting with digital bearer assets on expressive – and hence, vulnerability-prone
– base layers.
Moving beyond custodial risks, more complex interactions between smart contracts as required
by DeFi protocols can introduce additional scope for potential vulnerabilities. DeFi is rife with
purely technical vulnerabilities, owing to the complexity of interactive blockchain-based smart
contracts and the difficulty of anticipating complete edge cases before deploying code. Smart
contracts, once deployed, are cumbersome to upgrade, creating significant initial burdens on
developers. In the case of certain irrevocable smart contracts - like Uniswap, developers have no
ability to take down a smart contract once it is deployed. Upgrading such a smart contract would
be a matter of deploying an alternative and persuading users to use it. As long as the underlying
Ethereum blockchain remains intact, certain classes of smart contracts will remain operable
regardless of administrator or user behavior. In certain other types of smart contracts,
administrators can insert provisions into the code of their smart contracts so they can be
upgraded, terminated, or deprecated. These code provisions grant developers additional
discretion and recourse if there are bugs in deployed contracts. However, this has the externality
97
Parity Technologies. “A Postmortem on the Parity Multi-Sig Library Self-Destruct” (Nov. 15, 2017).
https://www.parity.io/a-postmortem-on-the-parity-multi-sig-library-self-destruct/
98
Figures current as of April, 12, 2021
99
The wider the set of computational concepts that can be expressed with base layer transactions the
more expressive a blockchain is.
22

problems of potentially making administrators responsible for user funds as well as making the
entities controlling the administrative keys a target for attackers.
Technical exploits are common: Werner et al (2021) 100 identified 21 such attacks on DeFi
protocols between February to December 2020, costing users an aggregate of $144.3 million
(USD value at the time of exploit) – although in some cases, funds were returned by attackers.
These exploits are varied in their approach, taking advantage of reentrancy bugs, “transaction
sandwiches,” logical bugs, and governance. In each case, however, attackers take advantage of
the properties of DeFi: predictable algorithms managing large pools of capital with limited
human oversight, built on blockchain rails. The rigidity of certain DeFi primitives, like
Automated Market Makers, can facilitate many of these attacks. The natively on-chain nature of
the collateral – in the case of ether – grants attackers the ability to withdraw their profits with no
recourse. When the captured tokens include stablecoins or other assets that are the liability of a
third party (such as exchange tokens), the tokens can be frozen.
b. Oracle attacks
One class of vulnerabilities deserving special attention relates to failures resulting from oracles.
In DeFi, oracles are service providers that provide outside information to a smart contract. The
most common usage of oracles is to transmit market prices, drawn from one or many exchanges,
to a DeFi protocol that relies on outside pricing information. For example, protocols employing
tokens as collateral would need to know the value (in standard terms like USD) of the pledged
tokens and employ smart contracts that consume oracle-provided market information. 101
A number of DeFi protocols rely on oracles, and the price inputs are critical to trigger
liquidations, deleveraging, margin calls, and other forms of automated collateral management.
Consequently, manipulation of oracles can be catastrophic for these protocols. It would be
somewhat comparable to what would happen in traditional finance if Bloomberg were hacked
and data were manipulated / could no longer be trusted. 102 Due to the sensitivity of these
protocols to deviations between the spot price of an asset and an index price (opening up riskless
arbitrage opportunities), so-called ‘oracle attacks’ are among the most popular means of attack.
Similar to strategies that involve manipulating the spot reference price for a derivative, oracle
attacks involve manipulating the market price of collateral referenced by a DeFi protocol in order
to create riskless arbitrage or to trigger liquidations.
100
Werner, S., Perez, D., Lewis, G., Klages-Mundt, A., Harz, D., and Knottenbelt, W. “SoK: Decentralized
Finance (DeFi) arxiv.org (Mar. 2, 2021). arXiv:2101.08778v2 [cs.CR] 2 mar 2021
https://arxiv.org/pdf/2101.08778.pdf
101
Note: oracle issues are not unique to crypto, but are harder to remediate because of transaction
finality and lack of contractual priority among parties.
102
See https://www.mas.gov.sg/-/media/MAS/resource/publications/fsr/FSR-2018.pdf (box C).
23

As Liu et al (2020) 103 note, oracles introduce risk in a number of ways: their mechanics are
opaque and unaccountable; they introduce critical nexuses of trust and dependency into DeFi,
and malicious oracles can cause catastrophic harm. The authors find repeated operational failures
in the methodological approaches to aggregating data across multiple exchange venues,
introducing operational risks and producing poor outputs.
As pointed out by Werner et al (2021), 104 market dislocations at spot exchanges feed into oracles
and affect DeFi systems built atop these price feeds. When the thinly-traded stablecoin Dai
briefly traded at $1.30 (it is typically pegged to $1) on Coinbase, this unnaturally high premium
was fed into the Compound protocol’s price feed, leading Compound to automatically decree
that a number of accounts were in default and programmatically deleverage and liquidate $88
million worth of collateral. 105 These wrongful liquidations occurred because of poor index
construction on the part of Compound (lacking controls for tail events) combined with the false
assumption that Dai would not trade at a significant premium on referenced markets.
c. Excessive leverage: smart contract-based flash loans
Certain idiosyncratic features of DeFi introduce attack vectors that are entirely novel. Among
these is the flash loan concept. Proposed in 2020 by DeFi lender Aave, the flash loan is an
unsecured loan permitting a borrower to access an unlimited amount of liquidity (up to the size
of the loan pool) 106 with a very low interest rate. 107 The catch is that the loan must be paid back
within the same transaction that it is taken out. Since DeFi applications give rise to frequent
arbitrage opportunities, such short-term loans allow individual parties with limited access to
capital to obtain leverage and take advantage of mispricings as long as transactions can be
executed atomically (i.e., simultaneously). As transactions on Ethereum can invoke many
contracts synchronously, flash loans are a useful tool in inter-contract arbitrage, as described by
Wang, et al (2021). 108
103
Liu, B., Szalachowski, P., and Zhou, J. “A First Look into DeFi Oracles” arxiv.org (11 Dec 2020).
https://arxiv.org/pdf/2005.04377.pdf
104
Werner, et al (2021). https://arxiv.org/pdf/2101.08778.pdf
105
Khatri, Yogita. “DAI price increase led to a massive $88 million worth of liquidations at DeFi protocol
Compound” The Block (Nov. 26, 2020). https://www.theblockcrypto.com/post/85850/dai-compound-dydx-
liquidations-defi
106
The Aave flash loan pool, as of Apr. 12, 2021, offers over $2.5b worth of liquidity. See,
https://aavewatch.com/
107
Aave FAQ, “Flash Loans.” Accessed Apr. 13, 2021. https://docs.aave.com/faq/flash-loans
108
Wang, D., Wu, S., Lin Z., Wu, L., Yuan, X., Zhou, Y., Wang, H., and Ren, K. “Towards A First Step to
Understand Flash Loan and Its Applications in DeFi Ecosystem” arxiv.org (Mar. 3, 2021).
arXiv:2010.12252v2 [cs.CR] 3 Mar 2021 https://arxiv.org/pdf/2010.12252.pdf
24

Flash loans dramatically reduce the barriers to entry for potential attackers while increasing their
leverage, and hence, the financial impact of their attacks on DeFi. Ever since their introduction,
flash loans have become increasingly prevalent in DeFi attacks. Cao et al (2021) identified nine
separate instances between February and December 2020 in which attackers successfully
siphoned a total of $49.58 million (USD value at the time of exploit) from DeFi protocols
through flash loan-assisted exploits. 109 The largest of these, the Harvest Attack in October 2020,
saw the attackers extract $26 million from Harvest, using the Curve and Uniswap protocols
while relying on a flash loan from Uniswap v2. While flash loans could be a helpful tool, they
can be misused to dramatically empower would-be attackers by making trial and error costs
cheap and granting near-unlimited leverage – provided that transactions can be constructed so
that the loan is paid back instantly.
iv. Other governance and regulatory risks
a. Administrative key abuse
Many DeFi protocols retain the discretionary option for administrative teams or other entities to
shut them down, upgrade them, pause the contract, and in some cases, drain user funds. There
are a few exceptions, like Uniswap, which simply exists as deployed code on Ethereum that
users can freely choose to interact with. The Uniswap contracts themselves cannot be paused by
the developer team. 110 The vast majority of protocols, however, do retain some form of a control
feature, including kill switches. In some cases, critical smart contract decisions are delegated to
the community of token holders (although in practice this collapses back to granting decision-
making power to a small number of insiders and backers as voting weight is typically
proportional to one’s share of tokens held). Additionally, since tokens are generally available on
the open market (which trades 24/7 on DEXs with no identity verification), an attacker could
freely purchase or borrow tokens in order to influence a token holder vote. Thus, many projects
for which token holder votes can influence the contract choose to retain de facto control by
directly limiting the free float of tokens. As the St Louis Fed notes regarding admin keys, “If the
keyholders do not create or store their keys securely, malicious third parties could get their hands
on these keys and compromise the smart contract. Alternatively, the core team members
themselves may be malicious or corrupted by significant monetary incentives.” 111
A common practice in mitigating admin key risks is granting a consortium of delegates control
over critical smart contract decisions by distributing power over key-related decisions into a
109
Cao, Y., Zou, C. and Cheng, X. Shanghai Wanxiang Blockchain Inc. “Flashot: A Snapshot of Flash
Loan Attack on DeFi Ecosystem” arxiv.org (Feb. 1, 2021). https://arxiv.org/pdf/2102.00626.pdf
110
Adams, H., Zinsmeister, N. and Robinson, D. “Uniswap v2 Core” White Paper. (Mar. 2020).
https://uniswap.org/whitepaper.pdf
111
Schär (2021), supra.
25

multi-sig setup. Some other controls exist, including enforced timelocks on key-related decisions
as is the case with Yearn, 112 or by granting signatories a limited, pre-specified set of powers, as
done by Synthetix. 113
The existence of admin keys in the majority of high-profile, active DeFi projects raises a number
of risks. Chiefly, these include key loss, insider theft of deposits, theft through extortion or hacks
from outside parties, and regulatory pressure. In many setups, including that of Synthetix,
contracts can be unilaterally frozen for a period by insiders as a precautionary ‘rapid response’
mechanism in the case of a hack or exploit. However, while merely pausing a contract does
limited harm, it could adversely affect liquidity.
Ultimately, assets held in contracts mediated by admin keys should be understood as custodial
rather than wholly sovereign interactions between users and a protocol. Adding more signatories
to a multi-sig key setup simply means that user deposits are held in the custody of a consortium
of insiders, rather than by one entity.
Research into the presence of admin keys is sparse because practices are constantly evolving.
While major DeFi projects have made efforts to mitigate key man risk and eliminate critical
points of centralization with regards to admin keys, independent researchers have nevertheless
identified DeFi applications where administrators have unilateral control over user funds. 114
Today, the major DeFi protocols Synthetix, Yearn, Dharma, SushiSwap, Badger, Harvest, and
Ren – containing user deposits of $10.58 billion collectively as of Apr. 12, 2021 115 – maintain
admin keys enabling a discretionary freeze of user funds.
While the power vested in admin keys varies, in some cases, anonymous individuals retain the
right to siphon all liquidity allocated to contracts they administer. Notably, the anonymous admin
behind Harvest Finance, which once held over $1 billion in user deposits in its smart contract,
can drain user funds encumbered only by a 12-hour timelock. 116 Projects like these, while
deployed on public chains, are dubiously decentralized, as they are functionally indistinguishable
from centralized asset-taking institutions (albeit not regulated as such).
112
Coopahtroopha, Supra. https://gov.yearn.finance/t/yfi-minting-ownership/155
113
Synthetix. “Synthetix Foundation Decommissioned” (Jul. 28, 2020). https://blog.synthetix.io/synthetix-
foundation-decommissioned/
114
See, Blec, Chris. “The Trustlessness of DeFi’s Top 10 Richest Products” Surviving DeFi (Nov. 23,
2020). https://survivingdefi.substack.com/p/the-trustlessness-of-defis-top-10 See also, Mapperson,
Joshua. “How many DeFi projects still have ‘God Mode’ admin keys? More than you think” Cointelegraph
(Sept. 25, 2020). https://cointelegraph.com/news/how-many-defi-projects-still-have-god-mode-admin-
keys-more-than-you-think
115
See https://defipulse.com/ Accessed April 12, 2021.
116
Blec, Chris. “Hunting Harvest’s Admin Key” Surving DeFi (Oct. 23, 2020).
https://survivingdefi.substack.com/p/hunting-harvests-admin-key
26

Some efforts exist to quantify and monitor the risks that users face from the existence of admin
keys: the DeFi Watch project, for instance, is a crowdsourced community project that monitors
the presence of admin keys in DeFi systems and evaluates their trustlessness. 117
Disclosure of administrative powers by core teams, token holders, and other entities has been
poor. For most projects, the scope of powers afforded by admin keys remains opaque, as the
developers creating these contracts seek to avoid the perception that they have control over user
balances. 118
b. Governance attacks
As more blockchain-based projects aspire to transform corporate business models by undertaking

a decentralized governance model, they introduce new risks. In practice, development teams
have been sluggish to delegate genuine decision-making power over development decisions and
system parameters for keys, which has meant that few governance attacks have been observed
thus far. However, should regulators see through the veil of decentralization 119 erected to
obfuscate the true nexuses of control in DeFi protocols, they would see that certain development
teams have sought to distribute governance power to holders of “governance tokens.” These
governance tokens endow their holders with and often a claim – albeit, frequently a diffuse one –
on cash flows or fees generated by these systems, as voting power over system parameters.
Typically, these have been managed, limited experiences, whereby governance token holders
could not vote to, for instance, fire the core development team or redirect funding from the core
corporate entity or nonprofit managing the system.
As token holders assert themselves, however, and gain the capacity to be more activist investors,
new classes of governance attacks emerge. Activists could elect to exploit DeFi systems to
benefit token holders (through some established extractive mechanism) at the expense of the
users of these systems. 120 For example, one theoretical attack, according to Gudgeon, et al
(2020), 121 would be for a governance attacker to gain control of the MakerDAO system and
siphon off capital from the system. As governance tokens become more available for short-term
liquidity – particularly through flash loans (discussed above), activists can more easily exploit
governance token votes to manipulate system parameters. After interacting with the Gudgeon-led
117
See, https://defiwatch.net/
118
Walch, Angela. “Deconstructing ‘Decentralization’: Exploring the Core Claim of Crypto Systems”
SSRN (Feb. 13, 2019). https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3326244
119
Id.
120
See, LongForWisdom.“[Urgent] Flash Loans and securing the Maker Protocol” makerdao.com (Oc.
2020). https://forum.makerdao.com/t/urgent-flash-loans-and-securing-the-maker-protocol/4901
121
Gudgeon, L. Perez, D., Harz, D., Livshits, B., and Gervais, A. “The Decentralized Financial Crisis”
arXiv.org (Feb. 19, 2020). https://arxiv.org/abs/2002.08099
27

research team, 122 MakerDAO noted the potential usage of flash loans to influence the outcome of
governance votes, and the MakerDAO community consequently voted on subsequent changes to
the system’s risk parameters. 123
As noted with the STEEM/Hive case study (mentioned above in the subsection on Validator
Cartels), tokens held by exchanges on behalf of users have been employed to influence
governance outcomes, in some cases against the wishes of these users. Large caches of
governance-laden tokens sitting at exchanges could influence them to accept bribes in order to
vote favorably on specific proposals or simply could be borrowed on an extremely short-term
basis (likely not impairing the exchange’s liquidity requirements) to swing a governance vote.
For instance, exchanges would naturally look to monetize user tokens held on deposit via flash
loans because the flash loans do not impair their ability to process withdrawals (because the term
of flash loans is literally zero). Meanwhile, the borrower only needs to hold tokens for the
duration of an on-chain vote in order to influence the outcome.
c. Tainted liquidity
At its core, DeFi envisions novel ways to undertake financial transactions. The cryptographic
nature of digital assets permits increasingly sophisticated and intricate schemes for managing
custody and transactional workflows. Bitcoin, for instance, offers a native multi-sig
functionality, whereby transactions can specify advanced conditions required for an output to be
spent. As of the time of writing, there exists a lower bound of 900,000 BTC (worth $56 billion at
the time of writing) 124 held in known multi-sig setups. 125
Thus, a novel class has emerged of custodians that maintain keys as a service, allowing
individuals and entities engaging in self-custody to take advantage of the sovereign nature of
holding one’s keys while maintaining the possibility of recourse in the case of key loss. A
common collaborative custody model involves a client holding a key in a “hot wallet,” a third-
party custodian holding one key, and a third key held for recovery, with two keys required for a
valid spend.
Engaging in collaborative blockchain transactions, however, can cause custodians to incur

liability from regulators such as the Office of Foreign Asset Control (OFAC). Indeed, BitGo,
122 See fn. 119, p.2, Gudgeon (2020), supra: ([“The research team”] engaged in a process of responsible
disclosure with Maker … who since modified their governance parameters to mitigate the two attack
strategies we present”).
123 See, MakerDAO forum discussion. https://forum.makerdao.com/t/signal-request-should-we-have-
another-executive-vote-regarding-the-governance-security-module/1209 See also, MakerDAO executive

vote to Set Governance Delay Module, which passed in February 2020.
https://beta.mcdgov.info/yay/0xd24fbbb4497ad32308bda735683b55499ddc2cad
124
Figures current as of Apr. 13, 2021
125
See, https://txstats.com/dashboard/db/p2sh-repartition-by-type?orgId=1 Accessed April 13, 2021.
28

which provides key management as a service in multi-sig transactions, was sanctioned by OFAC
for providing such services to clients in OFAC-sanctioned Crimea, Cuba, Iran, Sudan, and
Syria. 126 Bitcoin payment processor BitPay also settled similar charges with OFAC. 127
Additionally, the Financial Action Task Force (FATF) recently clarified in their recently revised
draft guidance on virtual assets 128 that parties administering keys in a multi-sig setup risk being
considered virtual asset service providers (VASPs), which would subject these parties to
surveillance and disclosure obligations, as well as to Travel Rule compliance requirements. 129
According to Coin Center, the FATF draft guidance breaks with policy precedent from the
Financial Crimes Enforcement Network (FinCEN), which considers “only persons with
‘independent control’ over customer funds are treated as regulated money transmitters.” 130 The
revised FATF draft guidance would dramatically increase the scope of covered parties. 131
DeFi in its current form is largely incompatible with such regulations. Since most decentralized
contracts do not require any user identification beyond a valid blockchain address, there is
virtually no emphasis on centralized compliance. Products facilitating on-chain swaps, like
Uniswap, are simply blockchain contracts that permit users to collaboratively pool funds and
make trades with no central intermediary. 132 133 The nature of these “peer-to-pool” systems is
such that these contracts cannot meaningfully exclude any entity looking to participate in the
pooling, which is open and freely participatory by definition.
Uniswap, in particular, relies on “liquidity providers” that contribute assets to a pool in exchange
for fees. These are not designated entities; anyone can be a liquidity provider if they contribute
assets to the pool. At the time of writing, Uniswap v2 boasted 84,000 active liquidity providers
with 5,400 liquidity providers active in the most popular pair, UNI-WETH. 134 If some tainted
liquidity, for instance emanating from an OFAC-sanctioned party or an illicit source, were to
enter a Uniswap pool, regular users would effectively be undertaking a financial relationship
126
https://home.treasury.gov/system/files/126/20201230_bitgo.pdf
127
https://www.coindesk.com/bitpay-to-pay-500k-to-settle-ofac-sanction-violation-charges
128
FATF “Draft updated Guidance for a risk-based approach to virtual assets and VASPs” (March 2021).
https://www.fatf-gafi.org/media/fatf/documents/recommendations/March%202021%20-
%20VA%20Guidance%20update%20-%20Sixth%20draft%20-%20Public%20consultation.pdf
129
See, Van Valkenburgh, Peter. “A quick analysis of FATF’s 2021 draft cryptocurrency guidance” Coin
Center (Mar. 22, 2021). https://www.coincenter.org/a-quick-analysis-of-fatfs-2021-draft-cryptocurrency-
guidance/
130
Id.
131
Specifically, the recent FATF draft guidance notes in para. 54 that “[t]his control [over client assets]
does not have to be unilateral and multi-signature processes are not exempt” when evaluating whether an
entity should be considered a VASP.”
132
For an introduction to Uniswap, see, Angeris, G., Kao, H., Chiang, R., Noyes, C., and Chitra, T. “An
analysis of Uniswap markets” arxiv.org (Nov. 2019). https://arxiv.org/pdf/1911.03380.pdf
133
Note: Aave’s new permissioned pool is an exception.
134
Source: https://explore.duneanalytics.com/dashboard/uniswap-community as of April 13, 2021.
29

with these prohibited parties. As currently deployed, the smart contract has no means to whitelist
users or permission them a priori. The very nature of decentralized finance on public
blockchains like Ethereum is to facilitate permissionless exchange, but this open access is
generally incompatible with anti-money laundering/combating the financiang of terrorism
(AML/CFT) regulations as currently implemented in the U.S.
d. Pseudo-equities - regulatory uncertainty
Transactions that involve lending, investment trading, and derivative exposure are regulated in
traditional financial markets through the registration, licensing and examination of intermediaries
that broker, custody, clear or otherwise facilitate such transactions. In DeFi, intermediaries are
largely excluded in favor of transparent code, presenting regulators and policymakers with
complicated decisions as to how to assess transactions (often bilateral) for which no clearly
identified party may be regulated. These important issues regarding the regulatory uncertainty of
the underlying commercial transactions that are conducted through DeFi protocols are beyond
the scope of this paper. This sub-section, however, examines more specifically so-called
‘pseudo-equities’ and their inherent regulatory risks.
Despite considerable regulatory risks of issuing pseudo-equity tokens with little regard for the
requirements of securities law, many DeFi protocols are administered by U.S.-based firms or
nonprofits. 135 In many cases, these entities finance themselves through the issuance of a token
that represents a claim on some cash flows produced by the system. These tokens have proven to
be a meaningful financing vehicle for developing DeFi protocols. As the time of writing, the
aggregate market capitalization of tokens in the “decentralized finance” space is $85 billion, with
Uniswap, Synthetix, and Compound (collectively having taken in $12.29 billion 136 in allocated
collateral) being the largest pseudo-equity tokens. Many of these DeFi tokens endow token
holders with some rudimentary governance rights as well as either implicit or direct claims on
cash flows generated through DeFi protocols. None of these pseudo-equity tokens backstopping
DeFi are registered as securities, circulating instead on decentralized financial infrastructure like
Uniswap (and in some cases, on centralized crypto exchanges). If securities regulators deemed
such pseudo-equity tokens to be unregistered securities and pursued not only their issuers and
promoters but also the venues upon which they trade, the financing and governance model of
these DeFi projects would be significantly impaired. Additionally, numerous DeFi protocols
subsidize their liquidity by issuing new units of pseudo-equity to end users. If these tokens were
to be delisted and their liquidity and value suffered losses, the utility of these subsidized
protocols would decline. These token incentives built into DeFi protocols are the equivalent of
135
Precedent exists for SEC enforcement actions regarding U.S.-based entities administering smart
contracts, including the case of EtherDelta, in which the founder administered frontend smart contracts
governing token trading on the Ethereum network, is best known. The SEC considered EtherDelta an
unregistered national securities exchange. See, https://www.sec.gov/news/press-release/2018-258
136
Figures taken from https://defipulse.com/. Current as of April 12, 2021.
30

Uber compensating drivers for each mile driven with incremental units of Uber equity. As an
example, the compensation for supplying the stablecoin USDC to the money-market protocol
Compound is 6.71% annualized at the time of writing, supplemented by a 2.15% annualized
payout in COMP terms to suppliers of USDC. The combination of the two is described as the
‘net rate’ for USDC by Compound. 137 If these incentives were to expire or be withdrawn, interest
rates would look significantly less attractive, reducing the incentive for liquidity providers to put
their capital at risk.
Virtually all DeFi protocols require oversight, bug remediation, technical and economic audits,
governance stewardship, and leadership and direction from these administrative entities. Even if
there are no corporations or firms officially underwriting these decentralized protocols, virtually
all of these protocols have an entity, whether codified or not, effectively managing the protocol.
The elimination of the pseudo-equity token as a viable financing mechanism would significantly
impair the industry’s ability to operate. It is possible, however, that corporations could create the
majority of decentralized finance contracts and monetize them without the use of a token by
directly charging rents for usage of the DeFi contract, or that anonymous developers could
deploy these protocols to the blockchain.
v. Challenges associated with scalability
Scalability – the general process whereby public blockchains grow to handle an economically
meaningful volume of activity and more transactional data without compromising their
assurances – is held as one of the chief difficulties facing blockchains today. Adding more data
to the final ledger trades off against the computational difficulty of operating a full node and
staying current on the ledger. While no silver bullet solution to scalability of blockchain exists,
since the basic security model of blockchains requires that all participants store a full copy of the
state, various improvements have been proposed. 138
Approaches such as Bitcoin’s Lightning Network envision a network of payment channels with
only periodic final settlement to the Bitcoin layer itself. 139 Sidechains immobilize bitcoins (or
other native units) and create a subledger whereby claims on those Bitcoins can circulate
137
https://compound.finance/markets/USDC
138
For a more complete overview of scalability solutions, see Kim S., Kwon, Y., Cho, S. “A Survey of
Scalability Solutions on Blockchain” 2018 International Conference on Information and Communication
Technology Convergence (ICTC) (Oct. 17-19, 2018).
https://ieeexplore.ieee.org/abstract/document/8539529
139
Poon, Joseph and Dryja, Thaddeus. “The Bitcoin Lightning Network: Scalable Off-Chain Instant
Payments” Draft Version 0.5.9.2 (Jan . 14, 2016. http://lightning.network/lightning-network-paper.pdf
31

frictionlessly, effectively creating a new transactional space. 140 141 Sharding splits the state of the
blockchain into parts, with nodes finding consensus on a subset of the final state, periodically
reconciling with each other. 142 143 More creatively, rollups (primarily envisioned for
computation-heavy blockchains like Ethereum) bundle transactions, moving computation on-
chain, but retaining final transactional data on chain. The validity of these transactions is assured
through zero-knowledge proofs or mechanisms known as fraud proofs. 144
The commonality around all these approaches is transactional parsimony, or the reduction of a
transaction into as few final bytes as possible, since the costs associated with storing and
processing transactional data is the chief externality of blockchain transactions. Unless radically
new models for public blockchains are developed, the problem of scalability will be an inherent
constraint, as it follows naturally from the requirement that nodes must ingest and verify the
global state in order to become full participants on the ledger. Additionally, these approaches all
generally aim to defer final settlement by distinguishing a payment or financial message and the
settlement of that message or bundle of transactions. This deferred settlement mechanism will be
familiar to anyone with knowledge of established payment systems, but public blockchains have
only just begun to explore their implications.
As mentioned, public blockchains in their current form are effectively deterministic, single state
environments 145 in which each peer must process each and every transaction in order to stay in
sync in a trustless manner with the blockchain’s current state. Both Bitcoin and Ethereum have
committed to operational limits on the most data throughput that either system can handle, with
Ethereum adopting a looser constraint. These limits roughly correlate with the number of
transactions that consumer hardware can meaningfully process without falling behind. In order to
prioritize transactions, both systems employ fees. Naturally, some classes of users are more
willing to bear fees than others, with fee sensitivity generally being a function of the perceived
140
Back, A., Corallo, M., Dashjr, L., Friedenbach, M., Maxwell, G., Miller, A., Poelstra, A., Timon, J., and
Wuille, P. “Enabling Blockchain Innovations with Pegged Sidechains” (Oct. 22, 2014) (commit 5620e43).
http://kevinriggen.com/files/sidechains.pdf
141
A “sidechain” is a subledger where transactions occur off-ledger, and are periodically settled to the
base layer itself (blockstream paper on sidechains).
“Roll-ups” similarly involve performing computation off-ledger, then periodically posting the results of that
computation to the blockchain itself.
142
Schaffner, Tobias. “Scaling Public Blockchains: a comprehensive analysis of optimistic and zero-
knowledge rollups” University of Basel (Jan. 14, 2021).
https://wwz.unibas.ch/fileadmin/user_upload/wwz/00_Professuren/Schaer_DLTFintech/Lehre/Tobias_Sch
affner_Masterthesis.pdf
143
“Sharding” is a process whereby the ledger state is partitioned and transfers occur locally within the
shard, and are periodically reconciled with the global state.
Each model involves effectively partitioning state in order to gain transactional efficiency and
subsequently reconciling it with the global ledger
144
Schaffner, supra.
145
‘Single-state’ means that all participants share the same ledger. ‘Determinism’ is the property whereby
the same inputs should always lead to the same outputs.
32

importance of a transaction. The consequence of this approach means that, in their current
format, heterogeneous demand requires that some applications are naturally priced out at any
given time. Applications of Ethereum are varied, running the gamut from using stablecoins for
remittances, to operating decentralized organizations, to minting non-fungible tokens that
represent unique pieces of artwork, to financial use cases like obtaining programmatic leverage.
Since all of these applications are competing for the same finite pool of blockspace, a burst of
adoption for one use-case can degrade the experience of using another (due to the effect of fee
hikes on the market for blockspace). High fees have the effect of pricing out smaller transacting
parties, who might deem an operation uneconomical if fees hit a certain threshold.
Due to relatively inelastic blockspace combined with volatile demand for blockchain resources,
fees are highly volatile. For instance, on February 23, 2021, mean per-transaction Ethereum fees
reached $38 while mean Bitcoin fees reached the equivalent of $25. 146 For comparison, in 2019,
Bitcoin and Ethereum per-transaction fees averaged the equivalent of only $1.24 and $0.13, 147
respectively. 148 In addition to the general drag that fees introduce on transactional usage,
blockchain congestion can facilitate specific attacks against DeFi protocols, which sometimes
need to execute transactions within a specific period. On March 12, 2020, for instance, the
MakerDAO system became insolvent when third-party participants, known as “keepers,” were
able to bid $0 during on-chain liquidations of ether, even though the ETH market price was
fluctuating between $80 and $120 that day. In addition, the Ethereum blockchain was
experiencing significant lags during which transactions were not processed in time due to high
fees, 149 precluding other keepers from participating in the “$0 bidder” auctions. This series of
events affected liquidations of roughly $8 million of collateral backing users’ debt positions in
the MakerDAO system, leaving the Dai stablecoin undercollateralized by $4.5 million.
Additionally, as fees rise on blockchains, they price out certain classes of activity, especially
more computationally expensive actions (particularly for blockchains like Ethereum where the
cost to transact is a function of computational demand). This systematically prices out users with
smaller balances and has the net effect of trapping funds held in accounts (or UTXOs, in the case
146
Coin Metrics. Accessed Apr. 12, 2021. https://network-charts.coinmetrics.io/
147
See, https://bitinfocharts.com/comparison/ethereum-transactionfees.html
148
And for further comparison - wire transfer fees generally range from $15-$30. Credit card processing
fees range from 1.43 to 3.5 percent of the transaction amount. For Venmo, zero transaction fees for
transfers to linked bank account, debit card or Venmo account, but if Venmo users want to access cash
earlier than 1-2 days, then instant transfers from Venmo to linked debit card requires a fee of 25 cents or
1% of the total transfer, whichever number is higher. https://www.mybanktracker.com/news/wire-transfer-
fee-comparison-top-10-us-banks , https://www.bankrate.com/finance/credit-cards/merchants-guide-to-
credit-card-processing-fees/ , https://www.credit.com/blog/the-app-your-kids-are-using-to-pay-people-
back-venmo/
149
See, Eichholz, Liesl. “What Really Happened to MakerDAO” Glassnode.com (Mar. 17, 2020).
https://insights.glassnode.com/what-really-happened-to-makerdao/ See also, Blocknative. “Evidence of
Mempool Manipulation on Black Thursday: Hammerbots, Mempool, Compression, and Spontaneous
Stuck Transactions. (July 22, 2020). https://www.blocknative.com/blog/mempool-forensics
33

of Bitcoin), stranding those assets. As fees rise on the base layer, retail users can no longer
economically engage in DeFi operating on the base layer, affecting liquidity in decentralized
exchanges. 150 In Bitcoin, a negative feedback loop between transactional usage and fees was
evident in 2017-18, indicating that fees not only adversely affect users’ affinity to transact, but
they do so in an inherently unstable way, without finding equilibrium.151
DeFi protocols depend critically on seamless interoperability as well as on composability. 152

Gudegon, et al (2020) stress the importance of composability on these systems: “Assets that are
created in Maker, for example, can be used as collateral in other protocols such as Compound,
dY/dX, or in liquidity pools on Uniswap.” Thus, mitigating the impact of base-layer fees by
creating subledgers or deferred settlement processes like sidechains, rollups, or sharding does
little to solve the problem at its core. Deferring settlement introduces efficiencies by engaging in
periodic reconciliation with the base layer itself. However, DeFi as currently envisioned depends
on funds being available on the base layer and contracts being able to seamlessly communicate
and refer to each other. Thus, standard approaches to scaling appear to compromise the desirable
qualities of DeFi and do not represent a panacea to the fee drag. It remains to be seen whether the
standard approach to scaling blockchains – effectively mimicking the layered approach that
characterizes established payments systems – can be made consistent with the desirable qualities
of blockchains that distinguish themselves from these systems. If they cannot, then real estate on
the base layer of blockchains with capped throughput will be reserved only for well-capitalized
parties, which are able to outbid smaller users.
IV. Conclusion: “No Free Lunch”
The risks described in this chapter do not seek to provide a comprehensive list but to help readers
conceptually understand the drivers behind the risks inherent in DeFi. Many of the risks
described above stem from the decentralized nature of blockchains. The goal of automating the
delivery of financial services and reducing human dependencies also has the congruent effect of
reducing oversight and control. Disintermediating traditional intermediaries reduces high fees
and entry friction, but also creates new opportunities for new types of intermediaries. These new
150
Carter, Nic. “Public blockchain fee cyclicality and negative feedbakc loops” medium.com (Oct. 5,
2020). https://medium.com/@nic__carter/public-blockchain-fee-cyclicality-and-negative-feedback-loops-
1620141a8a87
151
More descriptive data regarding bitcoin fee dynamics can be found in Lehar, A. and Parlour, C.
“Liquidity Demand and BitCoin Transaction Fees” (May 2019) (preliminary and incomplete).
https://iwfsas.org/iwfsas2019/wp-content/uploads/2017/02/S4_P2.pdf
152
Composability refers to the assurance that a transaction being proposed which calls multiple distinct
contracts will execute, because each implies final settlement. Gudgeon, et al (2020) describe it as “the
ability to build a complex, multi component financial system on top of crypto-assets” while Wachter,
Jensen and Ross note that it limits the “role for central clearing counterparties in mitigating counterparty
risk.” https://arxiv.org/ftp/arxiv/papers/2102/2102.04227.pdf
34

types of intermediaries require the sufficient economic incentives and, thus, could be potentially
more costly and risky than the monopoly rents extracted by today’s centralized intermediaries.
Ultimately, this new host of intermediaries in a decentralized financial ecosystems could stymie
the drive toward the twin goals of democratizing financial services: lowering cost and improving
access.
External dependencies on traditional finance, namely banks, is another important source of risk
as well as a transmission channel for risk. Although one of the goals of DeFi is to create a new
kind of financial system without traditional intermediaries, the irony is that as DeFi struggles to
make itself more useful in the real world, its dependency on the established financial system
grows. Its reliance on traditional finance is not only a source of risk but can potentially serve as a
transmission channel for risk between traditional financial and DeFi systems. DeFi leveraging
stablecoins’ backing by fiat or other financial liabilities is an excellent example of this type of
risk. Another dependency is that the crypto industry still needs to bank with commercial banks in
order to conduct the cash legs of their transactions.
DeFi is developing in a direction different than originally intended and, thus, it is coming full
circle. The tools exist to wall off DeFi from the financial system, such as running everything on
native tokens DAI or ETH for instance. However, the opposite is occurring. Relying on
commercial banks for stablecoins, etc., is convenient and serves business needs.
Wholly new risks are introduced by DeFi stemming from the reliance on open protocols and the
fact that the underlying infrastructure is un-owned. Removing the back office and human
oversight results in many efficiencies, but they also introduce risks. Thus, it is up to the end user
or contract administrator to monitor the risk of the protocols themselves, and many would not
want the burden. These risks are amplified when financial primitives collide with automated,
hard-to-intervene contracts. Here is where all the chaos in DeFi is really from - systems that are
built to be scalable and automated but that are underspecified or not understood by their creators.
In sum, blockchain technologies bring many benefits. But the tools or processes used to
disintermediate or gain efficiency also have costs in recourse, reversibility, risk management, etc.
– the ‘paradox’ of DeFi.
In spite of this ‘paradox,’ DeFi is achieving something truly novel: it is facilitating the business
model experimentation and evolution in a very short time frame. While our traditional financial
system has evolved over centuries to reach our current institutional arrangements (i.e., banks,
financial markets, market participants, different types of regulation – each in response to past
crises and mis-steps), DeFi is allowing for significantly more rapid institutional innovation and
trial and error experimentation. There is a lot of error at the moment (such as speculative
bubbles, fraud, governance issues, etc.), but those innovations that persist over several years in
such a fast-moving environment could be much more robust. As DeFi becomes more
35

economically relevant, financial stability risks in DeFi could come not only from the links with
the traditional financial sector but also from the risks within the DeFi sector itself. More research
is needed to study how shocks in the DeFi sector can impact the greater financial system and the
real economy. There is a greater need than ever to research this understudied but increasingly
important area to help provide a better understanding of the evolution of financial services and
the risks inherent in DeFi for industry, regulators and policymakers.
36

Defi Protocol Risks

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Defi Protocol Risks

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Defi Protocol Risks

Uploaded by

Copyright:

Available Formats

DeFi Protocol Risks: the Paradox of DeFi 1

By Nic Carter 2 and Linda Jeng 3

Electronic copy available at: https://ssrn.com/abstract=3866699

Electronic copy available at: https://ssrn.com/abstract=3866699

i. Evolution of DeFi Movement to DeFi

Electronic copy available at: https://ssrn.com/abstract=3866699

Electronic copy available at: https://ssrn.com/abstract=3866699

3. Risk Factors in DeFi

Electronic copy available at: https://ssrn.com/abstract=3866699

Electronic copy available at: https://ssrn.com/abstract=3866699

Figure 1. DeFi Protocols: Map of Interconnected Risks

i. Interconnections with traditional financial system

a. Banks holding reserves backing stablecoins

Electronic copy available at: https://ssrn.com/abstract=3866699

i. Market risk in stablecoins’ underlying reserves

In November 2019, MakerDAO introduced “non-native” forms of collateral backing an upgraded

Electronic copy available at: https://ssrn.com/abstract=3866699

ii. Sources of market illiquidity

Electronic copy available at: https://ssrn.com/abstract=3866699

b. High interconnectedness: banking relationships with crypto trading

Electronic copy available at: https://ssrn.com/abstract=3866699

c. Retail exposure: consumer fintech apps

Electronic copy available at: https://ssrn.com/abstract=3866699

d. Corporate exposure: corporate treasuries

Electronic copy available at: https://ssrn.com/abstract=3866699

ii. Operational risks stemming from underlying blockchains

Electronic copy available at: https://ssrn.com/abstract=3866699

Consensus – the construction, approval and distribution of blocks of transaction across

b. Underlying protocol interventions

Electronic copy available at: https://ssrn.com/abstract=3866699

c. Proof of Work (PoW) consensus failures

Electronic copy available at: https://ssrn.com/abstract=3866699

Electronic copy available at: https://ssrn.com/abstract=3866699

d. Miner extractable value (MEV)

Electronic copy available at: https://ssrn.com/abstract=3866699

Electronic copy available at: https://ssrn.com/abstract=3866699

At present, MEV appears to be a fundamental feature of data-rich blockchains that facilitate

Electronic copy available at: https://ssrn.com/abstract=3866699

Electronic copy available at: https://ssrn.com/abstract=3866699

iii. Smart contract-based risks

a. Technical vulnerabilities of smart contracts

For these native cryptocurrencies, it is fully possible to destroy, permanently immobilize, or

Electronic copy available at: https://ssrn.com/abstract=3866699

Electronic copy available at: https://ssrn.com/abstract=3866699

Electronic copy available at: https://ssrn.com/abstract=3866699

c. Excessive leverage: smart contract-based flash loans

Electronic copy available at: https://ssrn.com/abstract=3866699

iv. Other governance and regulatory risks

a. Administrative key abuse

Electronic copy available at: https://ssrn.com/abstract=3866699

Electronic copy available at: https://ssrn.com/abstract=3866699

As more blockchain-based projects aspire to transform corporate business models by undertaking

Electronic copy available at: https://ssrn.com/abstract=3866699

Engaging in collaborative blockchain transactions, however, can cause custodians to incur

another-executive-vote-regarding-the-governance-security-module/1209 See also, MakerDAO executive

Electronic copy available at: https://ssrn.com/abstract=3866699

Electronic copy available at: https://ssrn.com/abstract=3866699

d. Pseudo-equities - regulatory uncertainty