top of page

The Collapse of FTX and Crypto Economics

This note is motivated by the collapse of FTX and the publication of

the fascinating book, Proof of Stake by Vtialik Buterin, the major

developer for Ethereum. Buterin, as well as the founders are FTX and

others in the crypto space are brilliant individuals. Yet, the space

is obviously facing growing pains. The message of this note is that

effective regulation of the crypto industry requires integrating

knowledge from experts working in a number of areas. I also point out

that we can use micro-economics, particularly the research discussed

in Chapters 1-4 of Advanced Micro-economics (subsequently cited as

AM), to identify four distinct areas of expertise that are discussed

in turn:

  • crypto-technology - the technical advances to the design of block-chain, smart contracts and digital currency.

  • crypto-markets - the public market for smart contracts, digital currency and other products that build upon the block chain and related technologies.

  • crypto-regulation - the public regulation of crypto-markets.

  • crypto-science - Measuring the causal effect of interventions in crypto markets, particularly the evaluation of regulations.

The collapse of FTX provides a reminder that markets are extremely

complex and cannot operate effectively except in the shadow of behind

the scenes regulation. The first welfare theorem of general

equilibrium theory (see appendix of AM) shows that when markets are

complete, then competitive markets are efficient. As discussed in

Chapter 1 of AM, this result very much depends upon having a thick

market for all commodities, combined with the requirement that the

exchange of the commodity can be enforced at no cost.

Neither of these requirements are satisfied in practice. Much of

modern economics can be viewed as understand the limits of markets and

working out appropriate regulation.


Crypto-technology refers to block-chain and related technologies.

Technically, the technology is a brilliant synthesis of game theory

and complexity theory. Standard game theory assumes that parties are

aware of all the strategies available to them. Individuals assumed to

choose what they believe is the optimal strategy from the available

set, given the anticipated behavior of their counter-parties.

Complexity theory is a sub-field of computer science that provides

methods to evaluate the difficulty making decisions that are modeled

using computer algorithms. Hence, when combined with game theory, it

can be viewed as providing a theoretical foundation for a research

agenda started by Nobel prize winner Herbert Simon in the 1950s.

Complexity theory provides a formal way to measure how long it will

take to comprise security systems, that in turn is an important

component of block-chain technologies. The block chain technology

builds on these ideas to ensure that the transfer of property rights

can be safely verified without government intervention (See appendix

of Buterin's book for a nice outline of the bitcoin technology). Smart

contracts build on this technology to provide mechanisms for the

transfer of property rights that can be conditioned upon certain

events that do not require the intervention of a judicial system. For

example, transfer funds to a seller once a product has been verified

as delivered to a buyer.

Conceptually the block-chain technology is not new - it can be thought

of as a super-charged property rights system that is part of the

evolution of property rights systems started thousands of years ago.

For example, early property rights systems relied upon individuals who

follow accepted norms of behavior. We see this in the book of

Deutoeronomy in the Old Testament (chapter 27, verse 17): "Cursed be

he that removeth his neighbour's landmark. And all the people shall

say, Amen."

The promise of the block-chain technology is to provide a

decentralized system for the trading of property rights that is

democratic and not controlled by any government, while at the same

time lowering the transactions costs associated with the verification

of property rights. Transactions costs are so low that one is able to

allocate property to low value digital coins - a process that is akin

to assigning a property right for every dollar bill in one's pocket.

In an important paper, Abadi, Joseph, and Markus Brunnermeier.

“Blockchain Economics” provide a

brilliant economic analysis of what is possible with a block chain

technology. Using ideas from game theory, mechanism design and

asymmetric information (based in part on the economic theory discussed

in chapters 3, 6 and 8 of AM) they show it is impossible to have a

consensus mechanism, such as implemented with proof of work or proof

of stake, that is resistant to computer network failures, resource

efficient and allows any transfer of value that parties have agreed


Thus, the paper proves that there are limits to decentralized finance

systems with self-interested individuals. In particular, any system

that is implemented must compromise between the three goals, a

trade-off that Abadi and Brunnermeier call a trilemma. Resolving the

trilemma is not a theoretical problem, but one that is ultimately

solved with practice in crypto-markets.


A competitive crypto-market is the place where the block-chain

technology can be used and tested. Note that the everyday term for a

competitive market differs from the term as used in economics.

Technically, economists define a "competitive market" as a situation

for which there are many sellers of every commodity, and that parties

can enter into agreements to trade specified quantities of a commodity

at a specific time and place. As discussed in Chapter 1 of AM.

competitive markets, as used in the welfare theorems of economics, are

an abstraction that helps us understand the stringent conditions

required for efficient production and distribution. Give that these

conditions are never satisfied in practice, observing market failures

may be viewed as consistent with economic theory!

In practice, as Hayek emphasized many years ago, markets are chaotic

institutions for innovation and the testing of new products. From this

perspective, crpto-markets provide value by exploring potential

applications of cypto currency. It also allows one to see how the

trilemma, highlighted by Abadi and Brunnermeier, is resolved in

practice. We see this in the fact that bitcoin (that uses proof of

work) and ethereum (that uses proof of stake) rely have very different

resource foot prints. It will be interesting to see over time whether

one system becomes dominant, or whether there will be many different

block-chain protocols co-existing at the same time.

Hence free markets provide a venue for new and exciting innovation,

and over time we learn which solutions are the most useful. However,

the very fact that information is expensive also implies that free

markets can be exploited by harmful agents who sell snake oil, or

other products of dubious quality. As the Noble laureate George

Akerlof showed, when the uncertainty regarding the quality of a

commodity is sufficiently high, then markets may even shut down, and

we may lose the benefit of productive innovation and exchange. In the

case of FTX, the market failed when market participants realized that

the exchange may not have sufficient assets to cover their

liabilities, leading to a run on the exchange.

Such runs are a familiar feature of financial markets. This year's

Nobel prize in economics was awarded to Bernanke, Diamond and Dybvig

for their seminal contributions on the sources of financial

instability. For example, they showed that in theory seemingly small

fluctuations in beliefs may lead to the inefficient failure of banks.

This research ultimately helps provide guidelines for regulation. The

question then is what should be the appropriate policy response in the

case of failures in crypto markets?


Many policy makers worry that excessive regulation may stifle

innovation. It is worth highlighting the fact that most regulation is

not prospective, but reactive, and a consequence of observed market

failures, and not a cause.

A good example of this is the regulation of food and drugs. In the

"Poison Squad", Deborah Blum provides an amazing history of food and

drug regulation in the US. In particular, most regulation did not get

enacted until many individuals were made ill or died from defective

products sold on a free market. Thus, while some may argue that crypto

should have been regulated earlier, the evidence suggests that policy

in United States takes a lassez faire attitude towards markets, and

only regulates when there is clear evidence of a need. Even then, as

Blum documents, the regulation may evolve very slowly over time.

Similarly, the failure of FTX is yet another example of a commodities

(such as investment into interest bearing accounts on the FTX

exchange) sold on a free markets that did not have the advertised

characteristics. The failure of free markets has a long history, as

Michael Cassidy brilliantly documents in /How Markets Fail/. The book

does a wonderful job of introducing the general reader to modern

economics in an accessible fashion. He views economic policy as a

trade-off between "utopia economics" which he characterizes as

unbridled support for free markets, versus behavioral approaches (see

the preface to the 2021 edition of his book).

This is a reasonable representation of the public debate, but not

representative of modern economics where it is recognized that all

markets are imperfect, in large part due in large part of information

failures. The reason that the information failure perspective is not

highlighted in the public debate is because it is /complicated/!

Chapters 6-10 of AM reviews a small subset of the research that

studies the implications of information constraints on the observed


The challenge is that a great deal more work needs to be done in order

to understand how to best trade-off more versus less regulation. In

particular, it is worth highlighting that FTX was not unregulated. The

FDIC sent them a letter on August 18th, 2022 stating that FTX was

making misleading statements regarding their products, and asking for

clarification. Moreover, the fact that the principals of FTX have been

charged with criminal fraud charges, and that two of these executives

have already pleaded guilty is evidence that there were consequences

associated with the miss-information regarding the commodities FTX

supplied to the market.

What Cassidy, and other commentators on economics appear to avoid

discussing is the tremendous progress that has been made in economics

in providing a framework for improving policy.


The 2021 Nobel prize in economics was awarded to David Card, Joshua

Angrist and Guido Imbens for their contributions to economic science.

By the term science I mean research that explores the actual

implications of policy, rather than the hoped for implications. For

example, in his work with Alan Krueger, David Card showed that there

are labor markets in the US where increases in the minimum wage

increases employment. Simple labor market models predict the opposite

and hence there were many criticisms of the work at the time. Yet in

the end, it should be the data and not ideology that determines the

quality of a theory.

A goal of economic science is to measure the causal impact of policy

changes. A contribution of Card, Angrist and Imbens is to clarify that

conditions necessary for a prediction on the effect of policy to be

valid using available data. One necessary condition is the ability to

have multiple observations of the effect of a policy change. This in

turn illustrates the point that science is very different from

innovation. When a new products is brought to market it is a unique

event, and hence in many cases it is impossible to predict how well

they do (for example, consider the many expensive movies brought to

market that are financial failures).

In contrast, a scientific claim depends upon repeated observations of

the same event subject to different treatments. What make economics

science so difficult is the challenge of having repeated observations

of comparable events in a rapidly changing environment. This problem

is particularly acute in the climate change debate. In order to be

able to build credible predictions regarding the impact of different

climate policies one would need to try them out a several similar

planets, something that is clearly impossible.

Rather, practical policy relies upon a combination of science and

models of the world based upon science that attempt to extrapolate

knowledge to new domains. Policy, like new products are tested in the

public domain. This highlights the point both markets and science are

institutions that generate knowledge, but through fundamentally

different processes. In particular, both entrepreneurs and scientists

can be highly skilled, and thus convincing because they are skilled.

In /Proof of Stake/, Buterin presents a number of ideas for crypto

markets that he argues will be useful. It is natural that the

entrepreneur have faith in their new ideas. However, since these

claims are not science, thus one needs to be cautious.

By crypto-science I mean claims regarding the industry that can be

validated through careful research design. Crypto-science needs to be

differentiated from crypto-technology. For example, the results of

Abadi and Brunnermeier are contributions to technology since the

results are based upon stylized models of behavior. In contrast, the

goal of science it to understand actual behavior. The science is not

always sufficiently developed for many important policy questions.

Hence actual policy must rely upon a combination of science and market

experimentation. It is useful to differentiate the two sources of

knowledge in order to be ready to modify policies as new information

is acquired.

Let me conclude by discussing two questions for micro-economic science

that the FTX case raises. The first of these is why did very large

firms invest in FTX? There were news reports of firms who considered

investing in FTX, but upon further investigation decided again

investing. They apparently made the right decision. What explains the

bad decisions? A similar question arise in the 2008 market failure

where some investors were aware of the low value of mortgage backed

securities months before the market price adjusted to reflect this

information. Basic finance models assume that information is

transmitted quickly and efficiently through modern financial markets.

This is clearly not the case and at the moment we still do not have

good theory or evidence to explain the transmission of information in

modern markets.

A second question is why did the principals of FTX commit fraud? Is it

because they did not expect to go to jail? Did they not understand

that the firm would eventually fail? Were the short run returns,

including being part of an interesting and exciting project, out

weight the cost from time they will spend in jail?

Obtaining scientific answers to these questions is central to how one

should regulate markets. The fact that in the case of FTX the

potential jail time did not deter the fraud may explain the demand for

/ex ante/ regulation. This is regulation that places reporting

requirements upon firms before any malfeasance has occurred. Such

regulation places costs on both well manged firms as well as firms

that might commit fraud. If fraud could be deterred with high future

penalties, but as we saw from the FTX case, deterrence is not always


One of the foundations of micro-economics is the hypothesis that

individuals respond in predictable ways to rewards and penalties. We

clearly need much more work on understanding the limits of deterrence,

and in turn the implications of these limits for the effective /ex

ante/ regulation of markets.

81 views0 comments

Recent Posts

See All

Deep Learning vs Statistical Learning

The purpose of this blog is to discuss the distinction between deep learning and statistical learning models. Discussions about regulating algorithms tends to lump together deep learning and statisti

Choice Architecture

The starting point for micro-economics is the hypothesis that individuals make choices consistent with well defined goals subject to resource constraints. Section of 3.2.1 of AM (Advanced Microeconomi


bottom of page