ICOs and Economics of Lemon Markets

16 min readAug 23, 2017

Extract from paper (published here) providing an in-depth analysis on the structural and dynamic economic properties of Initial Coin Offerings.

In this article we will discuss what happens when you apply the mental model of adverse selection and skewed market sentiment to Initial Coin Offerings (ICOs). But before we do so, we will provide an overview of Lemon Markets and their underlying mechanisms.

Market for Lemons

What would happen if buyers of a particular product couldn’t tell the difference between good and bad quality products? It would be fair to assume that sellers would try and push their products as good quality in order to maximise their returns. In this case of information asymmetry only sellers know the real quality of their products. We can then assume that uninformed or unsophisticated buyers will begin acting on overall market sentiment where the quality of the products are expected to be somewhere between the good and bad quality products i.e. there is uncertainty in product quality. What effect does this have on the market for a product? This is exactly the phenomenon explored by economist George Akerlof in his seminal paper published in 1970, ‘The Market for Lemons: Quality Uncertainty and the Market Mechanism’ [1], for which he jointly received the Nobel Prize.

Akerlof used a simple thought experiment related to the used car market in the United States, in which he quantitatively demonstrated the impacts of information asymmetry on the equilibrium state of a market. A high-level qualitative overview of his argument is as follows:

In American slang a “lemon” is a car that is found to be defective after it is bought.
Suppose buyers cannot distinguish between a good quality car and a lemon, therefore they are only willing to pay a fixed price for a car that averages their value i.e. prices will be between good quality cars and lemons.
Sellers know whether they hold a good quality car or a lemon i.e. they have more information than buyers, but there is no way for buyers to know whom to trust.
As buyers will only buy at a fixed average price, given quality uncertainty, means sellers will only sell when they hold “lemons” since the price of a lemon is less than the average sale price.
On the other hand, the fixed average selling price means that sellers will leave the market when they hold a good quality car, as the average sale price is less than the true value of the car they have for sale.
Eventually, as enough sellers of good quality cars leave the market, the average sale price for cars will decrease as quality deteriorates.
As the quality continually deteriorates eventually sellers of average quality cars will also exit, leading to further reduction in the average selling prices.
This continual lowering of the average sale price and resulting deterioration of quality leads to a downward cycle until the no-trade equilibrium is reached i.e. buyers cease to buy as they will expect only lemons.
Thus asymmetric information and the challenge of uninformed buyers making decisions based on general market sentiment rather than individual informed views results in the Adverse Selection Problem, driving good quality cars from the market and potentially leading to a market collapse.

This is obviously a highly simplified view of how markets operate; however it clearly illustrates how the interactions between variation of quality and asymmetric information can potentially lead to the disappearance of a market where guarantees are indefinite [2].

It must also be highlighted that The Lemon Market model is not a completely abstract thought experiment. One of the most prominent examples of its application was seen during the financial crisis between 2007–2012 [3]:

“Banks wanted to offload toxic assets. But they had better information than the market did about how toxic their assets were. The markets therefore discounted heavily what the banks offered for sale, so heavily that the Banks didn’t offer good quality assets for sale. But knowing this, and not wanting to buy worthless assets, the markets withdrew.”

Conditions for a Lemon Market

Akerlof’s simple but effective model of the impacts of information asymmetry on a market can be broken down to a number of conditions that can potentially lead to a lemon market:

Information Asymmetry: buyers cannot accurately assess the value of a product through examination before a sale but sellers can.
Skewed Incentives: An incentive exists for sellers to pass off low-quality product as higher quality.
Lack of Disclosure Framework: Sellers have no credible disclosure mechanism, so good quality is difficult to differentiate from low quality.
Low Quality or High Variation: There is a high variation of seller qualities or the average seller type is sufficiently low quality.
Lack of Assurance: There is a lack of quality assurance either through reputation, guarantees, regulations or other clear legal protections.

Trust in a Lemon Market

The interesting thing about Akerlof’s work is that he didn’t just leave us with a problem and no way of preventing or minimizing it. A key part of his work is the importance of trust for market stability. While he touched on how informal unwritten quality guarantees can be preconditions for trade and production, such as reputation and brand recognition, he also pointed to how formal licensing/regulations and other clarity around legal protection can help reduce uncertainty [4]. In this case there would be no adverse selection problem because an asset’s fundamental value would be known with some level of certainty.

Interestingly, five years after Akerlof’s paper was published, the United States put in place a federal “Lemon Law“ (Magnuson–Moss Warranty Act), and each state now also has laws that provide remedies to consumers for automobiles that repeatedly fail to meet certain standards of quality and performance, although these are not necessarily applicable to used cars that don’t have a warranty [5].

From Lemons to Bubbles

While Akerlof showed that in the extreme case a market subject to the adverse selection problem, resulting from information asymmetry, can lead to market collapse, such market “implosions” are not common, and something subtler takes place.

Douglas Emery took this adverse selection problem further and showed that a market that is under the “lemon” conditions will not necessarily collapse, but will go into what he referred to as a “negative bubble” i.e. where an asset will eventually be traded at a price that is below its fundamental value.

Emery demonstrates in his work ‘Negative Bubbles and the Market for Dreams’ [6] that:

“as an asset’s market price falls to a sufficiently low level, the potential value from buying the asset increases to a sufficiently high level to attract non-experts to purchase the asset — in spite of the risk of getting stuck with a lemon. Therefore, rather than a market collapse, assets in such markets can often sell for less than their fundamental value.”

The key conclusion of Emery’s work is that while adverse selection can lead to a market collapse this is not always so, in reality a “negative bubble” is more likely, where buyers can pick-up bargains. However, Emery extends this to show that a mirror image of negative bubbles can occur under certain conditions, where assets can start trading above their fundamental value — leading to positive bubbles. A condition that is more prevalent in markets compared to negative bubbles.

Emery demonstrates that information asymmetry and skewed positive market perceptions can lead to assets trading above their fundamental value. This can persist until all assets owned by experts have been sold to non-experts. In a more perverse situation the experts can then start to change their strategy and leverage market momentum to profit from short term investment horizons e.g. intraday trading. Therefore, assets start trading increasingly above their fundamental value, which can be considered a result of “experts exiting the fundamentals market”, or in other words those “in the know” trading on short term horizons and those “who don’t” holding for the long term. Eventually as an increasing number of market participants, experts and non-experts, start acting on shorter term investment horizons, severe distortions occur in the perceived value of assets, thus leading to an unstable situation and the formation of a positive bubble (diagram below).

Positive bubbles will persist until the short term investors, mostly the “experts acting on short term horizons”, completely exit the market, potentially due to negative news either directly or indirectly related to the market, which leads to a bursting of the bubble. In such situations, the last ones left holding the assets are normally the investors acting on medium/long term horizons, and who may also be least able to afford the losses [7].

Market Dynamics Based on Return and Horizon Preferences

Beware the ICO Lemons

Our foray into market dynamics and stability shows us that conditions that lead to lemon market scenarios may result in market collapse or more likely a negative bubble, but if such conditions are accompanied with positively skewed market sentiment e.g. rising asset values, a positive bubble can ensue where buyers chase dreams.

Ultimately, Lemon Market conditions, with skewed market perception, makes it difficult for buyers to determine the fundamental value of an asset, thus distorting the market and driving instability. Using this mental model, we can assess how each of the Lemon Market conditions applies to ICOs and tokens, and what context this provides us of the current state and an optimal path forward.

Information Asymmetry

Buyers cannot accurately assess the value of an ICO token through examination of the project details before a sale but sellers can. The concept of ICOs is not complex, it is just about users exchanging one asset (cryptocurrency such as bitcoin or ether) in return for another asset (tokens), which they believe will have higher value for them in the future e.g. through financial gain or utility by accessing some “decentralised” product or service. This is like an initial public offering (IPO) of financial assets, but here the asset is a novel instrument that has little or no precedence for comparison or valuation. In most cases the issuing organisations do not even have a product let alone established product market fit, so models for supply/demand are few and far between. Ultimately the only tangible basis to assess any form of utility and hence value offered by these assets is based on what the issuer provides, either through a white paper or other forms of communications.

In this context, the flow of information is mostly one-sided, from seller to buyer, and in many cases social influencers are leveraged to maximise on the overall hyped market sentiment that “cryptocurrencies are on an upward trend”. Therefore, the token buying public, who may know little or nothing about the issuing organisation and the technical concepts behind the offering, can only trust that the issuer and its spokespersons are truthful, competent and committed to delivering what is stated. No one, apart from the issuer can truly know the answer to this, and as the issuer has no fiduciary duty to the token holders, they can pretty much say anything to maximise interest in the ICO and the ongoing demand for the issued tokens!

Skewed Incentives

One of the key consequences of a lack of accountability and a minimal time/financial cost associated with executing an ICO is that there is a clear and indisputable incentive for issuers to pass off low quality products as high quality ones. Skewed incentives enable issuers to disproportionately profit from a particular behaviour as the potential gains outweigh any direct risks. Some of the key skewed incentives are as follows:

Perceived Zero Accountability: Issuing companies currently have a sense of zero accountability for claims made during an ICO process. This is supported by detailed disclaimers and small print.
Limited Ongoing Responsibilities: In many cases an ICO is structured to appear as an “open source infrastructure layer” and the financial contributions are deemed to be donations made to an abstract “non-profit” co-operative, even though an incorporated company may be used to deliver key solutions. As such the ongoing value of issued tokens is considered to be unconnected with the efforts of the delivering organisation or the team members linked to the ICO, and the ongoing performance is considered as a community responsibility.
Minimal Cost and Complexity: There is a minimal upfront cost and complexity to execute an ICO, with biggest cost being time to build up a community following and draft a white paper. Driving a community following can also be accelerated at minimal cost by leveraging social media influencers who feel there is very little accountability and or reputational impact on them if things go wrong.
Minimal Technical Requirements: Due to lack of requirements for a minimal viable product and or detailed technical specifications, technical barriers to entry are only limited to convincing and recruiting advisors and claims of future hires with key skills. In many cases key technical contributors are sought who provide “reputation as collateral” [8] to legitimize the offering and signal to consumers the capabilities behind the ICO.
Scarcity and Liquidity: Implementing a monetary policy that is based on token scarcity and with a liquid secondary market, combined with implied claims for demand of tokens for access to future products, there is a perception the issued tokens could increase in value in the short term and early buyers can sell out. This drives a fear of missing out (FOMO), and hence higher initial demand for the ICO digital token.
Perceptions of Growth: The industry currently has a positively skewed perception of the ICO phenomenon and cryptocurrency in general. While there is a growing sentiment that the market is reaching a peak, at which point either regulators will step in or buyers will start exiting due to over inflated prices for subpar tokens, there is still a view that short term token investors can make a quick gain and pass on their holdings to unwitting long term investors on a liquid secondary market before the bubble pops.

Lack of Disclosure Framework

Like most industries in their infancy there is little or no disclosure requirements for minimum standards on type, structure and quality of information provided for an ICO process. In addition, any information that is provided, normally through a simple whitepaper, undergoes no third party review for assurance on accuracy, currency, reliability and correctness of disclosures. In many cases, in particular for the larger well organised ICOs, whitepapers come bundled with numerous disclaimers and waivers, which make any information contained in documents next to useless for gaining an accurate and truthful state of the project, as the project owners pretty much absolve themselves of any accountability and or responsibility, the equivalent of using the term “as is” on the sale of a second hand car! Finally, any future action by holders of tokens may be futile as projects can also contain small print such as: “…CONTAINS A BINDING ARBITRATION CLAUSE AND CLASS ACTION WAIVER, WHICH AFFECT YOUR LEGAL RIGHTS.”

In a world where assessing risks is hard enough as it is, doing so for an early stage company or project is even more pronounced. Therefore, in these cases a robust or trusted disclosure mechanisms is normally a key barrier to manage risk. This could come in the form of mandatory information audits through a trusted source, which would then be supported by some form of ongoing disclosure (e.g. financial reporting) to ensure a project is managing its finances and risks prudently and meeting key milestones. Such frameworks would enable sellers to differentiate themselves and their good quality (credible claims) from low quality (inflated claims), and thus ensure some form of accountability and responsibility when it comes to transparency. However, such disclosure frameworks are unlikely to be viable for most early stage companies and projects, as any startup investor or entrepreneur can testify to. For a project owner to truly provide information that can differentiate the quality of their project in this industry is mostly through a working product/service and some form of market traction… maybe this is why Satoshi Nakomato released a product rather than doing a crowd fund on a white paper!

Low Quality or High Variation

There is no doubt that there is significant variation in the quality of ICOs. However, what is not so obvious is if the average quality across all projects is low. To make things even more complicated it’s also unclear how to actually quantify quality.

If we start with the assumption that in most ICO cases there is little or no product let alone a product market fit! Therefore, focusing on other tangible aspects, such as marketing material and project details related to current issuances, we can see there is significant variation in relative quality, which can be seen in the writing style and technical depth of whitepapers, quality of websites, social media communications, and most prominently in the skills and experience of the teams and advisors. There are obviously some well financed projects with major backers, which can afford the development of quality marketing material and hire experts in communications and legal matters. But like in any early stage investment, this is no firm indicator of a viable project or of a project that has a higher than average survival probability. As one VC noted (who shall remain anonymous), such factors normally make it harder to identify the real quality, or as he put it: “such shallow niceties are like polishing a turd for a beauty pageant!“

The truth of the matter is that there is currently a lack of data related to how quality can be determined for early stage projects in general, let alone ICOs, and if such funding mechanisms gain wider adoption we will see over time how the survival rates pan out for various project types. Using simple rules of thumb, as for traditional early stage equity and debt deals, investors normally attempt to qualify a project based on fundamentals such as product, team, market and in many cases on basic financials. While financials are pretty arbitrary for many early stage projects, they do provide a level of robust thinking on how value is created by the product or service, the business model that may be leveraged to commercialise that value, and how the size and competition in the market will influence the trajectory of operations and commercialisation. As noted, this can be a “finger in the air” process for early stage projects, but it does provide a somewhat structured framework to define a possible roadmap and key milestones, which can then be modified as further information comes to light as a startup progresses — akin to a hypothesis driven scientific process. However, such structured thinking is rarely (if ever) applied to ICO issuances, in particular for cases that are oddly marketed as “non-profit” and or “infrastructure” yet the issuing company may also be developing proprietary end-point applications that can plug into the “open-source” layer (which is often nothing more than a simple smart contract).

In general, within the ICO world, there is little or no economic analysis on the optimal monetary policy (supply release and cap) and fiscal policy (commercial benefits for holders), thus the current issuance structures are mostly based on what would drive an initial rush of investor interest with little thought of managing ongoing supply and flow. This is essentially like an early stage equity/debt investment being executed with no justification of valuation or returns based on possible market size and or demand for the product (even if such numbers may only provide an arbitrary baseline). Based on this thought process, even though we will likely have to wait a number of years for project failures to crystalise, we can say with a certain level of confidence that the depth of economic analysis performed for issuing digital tokens will likely be a sign of quality, and right now this is at an all time low across the board!

Lack of Assurance

The lack of assurances by issuers is one of the key points of the ICO token market that makes it a form of regulatory arbitrage. If funds were raised from a number of investors for an early stage project through a securities issuance, there would be a substantial amount of protection for investors. However, digital token issuances executed through an ICO come with numerous disclaimers and little or no guarantees, as highlighted in the Lack of Disclosure Framework above. There is a deliberate perceived distancing of responsibilities and accountabilities of the issuing persons and organisation from what is delivered and the value of the token being issued.

The lack of assurances combined with the skewed incentive structure discussed above provides a combination that can lead to companies (deliberately or through carelessness) driving up the perceived value of the tokens through grandiose statements, but at the same time absolving themselves of any responsibilities of the token’s market performance if such claims are not fulfilled.

So is it a Lemon?

If we have a crowd that is socially influenced, a market that suffers from skewed asymmetric information and little or no controls or assurances we will likely see assets trading outside of the bounds of their fundamental value. In such cases it is highly likely a positive bubble will persist until there is some abrupt news that forces “experts” to exit the market, which will leave “non-experts” holding assets at an inflated value. Without previous trading volumes to support the inflated value, asset prices will begin to drop until they are at or below the fundamental value. To some extent this simple mental model was highlighted by Hyman Minsky in his work “The Financial Instability Hypothesis” [9], albeit from a wider economic context where periods of prosperity encourage greater leverage and progressive recklessness, where excess optimism creates financial bubbles and the later busts. He argued that such up and down cycles are endemic and inevitable in a capitalistic economy, which is a type of market failure and requires some form of controls.

All signs point to the ICO and token market suffering from lemon conditions, combined with the positive hype of a new technology that created a small group of early winners (in Bitcoin and Ethereum). In addition, with the lack of controls and accountability on marketing of ICOs, social influence through widely accessible and open public channels, can drive crowds to become “ineffective”. As a result we’re seeing digital assets being created and traded that will likely never live up to their current perceived value.

However, this does not mean that market participants are “irrational fools” chasing dreams, who only end up holding lemons when the market plummets! All this means is that without a controlled environment all participants will act in their own self interest, and unfortunately the long term investors or late adopters will end up losing out as the market corrects to find its fundamental value. In many cases the fundamental value of many tokens may be zero as the market realises some projects are the equivalent of trying to create Netflix on 1995 infrastructure — where the early stages of the technology and or lack of adoption means timing is misaligned [10]!

So if lemon conditions are prevalent, there is skewed positive market perception and a surge of late adopters inflating prices, what can be done to drive market stability, or at least prevent an all out collapse? This is something we discuss in part 3 of the full paper on the Economics of Initial Coin Offerings here.

To receive thoughts, discussions and papers on finance, economics, law and the human condition follow on twitter: @avtarsehra

Thanks to Vic Arulchandran for his contributions to the refinement of this and upcoming work.