Shadow Banking 2.0
Shadow Banking 2.0
Shadow Banking 2.0
Hilary J. Allen1
The growth of so-called “shadow banking” was a significant contributor to the financial crisis
of 2008, which had huge social costs that we still grapple with today. Our financial regulatory
system still hasn’t fully figured out how to address the risks of the derivatives, securitizations,
and money market mutual funds that comprised Shadow Banking 1.0, but we’re already facing
the prospect of Shadow Banking 2.0 in the form of decentralized finance, or “DeFi.” DeFi’s
proponents speak of a future where sending money is as easy as sending a photograph – but
money is not the same as a photograph. The stakes are much higher when money is involved,
and if DeFi is permitted to develop without any regulatory intervention, it will magnify the
tendencies towards heightened leverage, rigidity, and runs that characterized Shadow Banking
1.0.
Fortunately, though, there is still time to prevent DeFi from becoming Shadow Banking 2.0.
This Essay argues for precautionary regulation of DeFi, designed to limit its growth and to
cordon off whatever remains from the established financial system and real-world economy.
While proponents of DeFi will contend that this will limit innovation, this Essay argues that
DeFi innovation has limited benefits for society. DeFi doesn’t aspire to provide new financial
products and services – it simply aspires to provide existing financial products and services in
a decentralized way (meaning, without intermediaries). This Essay will demonstrate that the
DeFi ecosystem is in fact full of intermediaries and explain why full disintermediation of
financial services is an entirely unrealistic aspiration. This Essay will then proceed from that
finding to argue that if DeFi cannot deliver on decentralization, regulators should feel
emboldened to clamp down on DeFi in order to protect the stability of our financial system
and broader economy.
I. INTRODUCTION ............................................................................................................................. 2
II. SHADOW BANKING 1.0 ................................................................................................................ 4
A. Credit Default Swaps .......................................................................................................... 5
B. Mortgage-Backed Securities ............................................................................................... 6
C. Money Market Mutual Funds.............................................................................................. 7
III. DEFI........................................................................................................................................... 8
A. Introduction to DeFi ........................................................................................................... 9
B. DeFi as Shadow Banking 2.0 ............................................................................................ 10
i. Leverage........................................................................................................................ 10
ii. Rigidity .......................................................................................................................... 12
iii. Runs........................................................................................................................... 13
IV. HOW TO RESPOND.................................................................................................................... 15
A. The Cost-Benefit Calculus ................................................................................................ 17
1
Professor, American University Washington College of Law. Many thanks to participants in the William & Mary
Law Review’s Cryptocurrency Symposium for their comments and perspectives, as well as to Ryan Clements,
Stephen Diehl, and Todd Philips for helpful engagement with earlier drafts.
I. INTRODUCTION
“DeFi” or “decentralized finance” has become one of the hottest trends in finance in the
last few years. DeFi is usually discussed in aspirational terms, invoking comparisons to other
types of technological innovations: we frequently hear that DeFi will make sending money as easy
as sending a photograph, or an email.2 But money is not the same as photographs and emails – the
consequences of losing money (both for the affected individual and for confidence in the financial
system as a whole) are much greater than the consequences of a lost photograph or email. Because
money and finance are the lifeblood of our economy, finance has always been highly regulated in
a way that Kodak’s provision of photographs, and FedEx’s delivery of couriered letters, never
were.3
The existence of strong financial regulation has often spurred attempts to arbitrage it – and
that regulatory arbitrage is sometimes facilitated by complex financial innovation.4 That’s what
happened in the lead up to the 2008 crisis, when credit default swaps and mortgage-backed
securitizations evolved around existing financial regulation, just as money market mutual funds
had decades earlier (because these services provided functional equivalents for banking products
but operated outside the regulated banking sphere, they came to be known as “shadow banking”,
and this Essay will refer to them as “Shadow Banking 1.0”). Few steps were taken to rein in these
types of innovation, and the increased leverage, rigidity, and fragility they created became evident
during the 2008 financial crisis – only in the aftermath of that crisis did legislators and regulators
step up with some regulatory fixes (which have helped, but not fully addressed, the problems
associated with Shadow Banking 1.0).
The crisis of 2008 had searing social consequences. The recession that followed had
obvious and immediate impacts on employment and wealth, but it also generated a lingering
mental and physical toll for the most vulnerable members of our society.5 Nearly fifteen years
after the financial crisis of 2008, we are still learning more about the damage that the crisis caused:
recent work has focused on how the crisis has exacerbated inequality;6 another developing area of
literature considers the political repercussions of the crisis (and financial crises more generally),
suggesting that such crises can lead to political radicalization.7 The 2008 crisis was not inevitable,
2
For example, see “Ethereum makes sending money around the world as easy as sending an email.” Ethereum,
Decentralized Finance (DeFi), available at https://ethereum.org/en/defi/.
3
Carnell et al., THE LAW OF FINANCIAL INSTITUTIONS 7TH Ed., 3 (2021).
4
Saule T. Omarova, License to Deal, 90 WASH U. L. REV. 63, 70 (2012).
5
Janet L. Yellen, A Painfully Slow Recovery for America's Workers: Causes, Implications, and the Federal
Reserve's Response (speech dated Feb. 11, 2013), available at
http://www.federalreserve.gov/newsevents/speech/yellen20130211a.htm.
6
Bridges et al., Credit, Crises and Inequality, Bank of England Staff Working Paper No. 949 (Nov. 2021).
7
For a literature review (as well as findings that the severe banking crisis in Germany in 1931 not only led to
“broad-based political radicalization shortly thereafter; once the Nazis were in power, both pogroms and
Confidence in our traditional financial system (and the agencies that oversee it) was
justifiably shaken by the crisis of 2008; this has understandably piqued interest in visions of an
alternative decentralized financial system where no one needs to trust any intermediary because
intermediaries have been rendered superfluous. Unfortunately, this is an entirely unrealistic goal
– DeFi users have to trust in some combination of ISPs, core software developers, miners, wallets,
exchanges, stablecoin issuers, oracles, providers of client APIs used to access distributed ledgers,
and concentrated owners of governance tokens.9 In short, DeFi doesn’t so much disintermediate
finance as replace trust in regulated banks with trust in new intermediaries who are often
unidentified and unregulated. This Essay will argue that DeFi innovations that are supposed to
displace the need for trust in intermediaries succeed only in making DeFi more fragile than
traditional financial services.
I have posed this Article’s title, “DeFi: Shadow Banking 2.0?”, as a question. There is
already abundant evidence that DeFi mirrors and magnifies the fragilities of shadow banking
innovations that resulted in the crisis of 2008; the question is whether policymakers will allow
DeFi to grow and become sufficiently integrated with the established financial system that it can
cause widespread harm. This Essay argues that such an outcome is not inevitable: policymakers
should take a precautionary approach to DeFi regulation, limiting the use of DeFi where financial
regulators are able to exercise jurisdiction, and then cordoning off whatever DeFi remains from
the established financial system and real-world economy.
This approach will admittedly limit innovation in the DeFi ecosystem, but not all
innovation is good innovation: if the risks of innovation outweigh any possible benefits it might
have, then preventing that innovation is good public policy. In this context, it is important to
understand that DeFi does not purport to provide any new types of financial products or services
– it just aspires to deliver existing financial products and services in a decentralized way. Given
that decentralization is an entirely unrealistic goal, we are left with technology that may be
interesting from an academic perspective,10 but in practical terms is inefficient in its complexity
(and as a result, doesn’t respond well to the needs of those who are underserved by our existing
financial system). As such, policymakers would serve us well by taking preemptive steps to
prevent the growth of Shadow Banking 2.0.
This Essay will proceed as follows. Section II will provide an overview of Shadow
Banking 1.0, with a focus on the fragilities of credit default swaps, mortgage-backed
securitizations, money market mutual funds, and their contributions to the financial crisis of 2008.
deportations were more common in places more affected by the banking crisis”), see Sebastian Doerr et al.,
Financial Crises and Political Radicalization: How Failing Banks Paved Hitler’s Path to Power, BIS WORKING
PAPER No. 978 (Nov. 2021).
8
Financial Crisis Inquiry Commission, THE FINANCIAL CRISIS INQUIRY REPORT, xviii (2011).
9
See Section IV.A.i infra for elaboration.
10
Cryptographic consensus mechanisms are an elegant solution to the double spending program associated with
digital assets. Edmund Schuster, Cloud Crypto Land, 84 MODERN L. REV. 974, 988 (2020).
Following the financial crisis of 2008, a significant amount of academic and policy work
was done on “shadow banking”. Generally speaking, shadow banking describes financial
activities that are the functional equivalent of activities carried out in the regulated banking system,
but which escape bank regulation.11 Around the time of the crisis, shadow banking included “such
familiar institutions as investment banks, money-market mutual funds (MMMFs), and mortgage
brokers; some rather old contractual forms, such as sale-and-repurchase agreements (repos); and
more esoteric instruments such as asset-backed securities (ABSs), collateralized debt obligations
(CDOs), and asset-backed commercial paper (ABCP).”12 Because they facilitate new forms of
leverage outside of the banking system,13 credit default swaps are also considered part of the
shadow banking system.14 This Section of the Essay will use credit default swaps, as well as
money market mutual funds and mortgage-backed securitization (a particular type of asset-backed
securitization), to illustrate some of the fragilities that the first generation of shadow banking
introduced into the financial system.
Although these forms of shadow banking differ in many respects, one thread that unites
them is their complexity, which is a destabilizing force in and of itself. Complexity can make
financial products (and their possible interactions with the broader financial system) harder to
understand, increasing the chance that risks will go unanticipated.15 Even if risks are anticipated,
complexity-induced opacity increases the chance that such risks will be underestimated in good
times (causing bubbles), and overestimated in bad times (making panics worse).16 More generally,
there is a whole discipline of complexity science that considers how increased complexity can
make systems more fragile (particularly by obscuring how steps that are taken to make individual
system components more robust can end up transmitting problems with those components
throughout the broader system).17 Increased complexity writ large is certainly part of the shadow
11
“Shadow banks are financial intermediaries that conduct maturity, credit, and liquidity transformation without
explicit access to central bank liquidity or public sector credit guarantees.” Zoltan Pozsar et al., Shadow Banking,
FRBNY ECONOMIC POLICY REVIEW, 1 (Dec. 2013).
12
Gary Gorton & Andrew Metrick, Regulating the Shadow Banking System, BROOKINGS PAPERS ON ECON.
ACTIVITY, 261-2 (2010).
13
John Geanakoplos, Solving The Present Crisis and Managing the Leverage Cycle, 16 (Dec. 22, 2009), available at
https://fraser.stlouisfed.org/files/docs/historical/fct/fcic/fcic_report_geanakoplos_20100226.pdf.
14
Zoltan Pozsar et al., supra Note 11 at 4.
15
Dan Awrey, Complexity, Innovation and the Regulation of Modern Markets, 2 HARV. BUS. L. REV. 235, 250
(2012).
16
Nicola Gennaioli et al., Financial Innovation and Financial Fragility, FONDAZIONE ENI ENRICO MATTEI
NOTA DI LAVORO 114.2010, 2 (May, 2010).
17
For an overview of this literature, see Hilary J. Allen, Payments Failure, 62 B.C. L. REV. 453, 463 et seq. (2021).
In finance, “leverage” refers to using debt to acquire financial assets. Banks loans are
perhaps the most familiar and simple form of debt: investors (including other financial institutions)
can use the money they borrow from banks to increase their exposure to the assets they want to
invest in. Another familiar form of leverage entails investors borrowing some of the purchase
price for an asset from their broker, which is known as trading on margin. Leverage can multiply
profits, but when an investor only puts down a little bit of their own money to buy an asset and
borrows the rest, their down payment can be quickly wiped out if the price of the asset falls.18
Then the investor may have to sell the asset (or something else) in order to repay their debt (or to
satisfy a lender’s demand for more loan collateral). From a financial stability perspective, too much
leverage is problematic both because of its ability to multiply exposure to assets (which can inflate
bubbles), and also because the deleveraging process once the market turns south generates
significant “fire sale externalities” as the borrower is forced to sell assets at a discount in order to
satisfy their lender.19 This drives down the market price for the assets that are being sold, which
may force other market participants to deleverage, and may even drive them into insolvency.
Economist John Geanakoplos has observed that “[a]ll leverage cycles end with (1) bad news that
creates uncertainty and disagreement, (2) sharply increasing collateral rates, and (3) losses and
bankruptcies among the leveraged optimists.”20
Because too much leverage makes the financial system more fragile, traditional bank
lending (as well as margin lending from brokers) has long been subject to regulatory requirements
that have the practical effect of ensuring that some kind of minimum down payment is always
made by the borrower, preventing unlimited leverage.21 The development of credit default swaps
(“CDS”) in the mid-1990s, however, created a new and initially unlimited way of creating
leverage.22 CDS work like a type of insurance policy that will pay out if the underlying bond
suffers some kind of credit-related problem – except that you don’t need to hold the underlying
bond to buy a CDS.23 For this reason, multiple CDSs can reference the same underlying bond,
and by doing so, multiply the number of people getting exposure to that bond. In the lead up to
2008, CDS buyers often failed to demand any “downpayment” of collateral from their
counterparties, and so an unlimited amount of leverage could be created: “many firms, like AIG,
were allowed to make naked bets, without any credible showing of collateral to back up their
promise to pay in the event the default they were “insuring” against came to pass”.24 And so the
development of CDS allowed for the creation of more leverage in the financial system, which came
to a head during the 2008 crisis. The Financial Crisis Inquiry Commission report on the causes of
18
Anat Admati & Martin Hellwig, THE BANKERS’ NEW CLOTHES, 17 (2013).
19
“Losses by leveraged buyers of assets can cause a chain reaction when a margin call forces a leveraged buyer to
sell, which lowers the price forcing another leveraged buyer to sell and so on.” Geanakoplos, supra note 13 at 20.
20
Id. at 2.
21
Id. at 19.
22
For a comprehensive recount of the development of credit default swaps, see Gillian Tett, FOOL’S GOLD (2009).
23
Geanakoplos, supra note 13 at 16.
24
Id. at 20.
In the aftermath of the 2008 crisis, Congress and regulators took steps to reduce the amount
of leverage that swaps could create in the financial system: Title VII of Dodd-Frank encouraged
clearing of swaps (with the expectation that clearinghouses would impose margin requirements as
well as net out obligations), and introduced margin requirements for uncleared swaps.26 We have
already discussed how the “down payment” required by margin requirements limits leverage;
netting is another way of reducing the amount of leverage in the system. When CDS obligations
are netted out against one another, they cancel each other out, reducing the amount of leverage
associated with an asset. Geanakoplos demonstrates this using the following example:
A firm F that was neutral, betting one way against firm A on [a bond], and betting the
opposite way on the same [bond] against firm C could come out a loser anyway. If firm A
defaults on its insurance payment, then F will be unpaid by A but still on the hook for
paying C. So instead of just one firm A going bankrupt and another firm C going unpaid
in the absence of collateral, as would happen with netting, another firm F might also go
bankrupt, closing shop, firing workers, and creating other social costs.27
Regulation requiring netting as part of the clearing process eliminates exposure for parties like F
and reduces the amount of leverage in the system overall (although it does concentrate a significant
amount of risk in clearinghouses themselves). Title VII of Dodd-Frank is an improvement over
the unregulated status quo that prevailed before it was enacted (which allowed CDS to create
almost unlimited leverage), but Title VII has many limitations and has come in for its fair share of
criticism (particularly regarding the amount of risk building up in clearinghouses).28
B. Mortgage-Backed Securities
When banks make loans and hold them on their books, they are required to meet regulatory
capital requirements with respect to those loans (in other words, to continue funding them with
specified amounts of equity).29 However, if banks make loans and then sell them, then they have
no continuing obligation to satisfy capital requirements with respect to those loans (they also avoid
any ongoing default risk associated with those loans). Securitization provides a way for banks to
sell their loans right away: shortly after the bank makes the loans, they are sold to a bankruptcy-
remote entity that pays for the loans by selling bonds or other debt instruments to investors (in
exchange, those investors receive payments of principal and interest over time that are derived
from the pool of loans).30 Where the assets are mortgage loans, payments to investors come
indirectly from borrowers’ repayments on their individual mortgages.
25
Financial Crisis Inquiry Commission, supra Note 8 at xx; xxiv.
26
Michael S. Barr et al., FINANCIAL REGULATION: LAW AND POLICY, 1084 (2016).
27
Geanakoplos, supra note 13 at 20.
28
See, for example, Adam J. Levitin, The Tenuous Case for Derivatives Clearinghouses, 101 GEO. L. J. 445 (2013).
29
For a more detailed description of regulatory capital requirements, see Hilary J. Allen, Let’s Talk About Tax:
Fixing Bank Incentives to Sabotage Stability, 18 FORDHAM J. OF CORP. & FIN. L. 821, 828 et seq. (2013).
30
Gorton & Metrick, supra Note 12 at 270.
When bad times came, in the form of a nationwide mortgage foreclosure crisis, “the
mortgage securitization pipeline lit and spread the flame of contagion and crisis.”34 The rigidities
of the securitization structure made it harder for the underlying mortgages to be modified (thereby
increasing the number of foreclosures); they also exacerbated the turmoil in the financial markets.
Because securitization contracts did not contemplate a nationwide foreclosure crisis, and because
they were so hard to renegotiate once such a crisis occurred, the value of the securities produced
became very unclear.35 This valuation uncertainty made MBS very difficult to trade (or at least,
to trade without a significant discount), and leveraged financial institutions that had significant
holdings of MBS were often forced to sell off other assets in fire sales, depressing the values of
other financial asset classes (forcing leveraged institutions exposed to those asset classes to engage
in fire sales, in a vicious cycle).36 In sum, Gelpern and Levitin observe that “although
securitization contracts generate significant externalities and impose costs on a wide range of
constituencies beyond the contracting parties, they are designed to limit the government’s capacity
to mitigate their potential adverse impact on the economy.”37 No real reform was made to the
regulation of securitizations after the 2008 crisis, so the rigidities associated with the structure
remain.
Deposit-taking banks used to face caps on the amount of interest they could pay, and as
interest rates rose in the 1970s, this proved very frustrating for depositors. Money market mutual
31
Barr et al., supra Note 26 at 1154.
32
Anna Gelpern & Adam Levitin, Rewriting Frankenstein Contracts: The Workout Prohibition in Residential
Mortgage-Backed Securities, 82 S. CAL. L. REV. 1077, 1079 (2009)
33
Id.
34
Financial Crisis Inquiry Commission, supra Note 8 at xxiii.
35
Gelpern & Levitin, supra Note 32 at 1124-7.
36
On fire sale externalities generally, see Anil S. Kashyap et al., The Macroprudential Toolkit, 59 IMF ECON.
REV. 145 (2011).
37
Gelpern & Levitin, supra Note 32 at 1127.
In September 2008, the Reserve Primary Fund (a MMF with exposure to Lehman
Brothers) broke the buck, and that event caused many investors in other money market mutual
funds to panic.40 A run ensued as panicked investors rushed to redeem their MMF shares as quickly
as possible. They feared that if they waited too long, their fund would have already sold its best
assets to satisfy other investors’ redemption requests, leaving them less likely to receive one dollar
per share – a calculation that mirrors the calculation that depositors make during bank runs (or at
least, a calculation they used to make before the introduction of deposit insurance).41 During a
run, redemption requests can force MMFs to start liquidating investments at fire sale prices in
order to satisfy redemption requests, depressing asset markets, and cutting off credit for the
corporations in which MMFs usually invest through the commercial paper market.42
Three days after the Reserve Primary Fund broke the buck, the Treasury Department
temporarily guaranteed the $1 share price for all MMMFs, and the Federal Reserve provided
emergency liquidity to MMMFs, in order to limit fire sales and prop up the commercial paper
market.43 Once these temporary measures expired, policymakers considered multiple reform
proposals that would make MMMFs less susceptible to runs. A variety of reforms were adopted
in 2010 and 201444 – but these measures were insufficient to prevent a run at the beginning of the
Covid pandemic in March 2020. As MMMF shareholders again began to increase their
redemptions, a repeat of the 2008 emergency intervention by the Federal Reserve was required.45
These multiple instances of government support have most likely created expectations among
managers of MMFs that they will receive similar support in the future – expectations of future
support may encourage managers to include riskier (and therefore more profitable) assets in their
reserves going forward.46 These kinds of perverse incentives are known as “moral hazard”.
III. DEFI
38
Gary B. Gorton & Jeffery Zhang, Wildcat Stablecoins, 21 (Jul. 19, 2021),
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3888752.
39
The special accounting treatment is authorized by 17 CFR § 270.2a-7(c).
40
Hilary J. Allen, Money Market Fund Reform Viewed Through a Systemic Risk Lens, 11 J. Bus. & Sec. L. 87, 94-5
(2010).
41
On the theory of bank runs, see Douglas W. Diamond & Philip H. Dybvig, Bank Runs, Deposit Insurance, and
Liquidity, 91 J. POLITICAL ECON. 401, 403 (1983).
42
Id. at 95-6.
43
Gorton & Zhang, supra Note 38 at 23.
44
Report of the President’s Working Group on Financial Markets, Overview of Recent Events and Potential Reform
Options for Money Market Funds, 6-8 (2020).
45
Gorton & Zhang, supra Note 38 at 24.
46
Barr et al., supra Note 26 at 1211.
A. Introduction to DeFi
DeFi, like any new and evolving business model or technology, is hard to pin down with a
precise definition.51 Right now, the term is typically used to describe a software application
(known as a “Dapp”) that serves as a simulacrum of traditional financial services provided using
coins and tokens hosted on a permissionless distributed ledger. A distributed ledger is at its core
a database hosted on multiple computers, and a distributed ledger is “permissionless” if there is no
central authority in charge of determining who has the right to record transactions on the ledger –
currently, the Ethereum blockchain (a permissionless ledger) is the ledger that is predominantly
used for DeFi Dapps52. Tokens and coins are computer files stored on the distributed ledger, and
payments in DeFi are often made using a type of coin known as a “stablecoin”53 (stablecoins try
to avoid the volatility associated with cryptocurrencies like Bitcoin by pegging their value to the
US Dollar or some other fiat currency).54 Dapps are built using smart contracts,55 which are
computer programs that run on the distributed ledger and govern the operation of tokens and coins
in a way that is intended to be self-executing and self-enforcing. However, because most users of
DeFi Dapps cannot easily access the distributed ledger directly from their phones or laptops, Dapps
usually integrate the smart contracts with more traditional forms of software to create “user-facing
interfaces.”56
Proponents of DeFi assert that these technologies will be used in concert to provide new
versions of “payments, lending, trading, investments, insurance, and asset management”
47
Financial Crisis Inquiry Commission, supra Note 8 at xxiii.
48
Id. at xx.
49
Id. at xxiv.
50
For another comparison of DeFi to credit default swaps, mortgage-backed securities, and money market mutual
funds, see Michael Hsu, Cryptocurrencies, Decentralized Finance, and Key Lessons from the 2008 Financial Crisis
(Sept. 21, 2021), https://www.occ.gov/news-issuances/speeches/2021/pub-speech-2021-101.pdf.
51
One widely used working definition describes DeFi as having four defining characteristics: 1. “Financial
Services”; 2. “Trust-minimized operation and settlement”; 3. “Non-custodial design”; and 4. “Open, programmable,
and composable architecture”. Wharton Blockchain and Digital Asset Project, DeFi Beyond the Hype, 2-3 (May
2021), https://wifpr.wharton.upenn.edu/wp-content/uploads/2021/05/DeFi-Beyond-the-Hype.pdf.
52
President’s Working Group on Financial Markets, REPORT ON STABLECOINS, 9 (November 2021).
53
Id. at 8
54
The President’s Working Group describes stablecoins as “digital assets that are designed to maintain a stable
value relative to a national currency or other reference assets.” Id. at 1.
55
Wharton Blockchain and Digital Asset Project, supra Note 51 at 3.
56
Id.
In addition to these fragilities, DeFi Dapps are highly complex. Most investors (including
established financial institutions) are used to reviewing balance sheets and written disclosures to
assess investments. Few are able to read the computer code of the smart contracts that make up
the Dapps – and even those who can will struggle to find flaws simply by looking at the code in
the abstract.60 While it’s possible for the operators of Dapps to provide written disclosures to their
users, written disclosure documents may prove to be highly inconsistent with how the code of the
relevant smart contracts actually functions61 – and there’s no way for investors (or regulators) to
verify this unless they can run a beta test, or at the very least read the code. Finally, added
complexity arises as a result of the convoluted governance structure that often controls the Dapps’
software, as well as the governance structure of the permissioned ledgers on which the Dapps run.
This means that if a problem were to occur and emergency intervention needed to be provided
within the DeFi ecosystem to head-off catastrophic spillover effects for the rest of the financial
system, it could be difficult to figure out who to provide emergency support to. DeFi therefore
has many complexity-related fragilities; the next Part will look more particularly at how DeFi
resembles Shadow Banking 1.0.
Credit default swaps can create leverage in the financial system by multiplying the number
of times someone can get exposure to the same underlying asset (typically, a bond). The amount
of leverage in the system can also be increased by simply multiplying the number of assets
available to borrow against. That is a significant concern with DeFi, where financial assets in the
form of tokens can be created out of thin air by anyone with computer programming knowledge,
57
Id. at 2.
58
“The problem is that all this fancy financial engineering has, as yet, no “real” economy to service. Instead it
underpins an incorporeal casino: most of those using DeFi do so to facilitate or leverage their bets on one of many
speculative tokens.” Alice Fulwood, Decentralized Finance is Booming, But it Has Yet to Find Its Purpose, THE
ECONOMIST (Nov. 8, 2021).
59
DeFi Scams 101: How to avoid the most common cryptocurrency frauds, COINTELEGRAPH,
https://cointelegraph.com/defi-101/defi-scams-101-how-to-avoid-the-most-common-cryptocurrency-frauds.
60
On the limitations of reviewing source code, see Joshua A. Kroll et al., Accountable Algorithms, 165 U. Pa. L.
REV. 633, 638, 647 (2017).
61
Shaanan Cohney et al., Coin-Operated Capitalism, 119 COLUM. L. REV. 591 (2019).
Financial regulators should be very wary of the possibility of unlimited leverage building
up in the DeFi ecosystem, particularly if there are channels of contagion that would allow
deleveraging/fire sales in the DeFi ecosystem to impact the mainstream financial system and
broader economy. Recent IMF research has found increasing correlation between the performance
of crypto investments and more traditional investments like equities (especially during market
volatility), and cautioned that “[i]ncreased crypto-stocks correlation raises the possibility of
62
Saule T. Omarova, New Tech v. New Deal: Fintech as a Systemic Phenomenon, 36 YALE J. REG. 735, 775
(2019).
63
For a discussion of spillovers from operational failures in payments systems, see Allen, supra Note 17.
64
Sirio Aramonte et al., DeFi Risks and the Decentralization Illusion, BIS QUARTERLY REVIEW, 29-30 (Dec.
2021).
65
Id. at 29.
66
For a discussion of why market participants don’t address systemic risks on their own, see Notes 134-136 and
accompanying text.
67
“Through the use of advanced smart contracts, the platform allows anyone to issue and trade synthetic assets that
monitor and track the price of arbitrary real-world assets. Impressively, this is all accomplished without requiring
physical backing… the goal of the project is to speed up the integration of traditional assets entering the blockchain
sector.” David Hamilton, Investing in the Mirror Protocol (MIR) – Everything You Need to Know, SECURITIES IO
(Jan. 29, 2022), https://www.securities.io/investing-in-the-mirror-protocol-mir-everything-you-need-to-know/.
68
Aramonte et al., supra Note 64 at 29.
69
Id. at 30.
What might spur a great deleveraging in the DeFi ecosystem? There are many possibilities.
There could be a problem with an intermediary on which DeFi relies (these intermediaries are
discussed in detail in Section IV). DeFi is also rife with scams, and “forks, hacks, rug pulls,
vampire attacks, and flash loans all have the potential to surprise, erode trust, and spark fear.”73
While these types of events have not yet destroyed confidence in DeFi, most current users are
likely to be “hardcore believers in the technology and thus are both understanding of the risks and
willing to forgive them.”74 That is unlikely to remain the case if DeFi is more widely adopted, in
which case these kinds of events could destroy confidence in the value of DeFi assets more
generally. Or deleveraging might start simply as a correction to a crypto bubble if the “irrational
exuberance” starts to wear off.75
ii. Rigidity
When critical parts of the financial system become overleveraged, flexibility may be
needed during the bust cycle to release the largest entities from obligations to respond to margin
calls or repay loans – otherwise the failures of intermediaries and fire sales will have ripple effects
that can drag down the whole system.76 Unfortunately, smart contracts may prove too rigid to
provide the flexibility needed to avoid such an outcome. Smart contracts are designed to execute
their preprogrammed instructions instantly, without waiting for input from the parties (or a
regulator, or a court). In good times, this makes things more efficient – but smart contracts will
execute just as quickly in bad situations, even if the people involved would be better off if they
didn’t. It has already become evident that flaws in smart contracts can be exploited by hackers to
steal tokens and coins,77 but less attention has been paid to the fact that in the future, there may be
situations where the stability of the financial system would benefit if smart contracts simply didn’t
execute.
70
Tobias Adrian et al., Crypto Prices Move More in Sync With Stocks, Posing New Risks, IMF BLOG (Jan. 11,
2022), https://blogs.imf.org/2022/01/11/crypto-prices-move-more-in-sync-with-stocks-posing-new-risks/.
71
Jamie Crawley, US Banks Form Group to Offer USDF Stablecoin, COINDESK (Jan. 12, 2022).
72
Ian Allison, Fireblock ‘Whitelists’ 30 Trading Firms for Aave’s Institutional DeFi Debut, COINDESK (Jan. 5,
2022).
73
Hsu, supra Note 50.
74
Id.
75
On the psychology of bubbles, see Robert J. Shiller, IRRATIONAL EXUBERANCE (2000).
76
On the need for legal elasticity to ensure the survival of the financial system, see Katharina Pistor, A Legal Theory
of Finance, 41 J. Comparative Econ. 315, 320 (2013).
77
For the discussion of the hack of the Poly Network, see Ephrat Livni, For Rules in Technology, the Challenge is
to Balance Code and Law, N.Y. TIMES (Nov. 23, 2011).
While there are steps that can be taken to better equip a smart contract to adapt to
unexpected events (for example, a smart contract can be programmed to consult another smart
contract, or a external data source known as an “oracle” that is controlled by a trusted party), taking
these kinds of steps will increase transaction costs. The Ethereum ledger charges a “gas cost” for
any computing done, and consulting an oracle would increase the amount of computing power
(and thus the gas charge) necessary to execute a smart contract.81 Participants in the DeFi markets
will probably be willing to bear these charges up to a certain point, but eventually, these ongoing
operational costs will discourage measures that cater for very unusual events. Unfortunately, when
we’re talking about financial stability, low-probability high-consequence tail events are the ones
we’re most concerned about. These are the types of events that turned mortgage-backed
securitizations into “suicide pacts” – smart contracts may prove to be even more dangerous in the
midst of such events, if the speed of their self-execution leaves no time for emergency intervention.
iii. Runs
78
Aramonte et al., supra Note 64 at 27.
79
andreessen horowitz, Letter to Senate Banking Committee re: Request for Proposals for Clarifying Laws
Concerning Cryptocurrency and Blockchain Technologies (Sept. 27, 2021), https://a16z.com/wp-
content/uploads/2021/10/Andreessen-Horowitz-Senate-Banking-Proposals.pdf.
80
Primavera De Filippi & Aaron Wright, BLOCKCHAIN AND THE LAW: THE RULE OF CODE, 188 (2018).
81
Hilary J. Allen, DRIVERLESS FINANCE: FINTECH’S IMPACT ON FINANCIAL STABILITY, 99 (2022).
82
See, for example, Gorton & Zhang, supra Note 38; Dan Awrey, Bad Money, 106 CORNELL L. REV. 1 (2020).
Tether, for example, does not allow U.S. resident holders of its stablecoins to redeem them
directly from Tether, so holders are forced to go to a crypto exchange (like Coinbase) if they want
to convert their Tether to fiat currency.83 It’s not clear whether these exchanges are contractually
obligated to exchange Tether for $1, or whether they could refuse to do so in some circumstances
(for example, only allowing Tether to be exchanged for another cryptocurrency). It’s also not clear
whether a crypto exchange could turn around after exchanging a Tether for a customer and
contractually demand that Tether give the exchange $1 for the stablecoin. Assuming, though, that
stablecoin holders could demand an exchange for fiat and that the exchanges could demand a
redemption for fiat from the stablecoin issuer (potentially forcing a liquidation of the reserve), then
centralized stablecoins would have many similarities to MMFs. Decentralized stablecoins like
DAI operate in a more complicated way, and their potential run dynamics are even harder to follow
as a result. Instead of having a centralized entity like Tether managing a reserve of real-world
assets, DAI is maintained by a DAO (MakerDAO), and relies on smart contracts buying and selling
a reserve of cryptoassets (including a significant amount of the centralized stablecoin USDC) to
stabilize DAI’s price.84 As with centralized stablecoins, holders of DAI can seek to convert them
into fiat using an exchange, but unlike centralized stablecoins, the ultimate decision to liquidate
the reserve of cryptoassets (and provide them to DAI holders) is not made by a centralized entity,
but instead must go through the MakerDAO governance process.85
Runs happen when people lose confidence that a particular asset (like a share in a MMF)
will continue to remain accessible at the expected value.86 This “confidence” aspect of runs means
that runs are unlikely to occur if the people holding an asset never expected it to have a stable
value in the first place. Right now, it appears that the vast majority of stablecoins are not being
used for payments for real-world goods and services. Instead, the recent exponential growth in
stablecoin usage has been driven by people who have purchased stablecoins to speculate in the
crypto markets87 – SEC Chair Gary Gensler has described them as “poker chips” that are the price
of admission to the “casino”.88 If something were to shake confidence in stablecoins’ acceptance
in the DeFi ecosystem (this ‘something’ could range from a hack, to a problem with the reserve of
assets backing a stablecoin, to a problem with the smart contracts managing the value of a
decentralized stablecoin), we could then expect holders to exchange their stablecoins for fiat
currency and exchanges to seek redemption, forcing stablecoin issuers to start liquidating the
83
Alexis Goldstein, Testimony before the Senate Committee on Banking, Housing, and Urban Affairs, Hearing on
Stablecoins: How Do They Work, How Are They Used, and What Are Their Risks?, 3 (Dec. 14, 2021).
84
Lyle Daly, 5 Things to Know Before You Buy DAI, MOTLEY FOOL (Sept. 1, 2021), https://www.fool.com/the-
ascent/cryptocurrency/articles/5-things-to-know-before-you-buy-dai/.
85
“A Global Settlement is a last resort process to guarantee the Target Price to the holders of DAI. When a Global
Settlement is triggered, it shuts down the system. This means that holders of DAI…receive the net value of assets
that they are entitled to. The process is fully decentralized and Maker voters govern the access to it in the case of an
emergency.” Sharon Manrique, What is DAI, and how does it work?, MEDIUM (Feb. 7, 2019),
https://medium.com/mycrypto/what-is-dai-and-how-does-it-work-742d09ba25d6.
86
Diamond & Dybvig, supra Note 41 at 403.
87
At the time of publication of this report, stablecoins are predominantly used in the United States
to facilitate trading, lending, and borrowing of other digital assets.” PWG Report, supra Note 52 at 8.
88
Cheyenne Ligon, SEC’s Gensler Calls Stablecoins ‘Poker Chips’ at the Wild West Crypto Casino, COINDESK
(Sept. 21, 2021).
The recent President’s Working Group Report on stablecoins observed that “[b]ased on
information available, stablecoins differ in the riskiness of their reserve assets, with some
stablecoin arrangements reportedly holding virtually all reserve assets in deposits at insured
depository institutions or in U.S. Treasury bills, and others reportedly holding riskier reserve
assets, including commercial paper, corporate and municipal bonds, and other digital assets.”89 It
is possible that mass withdrawals by stablecoins from insured deposit accounts could trigger runs
on the institutions (like banks) that provide those deposit accounts. I will argue in Section IV.B
that banks should therefore be prohibited from holding stablecoin reserves on deposit.
With regard to other types of assets in stablecoin reserves, fire sales of those assets could
be a significant concern, but their systemic impact will depend on the size of the stablecoin reserve.
A decentralized stablecoin like DAI, for example, holds its reserves in cryptoassets, and so the
impact of a sell-off on the prices of real-world assets might be limited (although DAI invests
heavily in USDC,90 so a run on DAI might trigger a run on USDC, which does invest its reserve
in real-world assets). Interestingly, the reserves of Tether, which currently has by far the largest
market value of any stablecoin, may not actually be as big as expected. As one recent report put it:
“[e]xactly how Tether is backed, or if it’s truly backed at all, has always been a mystery. For years
a persistent group of critics has argued that, despite the company’s assurances, Tether Holdings
doesn’t have enough assets to maintain the 1-to-1 exchange rate, meaning its coin is essentially a
fraud.”26 If true, this would be highly problematic for holders of Tether, but it would also limit the
systemic impact of any fire sale of Tether’s reserve assets – because there wouldn’t be so many of
them. However, if these centralized stablecoin issuers start to become an important source of
capital for the real economy (as money market mutual funds did before them) then runs on
stablecoins will be a potential source of systemic risk.
The previous Section gave an overview of DeFi’s inherent fragilities, and what they might
mean for the stability of our broader financial system in the future. However, in deciding how
regulation should respond to DeFi, it is important to take a step back from what DeFi aspires to
be, and consider what DeFi actually is right now. A recent report from the Bank for International
Settlements described DeFi as largely self-referential, and concluded that “[g]iven this self-
contained nature, the potential for DeFi-driven disruptions in the broader financial system and the
real economy seems limited for now.”91 DeFi is not yet an entrenched part of our financial system,
and regulators still have the opportunity to take a precautionary approach that will have a real
impact on how DeFi develops. Regulators may be able to ensure that DeFi never reaches a scale
at which it could threaten the stability of our broader financial system – and if steps are taken from
the outset to limit the growth of DeFi and its integration with the traditional financial system
regulation, regulators won’t need to respond directly to the destabilizing problems discussed in
89
PWG Report, supra Note 52 at 4.
90
https://daistats.com/#/
91
Aramonte et al., supra Note 64 at 21.
The growth of Shadow Banking 1.0 was not inevitable; a series of policy choices allowed
it to develop. This is illustrated most obviously by Congress’ passage of the Commodity Futures
Modernization Act in 2000, which prevented the SEC and CFTC from regulating swaps – and
which Congressman Bliley justified as necessary in part because “[d]erivative
instruments…reflect the unique strength and innovation of American capital markets” and because
“U.S. markets and market professionals have been global leaders in derivatives technology and
development.”93 We hear the same rhetoric with regard to stablecoins and other DeFi projects,94
and this kind of rhetoric could encourage regulators to accommodate the growth of DeFi. As I
have argued previously, though, this is “not a neutral approach. Instead, it stacks the deck in favor
of the innovators who get to profit by generating risks that, if they come to fruition, will be borne
primarily by the rest of society.”95 Regulators should instead pursue a precautionary approach to
DeFi, erring on the side of caution to protect society from the risks it would otherwise generate
(which include not only the financial stability risks discussed in Section III.B, but also serious
consumer protection concerns, environmental costs, and national security risks).96
Given these risks, regulators would be more than justified in taking steps to limit the growth
of DeFi and preventing its integration with the established financial system – unless there were
something truly transformative about DeFi innovation. As this Section will explore, many of the
touted benefits of DeFi are illusory, though, and so lawmakers and regulators should have few
qualms about using regulation to limit DeFi innovation. The latter half of this Section will provide
some brief discussion of the forms this kind of DeFi regulation could take.
92
Hilary J. Allen, Testimony before the Senate Committee on Banking, Housing, and Urban Affairs, Hearing on
Stablecoins: How Do They Work, How Are They Used, and What Are Their Risks?, 6 (Dec. 14, 2021).
93
Congressman Bliley, Report on the Commodity Futures Modernization Act of 2000, 45-46 (Sept. 6, 2000).
94
Senator Toomey, for example, recently said that stablecoin legislation should be “designed to promote innovation
in the rapidly evolving global digital economy” and “seek to maintain the international competitiveness of the
United States.” Toomey Outlines Stablecoin Principles to Guide Future Legislation (Dec. 14, 2021),
https://www.banking.senate.gov/newsroom/minority/toomey-outlines-stablecoin-principles-to-guide-future-
legislation.
95
Allen, supra Note 81 at 41.
96
“Bitcoin is notorious for consuming as much electricity as the Netherlands, but there are around 10,000 other
cryptocurrencies, most using similar infrastructure and thus also in aggregate consuming unsustainable amounts of
electricity. Bitcoin alone generates as much e-waste as the Netherlands, cryptocurrencies suffer an epidemic of
pump-and-dump schemes and wash trading, they enable a $5.2B/year ransomware industry, they have disrupted
supply chains for GPUs, hard disks, SSDs and other chips, they have made it impossible for web services to offer
free tiers, and they are responsible for a massive crime wave including fraud, theft, tax evasion, funding of rogue
states such as North Korea, drug smuggling, and even as documented by Jameson Lopp's list of physical attacks,
armed robbery, kidnapping, torture and murder.” David Rosenthal, EE380 Talk, (Feb. 9, 2022),
https://blog.dshr.org/2022/02/ee380-talk.html.
i. Decentralization
Recent research from the Bank for International Settlements has observed that there is “a
“decentralisation illusion” in DeFi due to the inescapable need for centralised governance and the
tendency of blockchain consensus mechanisms to concentrate power.”100 No less than internet
pioneer Tim O’Reilly has noted that “history teaches us that there will always be new avenues for
power to become centralized”, and that “Blockchain turned out to be the most rapid
recentralization of a decentralized technology that I've seen in my lifetime.”101 The “inescapable
need for centralized governance” derives in part from issues we’ve already discussed in the context
of smart contracts: because it’s not possible to address all possible eventualities in advance, an
intermediary is often needed to resolve unanticipated situations (for example, reversing erroneous
or problematic transactions).102 Where there are opportunities to profit from streamlining
unwieldy decentralized services for users (especially when venture capitalists are standing ready
to fund such projects), the evolution of centralized intermediaries seems inevitable.103 Ultimately,
the need for intermediaries is an economic issue, not a technological one, and therefore not
something that technology can fix: as tech veteran David Rosenthal puts it “economics forces
successful permissionless blockchains to centralize.”104
97
“web3 is a somewhat ambiguous term, which makes it difficult to rigorously evaluate what the ambitions for
web3 should be, but the general thesis seems to be that web1 was decentralized, web2 centralized everything into
platforms, and that web3 will decentralize everything again. web3 should give us the richness of web2, but
decentralized.” Moxie Marlinspike, My first impressions of Web3 (Jan. 7, 2022),
https://moxie.org/2022/01/07/web3-first-impressions.html
98
Molly White, It’s not still the early days (Jan. 14, 2022), https://blog.mollywhite.net/its-not-still-the-early-days/
99
The likely endgame is “a costly, inefficient database, which is not in fact decentralized.” Schuster, supra Note 10
at 992.
100
Aramonte et al., supra Note 64 at 22.
101
Dan Patterson, Internet guru Tim O’Reilly on Web3: “Get ready for the crash”, CBSNEWS (Feb. 10, 2022).
102
Aramonte et al., supra Note 64 at 27.
103
Rosenthal, supra Note 96.
104
Id.
105
This tendency is known as “automation bias”. See Linda J. Skitka et al., Accountability and Automation Bias, 52
Int. J. Human- Computer Studies 701 (2000).
The Dapps operate on top of another layer of infrastructure: a distributed ledger, like the
Ethereum blockchain, which is also dependent on many humans for its functioning (Grimmelman
and Windawi point out an important and often unappreciated layer of infrastructure needed to
support the distributed ledgers – the internet itself.111 The actions of ISPs could therefore impact
the operation of distributed ledgers, although we tend to take the neutral functioning of ISPs for
granted).112 Most decisions relating to the operation of a distributed ledger are made by the people
with the power to validate transactions on that ledger, and by the core developers of the computer
code governing that ledger. While the underlying code of ledgers like the Ethereum and Bitcoin
blockchains is open-source, that doesn’t mean that there is no hierarchy in terms of the computer
programmers able to modify that code. Instead, so-called “core developers” “function as the
leaders and decision makers in relation to the code.”113 Validators are also important actors,
because they determine the definitive version of the ledger (which is the definitive record of who
owns the cryptoassets associated with that ledger).
106
See notes 79 and accompanying text.
107
Tom Barberau et al., Decentralized Finance’s Unregulated Governance: Minority Rule in the Digital Wild West
(Feb. 8, 2022), https://ssrn.com/abstract=4001891.
108
Chainanalysis, DeFi Whales Turned Central, Northern & Western Europe into the World’s Biggest
Cryptocurrency Economy (Oct. 14, 2021) (available at https://blog.chainalysis.com/reports/central-northern-
western-europe-cryptocurrency-geography-report-2021-preview/).
109
Amy Deen Westbrook & David A. Westbrook, SnapChat’s Gift: Equity Culture in High Tech Firms, 46 FL. ST.
U. L. REV. 861, 871-2 (2019).
110
Dante Disparte, 10 Stablecoin Myths: Internet Wildcat Banks or Always-On Dollars, 6 (Dec. 2021),
https://www.circle.com/en/10-stablecoin-myths-busted.
111
Grimmelman & Windawi [in symposium volume]
112
De Filippi & Wright, supra Note 80 at 177-8.
113
Angela Walch, In Code(rs) We Trust: Software Developers as Fiduciaries in Public Blockchains, in
REGULATING BLOCKCHAIN: TECHNO-SOCIAL AND LEGAL CHALLENGES (Hacker et al, Eds), 61
(2019).
Ultimately, the computer code is not running the show on its own – DeFi is governed by
institutions and individuals, and we have already seen these institutions and individuals exercising
their power. In 2016, when $60 million of Ether were stolen from an early DAO (known as “The
DAO”), core developers and some miners banded together to “hard fork” the Ethereum distributed
ledger, which “effectively rolled back the Ethereum network’s history to before The DAO attack
and reallocated The DAO’s ether to a different smart contract so that investors could withdraw
their funds.”120 More recently, when “a software upgrade in Compound in September [2021]
114
De Filippi & Wright, supra Note 80 at 23-4.
115
Angela Walch, Testimony before the Senate Committee on Banking, Housing, and Urban Affairs, Hearing on
Cryptocurrencies: What Are They Good For?, 9 (Jul. 27, 2021).
116
“It only took six years for Bitcoin to fail Nakamto's goal of decentralization, with one mining pool controlling
more than half the mining power. In the seven years since no more than five pools have always controlled a majority
of the mining power.” Rosenthal, supra Note 96.
117
Id.
118
“[N]or [would] moving to proof-of-stake…solve the MEV problem on Ethereum.” Michelle Lim, Could
Ethereum’s upgrade affect miners’ manipulations for extra profits? FORKAST (Aug. 5, 2021),
https://forkast.news/how-will-ethereums-upgrade-affect-miners-mev/
119
Aramonte et al., supra Note 64 at 28.
120
Cryptopedia Staff, What Was The DAO? (Apr. 27, 2021), https://www.gemini.com/cryptopedia/the-dao-hack-
makerdao#section-what-is-a-dao.
Many centralized intermediaries are also critical to the DeFi ecosystem. This is not really
surprising: as encryption pioneer Moxie Marlinspike has observed, decentralized services do not
scale well,122 and many DeFi intermediaries exist to compensate for the difficulties associated with
decentralized technology. Exchanges, for example, are critical to DeFi, because they enable users
to exchange digital assets like cryptocurrencies (including stablecoins) for one another.123 While
exchanges with more decentralized governance structures (like Uniswap) are available, they
generally charge more per transaction (especially for smaller transactions), and process far fewer
transactions than exchanges operated by centralized intermediaries like Coinbase.124 Marlinspike
also observed that DeFi Dapps rely on APIs that allow users’ devices to access the distributed
ledger on which transactions take place, because “blockchains are designed to be a network of
peers, but not designed such that it’s really possible for your mobile device or your browser to be
one of those peers.”125 Marlinspike found that almost all DeFi users ultimately rely on client APIs
provided by either Infura or Alchemy for this purpose – Infura and Alchemy are therefore critical
intermediaries for the DeFi ecosystem, as are the wallet providers who provide users with access
to their digital assets126 (again, these are needed because most users can’t access assets on a
distributed ledger directly). All of these centralized intermediaries have the power to prevent users
from engaging in crypto transactions, and yet users trust them not to. Commenting specifically on
the intermediaries Infura and Alchemy, Marlinspike observed that “[s]o much work, energy, and
time has gone into creating a trustless distributed consensus mechanism, but virtually all clients
that wish to access it do so by simply trusting the outputs from these two companies without any
further verification.”127
Ultimately, trust is required in the DeFi ecosystem. A decentralized foundation just makes
financial services more convoluted and replaces trust in established institutions (particularly
government institutions and regulated banks) with trust in different – and sometimes unidentified
– actors.128 The DeFi ecosystem also depends heavily on stablecoins issued by centralized firms
like Tether and Circle129 (and these stablecoins in turn depend on traditional financial services like
banks and fiat currencies, in order to stabilize their value).130 Although there are more
121
Livni, supra Note 77.
122
“If something is truly decentralized, it becomes very difficult to change, and often remains stuck in time.”
Marlinspike, supra Note 97.
123
Wharton Blockchain and Digital Asset Project, supra Note 51 at 8.
124
Aramonte et al., supra Note 64 at 26. For a discussion of crypto exchanges’ conflicts of interest, see
https://www.ft.com/content/4e15d5b6-033b-4294-8aba-d95e02f51b3b.
125
Marlinspike, supra Note 97.
126
Wallets are “software interfaces for users to manage assets stored on a blockchain.” Wharton Blockchain and
Digital Asset Project, supra Note 51 at 2.
127
Marlinspike, supra Note 97.
128
It was recently revealed that the head of Treasury for the Wonderland DeFi project had previously been
incarcerated for financial fraud – and that revelation led to a significant reduction in the value of the related TIME
token. As one commentator put it, “If the anonymous nature of DeFi means that a person like Michael Patryn can be
in charge of a major DeFi treasury, that’s a pretty big problem.” Emily Nicolle, Crypto Secrecy Makes DeFi a
Wonderland to Felon Tied to Quadriga, BLOOMBERG QUINT (Jan. 27, 2022).
129
PWG Report, supra Note 52 at 9.
130
Aramonte, supra Note 64 at 25.
The conflicts of interest that individuals and institutions in positions of authority face can
lead to suboptimal outcomes for crypto investors. We can also expect that these individuals and
institutions will fail to take financial stability into account. There is little incentive for them to
protect financial stability, because it is a public good that people can’t be excluded from or forced
to pay for, and even if the members of the crypto community were unusually altruistic, they would
not have enough information about other parts of the financial system to gauge the impact of their
actions (or be able to force their competitors to join them even if they did know how to minimize
systemic risk).134 Although SEC Commissioner Hester Peirce has complimented the crypto
community on its ability to “collectively figure out how to deal with unanticipated problems,”135
this view of self-correcting markets neglects the fact that crypto intermediaries lack both the
incentives and the information needed to address the negative externalities that crypto can create
for the broader economy.136
While it seems implausible to suggest that DeFi will ever deliver financial services entirely
“without centralized intermediaries or institutions”,137 using the term “decentralized” to describe
these services does serve marketing and political functions. The word “decentralized” taps into
the current fervor for “Web3”, maximizing chances that startups will receive funding from venture
capital firms,138 and also appeals to potential customers interested in “decentralized” products.
The commercial appeal of decentralization is driven in part by the political significance of the
term: “[t]he promise is a financial system that is democratized, decentralized, and secure. No
banks. No bailouts. No more being ignored or betrayed.”139 The appeal of this kind of rhetoric lies
in the belief that the internet works as a countervailing force against government entities and
regulated financial institutions (and neglects the reality that internet services can be another source
of concentrated power with their own conflicts of interest).140 In a book titled The Politics of
Bitcoin, David Golumbia argues that much of crypto’s pro-decentralization rhetoric actually
derives from extreme right-wing talking points about the evils of government: the existence of
DeFi intermediaries can be more easily reconciled with decentralization rhetoric if DeFi
131
Ian Allison, Fireblocks ‘Whitelists’ 30 Trading Firms for Aave’s Institutional DeFi Debut, COINDESK (Jan. 5.
2022).
132
Oracles are “Data feeds that allow information from sources off the blockchain, such as the current price of a
stock or a fiat currency, to be integrated into DeFi services.” Wharton Blockchain and Digital Asset Project, supra
Note 51 at 3.
133
Emily Perryman, What is Etherscan?, YAHOO (Oct. 28, 2019).
134
Allen, supra Note 81 at 20.
135
Hester M. Peirce, Lawless in Austin, (Oct. 8, 2021), https://www.sec.gov/news/speech/peirce-2021-10-08.
136
As one commentator put it, “Libertarianism's attraction is based on ignoring externalities, and cryptocurrencies
are no exception.” Rosenthal, supra Note 96.
137
Wharton Blockchain and Digital Asset Project, supra Note 51 at 2.
138
McMillan Cotton, supra Note 148.
139
Hsu, supra Note 50.
140
David Golumbia, THE POLITICS OF BITCOIN: SOFTWARE AS RIGHT-WING EXTREMISM, 7 (2016).
Returning to our core question of whether the likely benefits of DeFi are sufficient to justify
the financial stability risks associated with Shadow Banking 2.0, we also need to consider whether
DeFi may have other benefits that are more prosaic than decentralization. In particular, there is
interest in making transactions quicker and cheaper, and this “increased efficiency” is sometimes
pitched as a way to promote financial inclusion.142 However, it does not seem possible that a
technology that has been intentionally made more complex (in order to nominally decentralize)
could ever be more efficient than a simpler, centralized alternative.143 No matter which validation
mechanism is chosen for a decentralized ledger (proof-of-work, proof-of-stake, or something else),
it will always be slower and more cumbersome than validation by a centralized intermediary –
costly computations are the sinequanone of decentralized consensus mechanisms.144 And yet,
DeFi innovation is proliferating. This Part will argue that technological superiority is not the
primary driver of this innovation.
It became apparent following the 2008 crisis that some Shadow Banking 1.0 innovation
was not a “rational demand-side response to market imperfections”.145 Instead, the innovation was
often driven by supply-side incentives: financial institutions could profit from offering financial
products that capitalized on interest in the “next big thing”, notwithstanding that the result was
sometimes socially-useless over-innovation that hid risks from purchasers and created risks for the
broader economy.146 There are similarly perverse incentives for innovation in the tech industry,
which can encourage firms to pursue innovation that is “buzzy” enough to attract venture capital
funding, even if it is not particularly good technology.147
141
Id. at 10
142
Id. at 7-8. See also, Disparte, supra Note 110.
143
Schuster, supra Note 10 at 992.
144
Id. at 981.
145
Awrey, supra Note 15 at 260.
146
Id. at 263-4.
147
For a discussion of the limitations of the venture capital funding model that result in suboptimal innovation, see
Peter Lee, Enhancing the Innovative Capacity of Venture Capital (forthcoming YALE J. L. & TECH.). In
particular, “While VCs enjoy an iconoclastic reputation, in many contexts they tend to invest in the same popular
technologies while eschewing truly revolutionary innovations. Historical evidence reveals several trends of “hot”
technologies receiving significant VC funding and then losing favor.”
148
“[S]ome V.C.s have invested a lot in making blockchain inevitable. And the amount that V.C.s have available to
throw at investments has ballooned over the past 10 years in a way the average person can’t appreciate.” Tressie
McMillan Cotton, Wealth Inequality Drives the Appeal of Crypto, N.Y. TIMES (Jan. 31, 2022).
The real world has fundamental constraints that make the technology unworkable,
whenever it has to interact with the outside world the benefits of decentralization disappear
and the solutions end up simply recreating slower and worse versions of processes and
structures that already exist…There are fundamental limitations to the scalability of
blockchain-based technologies, and every use case is better served by another simpler
technology except for crime, ransomware, extralegal gambling, and sanctions evasion; all
of which are a drain on society not a benefit. Taken as a whole the technology has no
tangible benefits over simply using trusted parties and centralized databases.150
Another software engineer and blogger, Molly White, similarly describes blockchain technology
as “inefficient in every sense of the word” and also challenges assertions that this technology is in
its infancy and just needs more time to develop useful applications, asking whether “we are to
believe that as technology soared forward over the past decade, blockchain technologies spent that
time tripping over their own feet?”151
It goes without saying that there are technologists who take the opposite view, but the
burden should be on them to demonstrate why this aspirational technology is – in reality – superior
to the simpler, centralized alternatives we could develop with the venture capital funding that is
currently being poured into DeFi. In particular, the idea that DeFi can be used to improve financial
inclusion is a dubious claim that the industry should be required to support with concrete examples,
because, as a recent World Economic Forum report on stablecoins found:
stablecoins are subject to many of the same barriers that constrain citizens from accessing
other financial products and services, such as bank accounts, mobile money accounts or
fully digital remittance providers. Where stablecoins are accessible, they generally address
financial inclusion barriers to a similar degree as other digital financial
services...stablecoins as currently deployed would not provide compelling new benefits for
financial inclusion beyond those offered by pre-existing options.
Barriers (like the need for internet access) apply not just to stablecoins, but also limit the utility of
other DeFi Dapps for underserved communities. More generally, we should think about what DeFi
asks of underserved communities: reading financial disclosures is already hard enough, should
people really be expected to understand the ins and outs of code as well before they can understand
their financial services? As White puts it, “[h]ow long must the laymen, who are so eagerly hustled
into blockchain-based projects that promise to make them millionaires, be scolded as though it is
their fault when they are scammed as if they should be capable of auditing smart contracts
149
See, for example, Rosenthal, supra Note 96.
150
Stephen Diehl, The Case Against Crypto, https://www.stephendiehl.com/blog/against-crypto.html.
151
White, supra Note 98.
B. Regulatory Proposals
To summarize the previous Part, the inefficiencies and complexities of DeFi technology
simply do not make sense outside of the decentralization narrative, and the decentralization
narrative does not hold up to scrutiny. Given the financial stability risks that DeFi would create if
it were allowed to grow into Shadow Banking 2.0, and given that proponents of the technology
involved struggle to demonstrate any concrete superiority over simpler centralized alternatives,
policymakers should pursue policies that prevent DeFi from growing. This Part will sketch in
broad strokes some possible ways to achieve this. If DeFi remains largely disconnected from both
real-world economic applications and the established financial system, then the risks articulated in
Section III.B will not be a significant concern.
Because negative spillover effects from DeFi will wreak the most havoc on the real
economy if regulated banks become integrated into the DeFi ecosystem, steps should be taken to
insulate regulated banks from DeFi.154 As a priority, regulated banks should be prohibited from:
issuing stablecoins or providing any Dapps; holding stablecoin reserves in a deposit account; or
investing in any Dapp or stablecoin (banking regulators already have the authority they need to
take these steps).155 Some of these recommendations run counter to the President’s Working
Group Report, which recommended that “legislation should require stablecoin issuers to be insured
depository institutions.”156 This recommendation seeks to address stablecoin-related run risk, but
if followed, would create moral hazard by extending the public safety net of deposit insurance to
the DeFi ecosystem in which stablecoins are deployed. I have argued against pursuing the PWG’s
recommendation at this point in time, because taking this step now would legitimize stablecoins
in a way that would likely fuel, rather than limit, the growth of DeFi.157
We should also explore other regulatory strategies designed to prevent DeFi from growing
into Shadow Banking 2.0. Gorton and Zhang have noted that when it comes to stablecoins at least,
Congress has the authority to “tax competitors of [the US dollar] out of existence.”158 An
alternative or complementary strategy would be for Congress to adopt a licensing regime for Dapps
and stablecoins where the applicant would need to demonstrate: (i) that the Dapp/stablecoin has a
purpose that is connected to real-world economic growth; (ii) that the applicant has the institutional
capacity to manage both the financial and technological risks associated with the Dapp/stablecoin;
and (iii) that the Dapp/stablecoin is unlikely to have a negative impact on the stability of the
152
Id.
153
“I remember the days when subprime mortgage lending was similarly celebrated — when it was hailed as a way
to open up the benefits of homeownership to previously excluded groups.” Paul Krugman, How Crypto Became the
New Subprime, N.Y. TIMES (Jan. 27, 2022).
154
For an exploration of the financial stability and broader economic threats likely to arise from the integration of
traditional finance and crypto, see Allen supra Note 81,
155
For further elaboration on these proposals, see Allen, supra Note 92 at 18.
156
PWG Report, supra Note 52 at 2.
157
Allen, supra Note 92 at 2. It is also critical that antitrust measures be used to prevent any large tech firm from
leveraging its network of users into a stablecoin platform, which would also boost DeFi growth. Id. at 18.
158
Gorton & Zhang, supra Note 38 at 40.
Most of the stablecoins and Dapps currently available would struggle to satisfy these
licensing requirements, and so such a licensing regime would limit the growth of DeFi (if a license
were ultimately awarded, the licensing process would still offer regulators the opportunity to make
interventions to protect consumers and the financial system). However, with a more decentralized
Dapp or stablecoin (for example, DAI), there may be some confusion about who should apply for
the license. If such a Dapp or stablecoin were launched without a license, enforcement action
could be brought against the original founder (sticking with the DAI example, that would be Rune
Christensen, the entrepreneur who established MakerDAO, which is the DAO that maintains DAI),
or if control has been handed over to a DAO, against the managers of the DAO (Christensen acted
as CEO of the Maker Foundation, which managed MakerDAO at least into 2021) or significant
beneficial owners of DAO governance tokens (MakerDAO uses MKR tokens; venture capital firm
Andreessen Horowitz appears to have a significant holding of these MKR tokens).160
While regulators may sometimes struggle to assert jurisdiction over the relevant people
(either because regulators cannot determine their identities, or because they are located outside of
the United States and lack US assets to enforce judgments against), the licensing regime could still
help contain Dapps and stablecoins by prohibiting centralized intermediaries (like wallets and
exchanges) from providing any services in connection with an unlicensed Dapp or stablecoin.
Admittedly, there could also be jurisdictional issues associated with ancillary services that are
provided in more decentralized ways (the Uniswap exchange, for example, is more decentralized),
and so there is no single silver bullet measure that can stop the growth of DeFi. However, if DeFi
were forced to live up to its claims of decentralization by operating without any centralized
intermediaries, it would be very difficult for users to access DeFi or for DeFi services to scale
up,161 and this would limit the real-world fallout from any DeFi failures.
Until such licensing measures can be put in place, the SEC and CFTC should continue to
regulate stablecoins and Dapps as speculative investments where appropriate, and the Financial
Stability Oversight Council and the Office of Financial Research should continue to monitor the
DeFi ecosystem for potential spillovers that could harm the financial system and real economy. If
159
For a more detailed proposal for such a licensing regime, see Allen, supra Note 81 at 182 et seq. This licensing
regime builds on proposals made in Omarova, supra Note 4.
160
The Cointelegraph Top 100: Rune Christensen, COINTELEGRAPH, https://cointelegraph.com/top-people-in-
crypto-and-blockchain-2021/rune-christensen; andreessen horowitz, Maker, https://a16z.com/2018/09/24/maker/.
161
See Notes 122-127 and accompanying text.
V. CONCLUSION
For example, part of DeFi’s appeal derives from suspicions about concentrations of power
in the largest banks, yet some of these largest banks are considering how they can profit in the
DeFi ecosystem165 – DeFi could ultimately enlarge, rather than disrupt, the biggest banks. Real
solutions to the problem of “too big to fail” require actually shrinking the largest banks, and many
reforms have already been suggested to this end – what is lacking is the political will to implement
them.166 Interest in decentralization (particularly Web3) is also being driven by distrust of large
tech companies like Amazon, Google, and Meta (formerly Facebook) – but tech giants (and
venture capitalists) are already eyeing Web3 as an opportunity to profit.167 The large tech
companies pose very real threats as a result of their market power and dominance as disseminators
of information: these threats are beyond the scope of this paper, but as a start, legislation could be
adopted that prevents these firms from issuing stablecoins or providing any other financial
services.168
162
12 USC S 5463.
163
Hsu, supra Note 50.
164
Elizabeth M. Renieris, Amid the Hype Over Web3, Informed Skepticism is Critical (Jan. 14, 2022),
https://www.cigionline.org/articles/amid-the-hype-over-web3-informed-skepticism-is-critical/
165
See Notes 71-72 and accompanying text.
166
For a survey of these proposals and their political challenges, see Jeremy C. Kress, Solving Banking’s “Too Big
to Manage” Problem, 104 MINN. L. REV. 171 (2019).
167
Ephrat Livni, Tales from Crypto: A Billionaire Meme Feud Threatens Industry Unity, N.Y. TIMES (Jan. 18,
2022). For example, venture capital firm andreessen horowitz is one of the most influential promoters of Web3; its
founder Marc Andreessen also sits on Meta’s board. Id. Meta is aggressively moving into the web3 space. Brian
Quarmby, Rise of Web3: Metaverse tokens surge as Meta’s share price plunges, COINTELEGRAPH (Feb. 4, 2022)
168
Allen, supra Note 81 at 208-212.
169
Aaron Klein, Opening statement at roundtable on America’s unbanked and underbanked (Dec. 15, 2021),
https://www.brookings.edu/opinions/opening-statement-of-aaron-klein-at-roundtable-on-americas-unbanked-and-
underbanked/
170
Schuster, supra Note 10 at 997.
171
Klein, supra Note 169. For example, “[t]he single most impactful thing the federal government could do is to
give people access to their own money immediately. This can be done by simply amending the Expedited Funds
Availability Act to require immediate access for the first several thousand dollars of a deposit, instead of permitting
the lengthy, costly delays that harm people living paycheck to paycheck.” Id.