[Binance Research] DeFi Series #1 - Decentralized Cryptoasset Lending & Borrowing
摘要： Studying the landscape of the (Ethereum) decentralized cryptoasset lending industry.
- Decentralized Finance (DeFi) has grown to become one of the main applications of
Ethereum. While the Open Finance initiative is, by nature, platform-agnostic, DeFi
applications are almost exclusively being developed and used on Ethereum. One of
DeFi’s cornerstones is Dai, a decentralized cryptoasset-backed stablecoin minted in the
- Lending/borrowing protocols & platforms offer different incentives for market
- For borrowers: ability to short an asset or borrow utility (e.g. governance rights).
- For lenders: ability to insert capital to use and earn interest.
- For both: arbitrage opportunities across platforms.
- Decentralized, non-custodial protocols offer several promising advantages over
traditional financial products such as:
- Transparency and price efficiency as prices are subject to market demand
- Ease of access and speed when borrowing/lending capital
- Censorship resistance and immutability
- However, given the experimental nature of these financial primitives, they do exhibit
specific disadvantages relative to their custodial counterparts including:
- Technology risk (smart contract risk replaces counterparty risk)
- Low liquidity (limits to borrowing and lending at current interest rate without
materially affecting the equilibrium interest rates)
- Overall, the crypto-powered lending space is still nascent, but offers a compelling
value proposition whereby individuals and institutions can move past outdated credit
models to provide broader access to capital without the need for a trusted
intermediary. While it may be difficult to determine which platforms and protocols will
garner the most usage long-term, the various tools together provide a growing
assortment of useful decentralized levers for access to capital, together forming an
evermore comprehensive toolbox of DeFi.
Decentralized Finance, also referred to as "Open Finance" or "DeFi", has grown to become one of the core drivers of usage on the Ethereum network. At the core of DeFi’s principles is the provision of a brand new, permissionless financial service ecosystem without any central authority, available to everyone in the world. In this ecosystem, the user acts as his own custodian, maintaining full control over his assets, encouraging full ownership and access to all disintermediated marketplaces and platforms.
This report is the first article in our DeFi Series that will cover different platforms and protocols that aim at disrupting the existing financial industry and infrastructure. Specifically, this report will discuss the foundational cornerstones of DeFi: decentralized cryptoasset lending and borrowing platforms.
As of May 31st 2019, Ethereum dominates in amount of existing applications, transaction count, and volume traded/locked on existing platforms (e.g. Maker, Compound). As a result, Ethereum-based dApps will be the focal point of this report.
As loans have historically been built on a single pillar of trust between two parties, how do decentralized cryptoasset lending and borrowing platforms match loans between anonymous individuals with no predefined trust or relationship?
1. Lending & borrowing decentralized landscape
1.1 What is DeFi? What blockchains are included?
“An ecosystem comprised of applications built on decentralized networks,permissionless blockchains, and peer-to-peer protocols for the facilitationof lending/borrowing or trading with financial instruments.”
Today, a large majority of Defi protocols are being built on Ethereum. Collateral locked on Ethereum-based DeFi applications are collectively worth over USD 500 million (more than 1.5 millions of ether), as of June 5th.
Bitcoin's Lightning Network solution can be considered as a DeFi application running on the
Bitcoin blockchain. As of June 5th 2019, the Lightning Network has its value locked worth
around USD 8 million.
Regarding EOS, EOSRex enables users to borrow and lend EOSIO resources (e.g. CPU and
Bandwidth/NET). Despite it being released a month ago, this single application has become the largest platform based on the total collateral value, with more than 90 million EOS being locked. Other platforms for EOS include a permissionless money-market protocol named BUCK Protocol that went live recently. Furthermore, other blockchains will also build decentralized applications, part of the Open Finance initiative. For instance, Stellar aims at developing the world’s new financial system, but has yet to see increasing financial-related dApps created by third-party developers.
Another example is Ontology which recently partnered with Paxos, the stable coin issuer, to
release up to 100 million PAX tokens on its blockchain. In its press release, Ontology mentioned “to explore application scenarios on Defi [...] beyond exchanges and are expected tolaunch the PAX token on Ontology in May”.
Regardless, this report will focus on Ethereum-based DeFi applications. What are the assets
supported on these platforms and protocols?
> What assets are supported on Ethereum-based DeFi platforms?
A wide range of assets are being supported on Ethereum-based DeFi platforms. They can be
classified in two main categories: native blockchain assets and non-native blockassets.
“Native blockchains assets are defined as assets whose value is not backed
by any non-native blockchain asset such as a commodity, equity or fiat
currency. Examples include Basic Attention Token (BAT) or OmiseGo (OMG).
These assets typically allow minting/burning of new assets on the blockchain (i.e on-chain
creation of asset). Across the DeFi ecosystem, one of the core pillars (and conversely an
example of a native blockchain asset) is MakerDao (MKR) and its Dai (DAI) stablecoin.
As a crypto-backed stablecoin, Dai’s value is pegged to the USD dollar through a complex
collateralization mechanism and a stability fee that incentivizes market participants to
arbitrage any long-lasting mispricing issues. This will be discussed further in the next section.
Other native assets are also used as a component of these applications, such as ERC-20 tokens
on the Ethereum blockchain like Augur (REP), Basic Attention Token (BAT) or 0x (ZRP).
These native blockchain assets can be used as:
- Collateral for a loan
- Lendable assets
- Borrowable assets
- Governance assets (e.g Maker/MKR)
Interestingly, illiquid (native) blockchain assets cannot be used as collateral regardless of whether or not they are fungible or non fungible (e.g. CryptoKitties).
“Non-native blockchain assets include different assets that run on a public blockchain but whose value is collateralized by an other asset that is non-native to the blockchain, such as a commodity, equity or fiat currency.”
Some of the most popular non-native blockchain assets are stablecoins collateralized by fiat currencies on bank accounts such as USD Coin (USDC) or TrueUSD (TUSD).
Within the DeFi ecosystem, USDC4 was included into many lending and borrowing protocols. Notably, Dharma added USDC to its supported assets, following an investment from Coinbase in February 20195.
What are the protocols and platforms running on the Ethereum blockchain for non-custodial lending and borrowing platforms?
1.2 Protocols and platforms
In this section, only non-custodial platforms and protocols were included. As a consequence, it excludes platforms such as BlockFi or Nexo.
It is worth noting that other platforms have failed or seem to no longer be supported anymore, as illustrated by CDx.
1.3 How does the lending process work on non-custodial platforms?
In this section, the three largest platforms, mentioned in the previous section, will be discussed: Maker, Compound and Dharma. Interestingly they all have different models and as a result, offer interesting insights about how the general lending process can exist on non-custodial platforms.
In general, they rely on overcollateralization from borrowers as the key mechanism to initiate a loan and strict rules to decide whether the loan must be closed during the duration of the loan.
Maker is a unique “issuer”, in which an individual can borrow Dai, a stablecoin whose peg targets a value of 1 USD per token, directly by placing ETH in reserve in their collateralized debt position (CDP).
The Maker token (MKR) allows for individuals to participate in the operational earnings through “governance fees” that act as interest rates for the network. This structure is different from the peer-to-peer model, as it pools reserves together to actively issue coins, rather than transfer existing coins.
Unlike fiat-collateralized tokens, cryptoasset collateralized stablecoins rely on different mechanisms to maintain their peg against a fiat currency. Yet, it is worth mentioning that any stablecoin, regardless of its peg mechanism, has its intrinsic value fully dependent on the centralized decision-making process of central banks (e.g US Federal Reserve), which are in charge of monetary policy that ultimately impacts the relative value of the fiat currency against other currencies and assets.
In MakerDAO’s example, Dai a cryptoasset collateralized stablecoin that works with the Maker ecosystem, built on Ethereum. Using smart contracts written in Solidity, the Maker ecosystem currently allows Ethereum holders to park their Ethereum in a contract, in exchange for the ability to print or mint the DAI stablecoin, thus taking out a stablecoin loan against their volatile underlying collateral of ETH. It is worth noting that the Maker platform will soon allow borrowers to park multi-asset collateral in order to mitigate the volatility of a single asset, thus introducing diversification at the collateral level. This, along with a rebound in overall market prices, could further push the circulating supply of the world’s largest crypto-backed stablecoins to new highs.
In Maker, loans are overcollateralized with the collateralization ratio being required, at the loan inception, to be above 150%.
Meanwhile, peer-to-peer pools can be separated into two types:
- Direct matchmaking (e.g. Dharma, Ethlend, etc.)
- Liquidity pools (e.g. Compound, dydx)
We will have a look at the two largest peer-to-peer platforms: Compound and Dharma.
In a majority of these platforms, the platforms are matching lenders and borrowers directly, with a spread between the interest rates, so that the protocol creator or platform creator, can sustain operations.
Compound is a protocol which creates money markets for various tokens, running on the Ethereum blockchain.
Each market is linked to a cToken (i.e. cBAT) that acts as the intermediary for any asset being lent on the protocol. Through the cToken, lenders earn interest that compounds over time. Specifically, interest rates are compounded at the block level. With a block being created every 15 seconds, the cTokens would increase continuously over the duration of the loan.
Within Compound, there is a withdraw function which allows users to convert the cTokens to the original assets (e.g. from cBAT to BAT).
Interest rates exist for each asset based on real-time market dynamics. When there is an excess of demand from borrowers, the interest rate would increase whereas an excess of lendable amount would lead to lower interest rates. Furthermore, the supply rate (i.e the lending rate) is always lower than the borrowing rate, by design, to create liquidity on the platform.
Dharma is a platform that allows users to borrow and lend several assets at a fixed interest rate for 90 days. Supported assets include Ethereum, USD Coin and Dai.
In short, this platform handles and matches trades manually without acting as custodian at any single point of time. A user can request to lend funds and then he/she will need to wait for their offer to be matched.
Interest rates are currently determined manually by the team in a black-box process. Interestingly, borrowing and lending rates are set equal which contrasts sharply with other platforms like Compound.
If a borrower decides to repay his loan before the maturity date, he must pay the entire interest on the loan over 90-day. As a result, the only incentive for a borrower (to repay the loan early) is to get full access to his collateral immediately.
1.4 Some additional metrics
2. Market participants
In this section, we will discuss the perspectives of DeFi market participants, including what type of individuals/institutions may want to participate in these protocols, and their potential motivations to do so, such as arbitrage.
2.1 Lender case
From a lender perspective, the key benefit is the ability to put existing capital to use and generate yields versus simply holding their assets (“basic HODL strategy”) in the long-term. The two key yields that users may see in this case are:
- “Rewards” for long-term investments. Long-term position holders (aka value investors) can be rewarded through these interest rates (typically denominated in the same currency) and as a result, it leads to a new stream of income for them (besides capital appreciation or a “HODL strategy”). For instance, an investor who is long on Ethereum with a long-term investment horizon may benefit from these lending platforms and earn an additional interest rate.
- USD interest rate on stablecoin. Unlike most regular bank accounts at a US bank, stablecoins do not generate any interest rate. Storing USD-denominated stablecoins on a digital wallet is similar to keeping USD notes in a leather wallet: none of them return interests. This translates into a depreciation of the effective value of the stablecoin owing to inflation risks. For stablecoin issuers competing for market share, there could be an incentive to distribute part of their revenue from floating interest rates collected on their bank deposits12 to users that support the circulating supply through these decentralized platforms.
2.2 Borrower case
From a borrower perspective, there are several reasons why borrowing an asset through a decentralized protocol could be useful.
- Ability to short an asset. A borrower can short an asset by borrowing it and immediately selling it on any exchange for another coin, typically a stablecoin. This serves a similar function to margin trading on a centralized exchange. Similarly, it also provides an ability to enter into margin trading on platforms that don’t support this functionality. For instance, a trader is able to short an asset on Coinbase (as it doesn’t allow margin trading). As the list of borrowable assets expands over time, this ability will also become increasingly valuable as well.
- Borrow utility for a token (e.g. borrow governance rights). The borrower can decide to (temporarily) borrow an asset in order to achieve more power or governance rights on the blockchain. For example, someone could borrow REP (Augur token) in order to participate in resolution mechanisms on-chain13.
2.3 Arbitrage opportunities
Assuming no transaction cost, an arbitrage opportunity exists if the following inequality is true:
borrowingRate (%) < lendingRate (%)
However, because all loans are overcollateralized in today’s existing protocols, the spread between borrowing and lending rates will never actually be zero. A more thorough equation can be written to account for additional fees such as transaction costs or gas fees:
lendingRate (%) -minOvercollateralizationRate (%) * borrowingRate (%) - aggregateFees (%) >0
with aggregateFees (%) >0 being the sum of any additional fees such as gas fee or any transaction fee on the lending/borrowing platforms.
The product of the minOvercollateralizationRate (%) and the borrowing rate assumes a 100% utilization rate of all the possible capital that can be lent. However, in practice, the network in aggregate tends to be overcollateralized at a much higher rate than the minimum, providing buffer room if the later leading to a less than 100% utilization rate.
3. Risks and benefits of decentralized protocols
3.1 In this section, we compare the risks and advantages of these decentralized protocols compared to similar options and platforms from the traditional financial industry and with centralized cryptoasset custodial platforms.
3.1. Advantages and benefits of these platforms & protocols
> Benefits for the cryptoasset industry
- Improved price efficiency: as lending and borrowing protocols allow anyone to take a short position against the asset, it results in greater price discovery and efficiency in the market.
- Immutability and censorship resistance: a single transaction cannot be reversed and no third party can stop the borrowing process. This results in efficient markets that are censorship-free without discrimination for users.
> Advantages for decentralized systems versus centralized banking platforms
- Greater capital access: participants residing in capital-controlled countries can obtain access to unseizable stablecoins that give them exposure to other currencies, such as the US dollar or any fiat-denominated stablecoin (e.g. GBP, EUR).
- Transparency and efficiency: on pure P2P lending and borrowing platforms, interest rates are determined solely by market participants and loans are secured through overcollateralization. It is easy and costless to access information publicly about loans.
- Flexibility and loan segregation: using one wallet to borrow multiple assets simultaneously. Different risks can be considered independently. For instance, it is possible to borrow ETH with USDC with one wallet while another wallet is borrowing BAT with ETH. If one loan defaults (i.e. the collateral value gets below its liquidation threshold), it only impacts the locked collateral as the risk exposure is segregated. However for platforms allowing multi-collateral support, it is possible to mitigate the risk exposure by aggregating different assets as a basket of collateral.
- Lower set up costs/turnaround time: unlike the traditional financial industry, it is extremely fast for any user to borrow funds at market rates, removing any intermediary crediting agencies.
> Advantages for decentralized systems versus custodial lending platforms (e.g. centralized exchanges)
Custodial lending platforms refer to centralized platforms in the cryptoasset industry such as custodial lending platforms (e.g. BlockFi) or centralized exchanges (e.g Bitfinex, Gate.io) allowing margin lending & borrowing on their exchanges.
- Ability to transfer borrowed capital across platforms and trading venues: unlike centralized lending trading venues, an individual can borrow capital and transfer it to multiple venues freely as only his collateral would be locked in a smart contract.
- Full custody of the funds: someone wishing to enter margin trading can effectively short an asset on a decentralized exchange without giving away his tokens.
- Ability to enter into margin trading in restricted jurisdictions. On fully decentralized lending protocols, the whole process occurs on-chain without KYC. As a result, any market participant -regardless of its home jurisdiction- can get exposure to an asset (either long or short) and, subsequently, enters into margin trading.
3.2 Disadvantages and inherent risks
> General disadvantages
- Lack of insurance mechanism: unlike the traditional financial industry, there is no insurance mechanism on a loan. As a result, default risk exists and while some platforms may insure users in case of default of a counterparty, this remains a general flaw in these decentralized platforms.
- Difficult to redeem to fiat currencies: as loans are denominated in cryptoassets, it is generally hard to redeem these loans for fiat currencies. As a result, it appears difficult to make use of these borrowed funds in the real economy. However this issue is broader and related to the general use of blockchain assets in the real economy. Furthermore the support of collateralized stablecoins (particularly USD Coin) may further incentivize users to borrow funds based on native blockchain assets (such as Ether) and to redeem for fiat currencies .
- Overcollateralization does not help the unbanked: as there is currently no credit scoring available, loans must be overcollateralized14, by design, which reduces the opportunity to obtain access to capital that the user does not own. As a result, these lending and borrowing platforms do not support the growth of an economy as one of the pre-conditions to contract a loan is to be overcollateralized (sometimes by as high as 150%). Alternative systems relying on credit scoring are being built but it appears unclear how a system can be designed without clear KYC policies which may further discriminate some of the market participants15.
- Overcollateralization prevents leveraged trading: while it is possible to borrow and transfer these funds onto other platforms, it remains less interesting for traders to borrow on decentralized platforms and protocols funds as traders need to be overcollateralized. Centralized exchanges make margin trading easier as they rely on automatic deleveraging system, sophisticated liquidation algorithms and insurance funds that allow traders to build highly leveraged positions.
Yet, if these disadvantages are important, most of them are explained by the nascency of the Open Finance industry. These applications and protocols remain in an experiment stage and are expected to mature, solving some of the key issues with the creation of insurance protocols, further gateway with the fiat world and new smart contract mechanisms to avoid the need for overcollateralization by design.
> Blockchain-specific risks
- Network congestion: Ethereum has had a few clogging issues on its blockchain owing to different applications such as CryptoKitties16 or a Sybil Attack owing to FCoin’s on-chain vote-driven listing policy17. If the network gets congested, transaction may remain in a pending state, which ultimately results in market inefficiency and information delays.
- Transaction costs such as on-chain gas fees: as transactions are competing based on gas fees, transactions with lower gas fees may be left pending at lower priorities. Conversely, there are sometimes no definitive or locked gas fees for some products, leading to potential higher costs for the user.
Even if these issues are related to Ethereum specifically, similar issues can exist on any blockchain. Specifically, Ethereum sometimes faces these network performance issues owing to its popularity and usage. On the contrary, most of the other existing blockchains do not currently face scalability issues simply because they do not have enough traffic or are much more centralized by design, allowing for higher speeds and better performance.
> Platform-specific risks
>> Semi-centralized platform risks
- Lack of full transparency. For instance, Dharma is the third largest DeFi application but still lacks some transparency about specific aspects. It is unclear how the fixed interest rate for each asset is defined as it is not subject to market environment (supply and offer dynamics). Furthermore, the loan matching process is not transparent. There is no real-time information about how many loans are pending and no way to audit whether loans are matched based on fair principles such as “FIFO”.
- Interest rate governance.: centralized platform can change the interest rate at any time which lead to uncertainty for lenders and borrowers.
- Counterparty risk: if the platform is semi centralized (which funds transiting through an intermediary vs being directly sent to a smart contract), the platform itself has control over the assets and may use the funds maliciously.
>> Fully decentralized platform risks
- Failure in price mechanism: some Oracles, aka data providers to value collaterals, are subject to operational risks, such as data manipulation or failure to retrieve the correct market data owing to price source issues. This could potentially lead to issues on how the collaterals are valued.
- Volatility risk: if the price fluctuates too fast, some users may have their collateral liquidated and the loan refunded.
- Smart contract flaw: as transactions cannot be reversed on the blockchain, any failure in the smart contract, leading to a hack, could lead to massive losses for users of a decentralized protocol. As Ethereum is being developed and continues to evolve at a fast pace, platforms need to constantly remain up-to-date with development and potential migration of smart contracts to prevent fund losses.
- Timing flaw: implied interest rates can only be calculated at the block level when the platform’s block time is fairly consistent. Otherwise, the interest rate may differ if the blockchain is stalled or slowed (or conversely accelerated), and the logic of the smart contract to calculate interest is tied to a rate per block.
> Regulatory and taxation risks
- Security classification: there is growing uncertainty about whether stablecoins could be classified as a security.19 The SEC may consider these as “demand notes” whereas the CFCT could consider these as “swaps”.
- Custody: non-custodial platforms brings uncertainty about what is the actual custodian of the assets lent. Many professional investors are requested to rely on third-party custodians and the use of these platforms could create potential regulatory loopholes where some regulatory requirements may not be fulfilled. Interestingly, platforms like Compound have introduced the cTokens that grow in balance directly in account so you can earn interest from your cold wallet.
- Licences: decentralized platforms operate without licence in most jurisdictions, regardless of where the end-user is based . In the medium term, the recent support of USD Coin by many platforms may create potential “licence risks” which would ultimately affect costs for the operators, as a result, would lead to either higher interest rate spread or increased transaction fees.
- Taxation uncertainty: there is currently a gray area for these platforms about the location and what local jurisdiction these platforms may be requested to follow.
Ethereum is the largest programmable blockchain by market capitalization. As a result, it is no surprise that it is currently the standard default platform for many decentralized applications, which include lending and borrowing protocols such as Compound, Dharma and others. Soon, other programmable blockchains (such as EOS) are likely going to be increasingly used for many other decentralized lending protocols.
Owing to its nascency, the lending and borrowing industry carries a few risks that are expected to be mitigated as the industry matures (notably with increasing inflows and higher volumes). More products in a more mature industry should give more options for participants to select, giving users a more complete access to financial services.
Furthermore, these decentralized financial platforms may be a foundation or data point for centralized institutions to make their access decisions, hopefully leading to greater access to financial services through increased competition for traditional financial players, the gatekeepers of the financial world today.
While Quorum, JP Morgan’s permissionless fork of Ethereum, is being developed with initiatives such as JPM Coin, Ethereum remains the pacesetter of all blockchains for decentralized applications that will unbank many individuals.
DeFi, as illustrated by these lending and borrowing platforms, appears as one of the best use-cases of blockchain technology which could reach up to billions of users across the globe and allow access to basic financial services at efficient rates.
Circle Research (Ria Bhutoria). A closer look - Dharma-larm (2019). https://medium.com/circle-research/a-closer-look-dharma-larm-8088e51c0886
Delphi Digital. Decentralized Finance (DeFi) - Thematic Insights (2019). https://www.delphidigital.io/defi
Kyle J Kistner. How Decentralized is DeFi? A Framework for Classifying Lending Protocols (2019). https://hackernoon.com/how-decentralized-is-defi-a-framework-for-classifying-lending-protocols-90981f2c007f
Macro Narratives in Blockchain (Andrew Wong). The next FinTech: Global “Open Finance” Infrastructure (2019). https://medium.com/macro-narratives-in-blockchain/the-next-fintech-global-open-finance-infrastructure-90ac093a411b
Maker Team. The Dai Stablecoin System (2017). https://makerdao.com/whitepaper/DaiDec17WP.pdf
Multicoin Capital. Investment Thesis (2019). https://multicoin.capital/2019/04/24/multicoin-investment-thesis/
The Block (Ryan Todd). Block by Block: Crypto Lending (2019). https://www.theblockcrypto.com/2019/01/18/block-by-block-crypto-lending/
The Block (Contributor Network). The benefits of trustless lending (2019). https://www.theblockcrypto.com/2019/01/31/the-benefits-of-trustless-lending/
Wave Financial (Roy Learner). Crypto lending: too good to be true? (2019) https://medium.com/wave-financial/crypto-lending-too-good-to-be-true-fc010e7fc86c