DeFi: What it Is and Isn’t (Part 1)

via medium

I interviewed industry professionals at top law firms, investment funds, exchanges, and various decentralized finance (DeFi) projects, in addition to attending panels and conducting independent research, to better understand the the emerging DeFi industry. DeFi: What it Is and Isn’t is the first of a three part series that draws upon the insights derived from this research.

The article that follows will outline the differences between the DeFi dream and the DeFi reality, and highlights 7 of the most pertinent challenges the industry must overcome before bridging the gap. Subsequent articles will analyze the decentralized exchange of assets and decentralized lending and derivatives. These articles assume a foundational knowledge of both financial markets and blockchain technology.

The most widely used application of blockchain technology is in the creation of digital currencies, which has required the development of financial markets to support their exchange. However, these financial markets, in their current state, prevent fair and open access. Furthermore, the infrastructure that supports blockchain-based markets is vulnerable to counterparty risk, censorship, a lack of transparency, and manipulation as it remains largely centralized. Current infrastructure flaws erode trust and inhibit adoption.

Just as the internet allowed for the creation of a new information infrastructure, blockchain technology allows for the creation of a new financial infrastructure and the development of entirely new markets. The DeFi movement has emerged in an attempt to make the infrastructure that supports new and existing blockchain-based markets just as decentralized as the underlying technology.

The DeFi Dream

In a functioning decentralized financial system, internet connection would be the only prerequisite to accessing financial services, rather than geography or circumstance. A reduction in the centralization of those that control and own the infrastructure underpinning financial markets would increase transparency, decrease costs, and reduce the opportunity for censorship and / or manipulation. The global “unbanked” (both individuals and enterprises) would gain access to financial services. DeFi would not only facilitate markets for illiquid financial products that already exist but would also allow for the creation of new financial products and markets that don’t yet exist. The ability to more effectively arbitrage, borrow, hedge, and access liquidity would spur more institutions to join the movement without restricting the usage of these products and markets exclusively to them.

The DeFi Reality
While DeFi is referred to as “open finance” and is sometimes touted as a way to “bank the unbanked,” the truth of the matter is these products aren’t aimed at the mass retail investor, let alone the global “unbanked.” Technology is not usually the primary factor restricting access to financial services. More often than not, identity and/or oppressive regimes are. On the retail side, it is unreasonable to believe that the average retail investor would understand the risk profile of even the simplest DeFi products and I’ve yet to hear a compelling argument for why the average retail investor would need access to exotic financial derivatives. UX/UI challenges present further obstacles to retail adoption.

Product-market fit for most institutional investors is not any clearer. Most DeFi projects aren’t well suited for HFT (high-frequency trading) since they’re limited in terms of speed. They are not well suited to trading large positions either and traditional finance institutions won’t even consider entering into transactions in which their counterparty is unknown. At this point, institutional traders aren’t willing to make these trade-offs so long as they can trust at least one other person in the markets in which they operate.

As a result, product-market fit is currently constrained to crypto-native power users that are comfortable with the existing UX/UI challenges of crypto networks and blockchain-based assets. For user adoption to accelerate, the trade-offs required by DeFi interfaces must not outweigh the perceived benefit of ownership, access, and transparency when compared with centralized alternatives. A lack of developer education and tools is also restraining the industry. Tooling and services need to be further developed and a best-in-class stack needs to be established so that users don’t have to interact with a fragmented ecosystem marching in fragmented directions. Developers, regulators, lawyers, investors, professional traders, and retail users will have to work together to overcome existing challenges.

As a result, the transition towards a decentralized financial system is more likely to be evolutionary than revolutionary. That doesn’t diminish the potential of DeFi to reshape the way in which markets function and the way that the entire world interacts and transacts.

While the DeFi movement has the potential to provide meaningful benefits over centralized alternatives, there are many practical challenges the DeFi industry needs to overcome first. User adoption may be the biggest impediment to the development of the industry at present, new risks compounded across protocols may be the biggest threat to its sustainability. An in-depth look at the 7 most pertinent challenges follows.

1.) Identity and Reputation
The first step in entering into a financial transaction often requires the identification of transacting parties. However, a core tenant of DeFi is that one’s ability to access financial services should not be dependent on most aspects of identity. This is problematic as violations of KYC / AML / OFACregulations can not only result in large fines but could result in criminal charges. If a DeFi Relayer (an entity that hosts on order book on a DeFi protocol) facilitates an exchange between unknown parties and those parties turn out to violate any of these regulations, the consequences may be serious. Furthermore, without a way to enforce identity, most proposals for decentralized governance of these projects are quickly reduced to plutocracy.

While still far from complete solutions, projects are researching ways to allow for KYC (know your customer procedures) without introducing centralization. For example, Relayers on 0x can opt-in to implement a permissioned liquidity pool that ensures that pool is only accessible to whitelisted Ethereum addresses that meet certain requirements, such as those required by AML (anti-money laundering) and KYC policies. However, this method still doesn’t ensure identity in a way that allows one to know that a counterparty is trustworthy without excluding those outside of the traditional financial system and introducing centralization. Several parties have issued EIPs(Ethereum Improvement Proposal) to incorporate KYC/AML compliance into ERC-20 tokens. However, in many cases, these proposals would still require service providers to work together off-chain via a consortium to review each others’ KYC policies and it is still unclear whether these proposals would fully satisfy regulatory requirements.

Establishing reputation in blockchain networks is a distinct challenge. This is a strong industry focus as the range of possible products expands when there is a sense of on-chain reputation. There are currently two main ways to attempt to establish reputation in these networks:

  • Allowing everyone to start on an equal playing field under the assumption that network participants are good actors. Participants that prove to be untrustworthy / uncreditworthy would subsequently be slashed (punished.) Underwriters on the Dharma network fall into this category, wherein they gradually build reputation over time via an on-chain record of their accuracy and behavior.
  • Porting existing credit data to a blockchain network via an oracle. This is hardly an improvement over the traditional finance system in terms of allowing for fair access.

The lack of an on-chain reputation method that doesn’t require users to reveal too much about their personal identity means most DeFi projects require (over)collateralization in lieu of being able to establish trustworthiness.

2.) Capital Inefficiency

The overcollateralization required by DeFi projects is capital inefficient. MakerDAO requires users to deposit 1.5x the value of ETH to establish the collateralized debt position underpinning Dai (CDPs will be covered in more depth in Risk Off or On?: Decentralized Lending and Derivatives .) Even still, most people choose to keep their “loan-to-value” ratio at 300 percent in order to avoid double digit liquidation penalties.¹ Similarly, Compound requires a 2x collateralization ratio, which the company says will decrease over time.² However, some users indicated a willingness to post 4x–5x the required collateral.²

Until a decentralized reputation system is developed, there is little choice but to require users to lock up excess capital, dulling the benefit of taking out these positions to begin with. Even when/if reputation is solved, the volatility of the underlying positions could result in a persistent preference to overcollateralize.

3.) Oracles

Corruption of on-chain oracles (the mechanism that finds and submits real-world data to a smart contract) is a huge concern for these systems since liquidation occurs automatically in the event that collateral levels drop below their specified “loan-to-value” ratios. Different DeFi projects approach oracles in different ways, but many projects in the space are using MakerDAO’s oracle. MakerDAO’s oracle is currently designed to support single collateral Dai (backed entirely by ETH) but will be re-designed to support multi-collateral Dai (backed by a pool of different cryptocurrencies) in the near future. MakerDAO’s oracle pulls data from sixteen different sources for its oracle feed. These sources are comprised of Ethereum addresses voted on by MKR token holders, which are then submitted to an autonomous smart contract. The oracle chooses the median of all sixteen submitted data points. This system allows for 51% tolerance as it excludes the outliers which are more likely to be submitted by malicious actors.¹ Importantly , MakerDAO also utilizes an oracle security module in which the second layer of the protocol can activate an emergency shut down. This shut down freezes the system at its last known “safe state” if it has reason to believe the oracle may have been compromised. If an emergency shutdown occurs, users can convert their Dai to ETH at the equivalent of 1$/1 Dai, according to the state of the ledger at its last determined “safe state.¹”

Single collateral Dai oracles update every time the price of ETH fluctuates by +/-1.0% but multi-collateral Dai (MCD) oracles will update once an hour.¹ This allows the sixteen oracle inputs to be viewable for an hour before they are acted upon, increasing transparency. However, such a long lag time may not be appropriate considering the volatility of cryptoassets. The company’s argument that this delay can be compensated for by the risk model is questionable. Furthermore, liquidation of collateralized positions (essentially defaults) will be executed via auction with MCD, which means it will “six hours or more” to liquidate positions as the protocol accesses “all the arbitrageurs and liquidity across the whole marketplace and ecosystem.¹” The impact of having to wait 6 hours to unwind a single position during times of market distress, or failure, would be significant.

Compound takes a different approach with its oracle, aggregating and averaging price feeds from a series of exchanges and posting them on-chain consistently. The data updates every time the underlying value fluctuates by +/- 0.1%, but data is updated on-chain every 15–30 seconds, confined by the processing speed of Ethereum.² Given the importance of oracles in these systems, DeFi projects may want to more closely consider which method they use or choose to implement their own methods.

4.) Network: Platform, Liquidity, Scale

Most current DeFi solutions are built on top of Ethereum and therefore DeFi’s adoption is tied to the scalability and usability of the Ethereum network. The scalability debate is well known (and addressed below) while usability remains a challenge as mainstream users still struggle to easily interact with Web 3.0.

While the composability of protocols built on Ethereum creates even larger switching costs, it also introduces network risk. As more projects build on Ethereum, it may become harder to upgrade the base layer protocol in a way that allows for backwards compatibility.

Part of the power of DeFi is that it allows for the creation of new markets. However, decentralized markets suffer the same circular problem that all new markets do: adoption is required to generate liquidity, but liquidity is a driver of adoption. While DeFi can enable new markets and allow new participants to access them, it does not automatically create liquid markets for these products. This is a problem because assets that are illiquid tend to trade at a discount to their liquid counterparts.³ It also creates inefficient pricing as opportunities for arbitrage go uncaptured since it remains difficult to move quickly and seamlessly between crypto markets.

Alex Evans of Placeholder VC breaks down the models of current DeFi networks into three broad categories:

  • Those that require users to find peers to trade with. Augur , 0x , Dharm a
  • Those that pool “maker” assets and offer them to “takers” for a fee. Compound , Uniswap
  • Those that set parameters through governance, allowing users to trade directly with a smart contract. Ex: MakerDAO

Each model has implications for liquidity. The lack of requirement to find a specific peer with which to trade seems to be the design advantage of the top protocols. These protocols also tend to offer fewer options in terms of products / use cases, which pools demand, facilitating better liquidity. Alex Evans also believes automatic and consistent processes (MakerDAO) better facilitate liquidity than bespoke and varied ones (Augur.) This seems to have been one of the drivers behind UMA and Dharma deciding to set tighter parameters on their products (relative to a completely open system in which individual users set all parameters.)

“At least initially, the markets that have built deep pooled liquidity in a handful of important markets appear to have the adoption lead versus those that have tried to create a multi-asset infrastructure.” — Alex Evans

Assuming these markets find a way to bootstrap the necessary liquidity, blockchain infrastructure is not yet scalable enough to process volumes similar to those processed by centralized exchanges. For a sense of the limited scale of current DeFi networks, investor at Paradigm, Arjun Balaji, predictsthat December 2019’s aggregate volume on 0x will lag a single day’s volume on Coinbase. While advances are being made in Layer 2 scalability and innovative solutions such as StarkDEX (currently partnering with 0x) show promise, current blockchain infrastructure has a long way to go before it can support volumes similar to those supported in traditional markets.

Front-running, and other opportunities for manipulation, on DeFi networks will be addressed in Trade-Offs: Decentralized Exchange.

5.) Business Models Still Undefined

While there are many options, most DeFi projects have left their monetization method “undefined” and are focused on “defining the incentives of the protocol at large.⁴” However, at some point these projects will need to generate revenue if they are to persist.

dYdX highlights three main monetization models for DeFi projects:

  • Value accrual via a native token. MakerDAO (MKR)
  • Monetization via fees. P otentially Compound
  • Monetization via a user facing application. dYdX, Dharma, etc.

In most cases, a native token monetization model introduces another layer of friction to user adoption. For other projects it might not make sense. For example, a token monetization model doesn’t make much sense in networks where ownership / voting percentage can be determined by participation, which is recorded on-chain.² Nadav Hollander of Dharma points out that a fee model implemented at the protocol level, in addition to being somewhat anathema to blockchain ideology, could easily be forked away.⁴ However, Compound is not against keeping a small amount of the interest flowing through the system in a model akin to the AUM model in traditional finance.²

The latter appears to be the prevailing model. Dharma, dYdX, and others found that they needed to build out full stack products (Expo on dYdX, for example) because they found that developers weren’t willing to invest the time necessary to build on these new protocols. While the 0x model is often touted as the exemplary model, 0x’s success was enabled, in part, because there was already an existing market for DEXs (decentralized exchanges.) 0x’s protocol opened into an existing market, whereas these new DeFi protocols have to create new markets from scratch.

In an effort to bypass many of the challenges of creating a two-sided marketplace from scratch, it’s likely that new DeFi protocols will continue to build out full stack services and monetize those, at least over the near term.

It is important to remember that creating a marketplace is a service business and that is unlikely to change. Whatever entity enables a marketplace also has to offer services to both the demand and supply sides. Marketplaces can’t be created out of thin air, even by the smartest protocols. They will always require a team / company to support the ecosystem with the services that allow marketplaces to live and grow.

As a result, designing businesses with “minimal viable decentralization” may be a more efficient way of launching of products and approaching early governance⁶ although its likely to be viewed less favorably by those that prioritize decentralization above all else.

DeFi business models are not constrained to the above mentioned models. For example, Arwen is planning to monetize via a revenue sharing agreement with centralized exchanges for the trades referred by Arwen⁵ (further details will be provided in Trade-Offs: Decentralized Exchange . )

6.) New Risks Compounding Across Protocols

Cryptocurrencies and blockchain-based markets have fundamentally different characteristics than their traditional counterparts. DeFi protocols benefit from composibility which leads to faster innovation, but also results in higher levels of interdependancy. Therefore, it’s fair to assume that the risk profile of these products, especially in combination, is not yet fully understood. While each project claims to have developed its own robust risk models, the complexity of analyzing these new risks across interdependent protocols is non-trivial . It’s also worth noting that most risk models weren’t very useful in 2008. In some cases, these models failed because just one assumption was flawed.

Many of these projects utilize concepts that contributed to the 2008 financial crisis, but more importantly, they utilize them in new and untested ways. For example, the rehypothecation of collateral , fractional ownership of structured products, and pooling of risk were all elements of the 2008 financial crisis. DeFi takes these concepts and applies them to highly volatile and hard to value assets in relatively illiquid markets with insufficient safeguards. The combination of all of these factors, combined with the complexity of creating a cohesive view of collateral rehypothecated across protocols, creates an entirely new risk profile for which there is little precedent. Superfluid collateral? Let’s not do this.

In this context, it’s worth considering what a market failure would look like. MakerDAO is an experimental network upon which the success of many other projects depends. It is important to remember that in the case of market failure, CDPs and other DeFi products are not insured and a third-party is not likely to step in to recapitalize small cap crypto start-ups (moral hazard debate aside.) Investors, users, and token holders will be responsible for recapitalizing this highly interdependent DeFi system.

The systemic impact would likely be jarring. Dai is commonly traded on 0x Relayers as a pair with ETH. It is often deposited on Compound and then lent out again to hedge funds to be used for risk-on trades. dYdX is also dependent upon MakerDAO since its short ETH token is long Dai. dYdX further depends on Dharma (which lends in Dai) and the 0x protocol (which facilitates the trading of Dai) to access liquidity. The cascading effect of a failure of any one of these protocols would likely cause a systemic unwind that is rapid (due to volatility of underlying, rehypothecation, and automated execution of smart contracts), jarring (these markets are not as liquid as traditional markets), and significant. The rewards may be high, but the risk is at least commensurate.

Cryptoeconomics don’t defy the principals of regular economics and cryptofinance (DeFi) can’t escape the classic risk/reward constraint of regular finance. Superfluid collateral? Let’s not do this.

7.) Regulation

The primary concern regarding regulation of the industry doesn’t seem to be that current regulations are too restrictive, but rather the concern is related to the ambiguity as to how existing regulations will be applied in regards to blockchain-based networks and cryptocurrencies. Many of the startups in the space don’t know how to determine whether they should launch or not because the regulatory environment they are operating in is so unclear. The cost of all this uncertainty is high.

In fact, the regulatory burden is so high that some start- ups have determined that they don’t have enough capital to launch in a fully compliant manner, made all the more difficult by the recent crypto winter fundraising environment.

From Dream to Reality

Projects in the space are acutely aware of the above cited problems and are actively working to overcome them. For example, Dharma now pays gas fees for users to make the user experience easier while MakerDAO is planning to abstract the process of converting Dai to MKR upon redemption for MCD. Networks across the board are increasingly removing the requirement to hold native tokens, choosing to replace them with more widely used cryptocurrencies such as Dai or ETH. This is an important step since it removes what has historically been a hurdle to DeFi adoption.

New innovations and projects also enter the space frequently and could provide solutions to many of the weaknesses of current projects. For example, the Arwen protocol is ideally suited to HFT since only the opening and closing of escrow gets posted on-chain, while the rest of the protocol operates like a Layer 2 channel, allowing trading to happen much more quickly.⁵ The trade is also only visible to the trading party and the exchange. Since there is no on-chain execution of individual trades, miners do not see the transactions ahead of time, eliminating most of the front running issues that DEXs face.⁵ The protocol could also solve other critical painpoints, such as allowing for fractionally collateralized options (which dYdX is actively working towards), trustless fiat conversion (via integration with banks) and quicker movement of tokens across exchanges, allowing for effective arbitrage.⁵

In 2017–2018, the main focus in the DeFi industry was on building DEX scalability and liquidity. Solutions for each are quickly evolving. The focus in 2018–2019 has shifted to lending. The composability of DeFi products should allow the whole ecosystem to move forward at a faster pace and may allow for more mainstream adoption throughout 2019, as users are lured into the ecosystem via multiple protocols (trading, lending, derivatives) that all require users to transact with cryptocurrency.

Many of these dynamics could spur more usage and adoption in the next eighteen months, creating the flywheel effect common in two-sided markets. The key point is to acknowledge that while the challenges are persistent and numerous, the industry is far from stagnant, and real and rapid progress is being made. However, risks need to be seriously assessed and considered so that this progress can be sustained.


All interviews personally conducted are unattributed

  6. Idea credited to Brendan Forster as articulated in a panel at New York Blockchain Week
1 Like