- The Ethereum DeFi ecosystem has demonstrated strong growth, though highly concentrated in the lending and crypto-collateralized stablecoin sectors.
- While many cite Total Value Locked as evidence that all of DeFi is growing, a closer observation reveals that this metric is skewed by the importance of collateral and staking by design in the most popular protocols including MakerDAO, Uniswap, and Compound.
- Considering alternative growth measures such as trading volume or interest, DEXs and prediction markets have struggled to gain adoption. Measured by the value of loans originated, Compound is rapidly growing.
One of the major industry developments in the past year is the emergence of ‘Decentralized Finance’, or DeFi, applications that seek to offer alternatives to various traditional financial and banking services, such as lending, through a disintermediated, open protocol. The vast majority of DeFi projects are built on Ethereum smart contracts and use ETH, primarily as collateral, though the sector could easily be imagined to expand to other blockchains. Despite the sector being frequently represented in industry discussion as an uncomplicated story of growth, closer examination reveals significant divergence in traction across classes of application. This article highlights metrics to track DeFi’s growth across major projects and sectors, and offers brief commentary on particular issues facing projects.
What is 'DeFi'?
Semantically, ‘DeFi’ is a rather nebulous concept, alternatively called ‘Open Finance’ in some contexts. Most uses of blockchain technology are ‘DeFi’ in some way, as they generally involve some sort of financial activity arbitrated by a decentralized database. One could argue for an expansive definition that includes all cryptocurrencies, which function as general purpose monetary systems and operate in a decentralized fashion. Another edge case is fiat-collateralized stablecoins, such as USDT and USDC; while these projects utilize smart contracts to some degree, their reliance on the central issuing party for collateralization limits their functional degree of decentralization or censorship resistance. As such, industry discussions of DeFi typically use the term more narrowly, referring to smart contract platform applications that offer direct alternatives to services provided by traditional financial firms. This category primarily comprises of finance-focused Ethereum dApps with significant on-chain logic. This discussion follows this convention.
Measuring the Growth of DeFi
Across a range of metrics, the DeFi space has seen considerable growth over the past two years. The most commonly cited metric for tracking this growth is Total Value Locked (TVL), which is the equivalent ETH value for all tokens staked, most commonly as collateral, in various smart contracts related to DeFi applications. One argument for TVL as a valid growth metric is that since locking assets in a DeFi protocol necessarily has some opportunity cost and/or risk, it is less able to be cheaply manipulated than a metric such as Daily Active Users (DAUs) or transaction counts, which has occurred on no-fee smart contract platforms such as EOS. The majority of this amount is staked through ETH itself, but also includes the ETH-equivalent value of DAI or USDC loans, for example.
Source: DeFi Pulse
The vast majority of TVL is through MakerDAO’s Collateralized Debt Position (CDP) system, which currently requires ETH as collateral for the DAI stablecoin. Additionally, Compound has gained significant traction with ETH and other Ethereum-based assets ‘locked’ in loans. Presently, over 2% of the outstanding ETH supply is locked in DeFi.
However, TVL is just one metric to measure the ecosystem, and the measure has its shortcomings. TVL overweights applications where ETH is used specifically as collateral, which, given the broad design space for DeFi—roughly, financial services that can be specified as autonomous, secure code—may not accurately measure all facets of its growth. To some extent, the focus on and growth in this metric is a product of the very designs of these projects, particularly MakerDAO, which function as a direct result of high collateralization requirements. While growth in Total Value Locked is commonly cited as evidence of the broad growth of the entire DeFi sector, a closer examination of metrics within individual sectors of DeFi reveals a more nuanced evolution of the space.
A rapidly growing category with DeFi is crypto-collateralized stablecoins, led by MakerDAO. MakerDAO is a platform for crypto-collateralized loans and stablecoins, in which users create Collateralized Debt Positions (CDPs) by staking ETH and receive the USD-pegged DAI stablecoin. This system can be used as a decentralized margin trading platform, for example offering users a method to leverage a short or long trade on ETH. All CDPs are required to maintain a collateralization ratio of at least 150% issued DAI value staked in ETH, and CDP holders pay a ‘stability fee’—effectively an interest rate—for each loan, payable in MKR.
The DAI stablecoin has maintained a relatively stable value of within 5% of the $1 peg. While this degree of volatility certainly is not ideal or even the lowest within the stablecoin space, with Tether staying within 2% of the peg over the past year, it arguably does represent a general state of success for DAI in being a ‘stablecoin’.
This relative stability has occurred in conjunction with a volatile and generally declining ETH price, requiring CDP holders to continually fund their contracts with ETH to maintain a collateralization ratio of at least 150%. A notable upcoming change is the release of Multi-Collateral DAI, allowing a broader range of assets beyond just ETH, including those pegged to physical assets, to backstop the price. This feature is intended to lend more stability to the system, as collateral price fluctuations may be distributed across many assets, thus reducing ETH’s importance and potentially reducing the high collateral ratios necessary in the present single-collateral system.
The total supply of DAI has risen steadily to roughly 80 million, though growth has leveled off over the past few months. The stability fee, a network parameter effectively representing an annual interest rate that CDP holders must pay, increased to a peak of nearly 20% in mid-2019. Determined by an MKR holder vote, this stability fee increase is a response to the increased demand for new CDPs, and represents a key mechanism through which the system maintains a DAI price peg.
Smart contract-based lending protocols, such as Compound and dYdX, exemplify another application category that has grown significantly. To some extent, MakerDAO also falls in this category. These lending protocols use a variety of market designs to match borrowers and lenders. One useful metric to track growth is the nominal value of loans originated. The category has seen strong growth, driven primarily by Compound, with nearly all loans in DAI, ETH, or USDC.
Currently, the largest such project, Compound, offers pooled lending services through money market funds with algorithmically determined interest rates. Users can borrow and lend various Ethereum-based assets, presently DAI, ETH, USDC, REP, ZRX, WBTC, and BAT, through Compound’s non-custodial protocol. Compound’s approach contrasts with other peer-to-peer lending protocols that function primarily as matching engines between individual borrowers and lenders. Rather than matching, collateral is pooled for each asset’s money market and interest rates fluctuate for all users with an outstanding loan. All Compound loans accrue interest dynamically based on the current rate and do not have a fixed length. Compound’s strategy of providing pooled liquidity through money market funds and algorithmic variable interest markets is duplicated by other projects, such as dYdX, an alternative lending platform that also offers margin trading using the 0x protocol.
In contrast to Compound and dYdX, the original version of Dharma facilitated lending by matching individual borrowers P2P with fixed term loans. After significant early growth, Dharma pivoted to become an interface for stablecoin savings accounts, actually using Compound’s pooled lending liquidity as a backend protocol through which savings accounts are routed. While the sector is still early in development and alternative designs may emerge, this variation in adoption suggests that P2P loan matching faces significant challenges moving forward.
Despite the growth in the above metrics, one could argue that DeFi lending approaches are rather capital inefficient. In a traditional lending system, loan originators use some form of credit scoring to assess an individual or organization’s credit worthiness, examining their financial history and record of repayment. Blockchain-based applications presently lack a robust identity system to track this history; a user could default on a loan linked to one pseudonymous address, which may not be linked to alternative addresses and loans. Lacking a credit scoring system, the prevailing alternative is to require users to overcollateralize all loans and crypto-backed stablecoins. This is shown in Figure 5, Average Collateralization.
Source: Loanscan Note: Compound v1 had a collateralization ratio of over 6000% in late 2018, declining to 450% before v2 was released; this outlier is excluded in the interest of chart scale.
This overcollateralization requirement represents a significant hurdle for mainstream adoption of such lending protocols, since the resulting interest rates are generally higher than consumer loan alternatives from the traditional financial system. While this prevalence of overcollateralization is perhaps unsurprising given the lack of alternative methods, it does represent a key area in which future DeFi projects may innovate and attract new users. Assuming stablecoin stability and loan default rates stay the same, users would prefer to use systems with a lower cost of capital and subsequent lower interest rates.
In contrast, other sectors of DeFi do not have the same growth, particularly when tracked by metrics other than TVL. For example, DEX traction may be tracked by following changes in traded volume, shown in the following figures. DEXs as an alternative to centralized exchanges are expected to offer greater audibility and decreased custody risk. Despite these potential benefits, adoption to date is limited, with average monthly volume across all DEXs under $500 million over the past two years. By comparison, the leading centralized exchange, Binance, regularly processes over $1 billion in daily trading volume. For further discussion of prospects and challenges facing DEXs, see Smith + Crown’s 0x Deep Dive Report.
However, DEX traction is not uniform across projects. One DEX design that has attracted considerable attention is Uniswap, which uses an automated market maker (AMM) design similar to Bancor to pool liquidity for token pairs within a smart contract. In contrast to 0x and most other DEXs, Uniswap does not maintain an order book, and instead users effectively trade against a smart contract that holds a balance of both tokens. By supplying tokens to individual liquidity pools, users can earn trading fees. The result of such a design is that liquidity is always available at some price against the smart contract counterparty, though large trades may incur significant slippage.
Uniswap’s growth may be explained by a variety of compelling characteristics. Since all liquidity for a trading pair is pooled, there are no large bid/ask spreads across individual order books, thus offering a reduction in slippage for small traders. Relative to other AMM designs such as Bancor, Uniswap’s lack of a native token or listing approval process allows developers and traders to trade on a permissionless protocol using existing and more liquid assets, bypassing the friction of a native token.
Figure 7 shows growth in tokens staked in Uniswap liquidity pools in the first half of 2019.
Source: DeFi Pulse
The case of Uniswap highlights how the uncontextualized use of TVL as metric can mislead observers as to actual sector or project growth. While DEXs, including Uniswap, have seen relatively low trading volumes, Uniswap’s growth in TVL is presented elsewhere as evidence that the entire sector is rapidly growing. Much like overcollateralization is a direct product of MakerDAO’s design, locking tokens is a result of Uniswap’s pooled liquidity design, such that significant growth in Uniswap’s TVL is observed even when its trading volume is still quite minimal. By examining a variety of metrics in context, it becomes apparent that DEXs as a sector remain struggling for adoption amidst a variety of challenges, though Uniswap’s particular design emerges as a relative bright spot with some demonstrable traction. The case thus illustrates why TVL as a metric should be contextualized to the DeFi sector and particular applications.
Finally, prediction markets, an early and much-heralded application of blockchain technology, have thus far struggled to obtain significant volume. Most prominently, Augur is a decentralized oracle and prediction market platform that acts as a framework for the creation, trading, reporting, and settlement of individual, user-generated prediction markets. Augur launched in mid-2018 after multiple years of development. Despite prominence within the industry and a complex oracle design that has thus far correctly arbitrated all of over 2000 finalized markets, Augur’s trading volume has averaged less than 1000 ETH per day with open interest for all markets averaging under $1 million. Further, interfaces built on Augur such as Veil have quickly shut down after struggling to gain initial user traction.
Another member of the prediction market space, Gnosis has also struggled to gain widespread adoption, following a mid-2017 token sale that valued the network token supply at $300 million Though a codebase for the Gnosis protocol has been open sourced and is operational, various related prediction markets have seen minimal volume. A new platform, Sight, is currently under development.
There are a number of potential explanations for this lack of growth within prediction markets, including fragmented market liquidity, a difficult user onboarding experience, and regulatory uncertainty. The issue of fragmented liquidity may be of particular importance; while there are a large number of open markets on a wide range of topics, platforms such as Augur lack a strong incentive for market creators to seed the initial market order book. This prevents potential users from taking large positions without significant slippage. A second open issue for prediction market sector is the inherent possibility of illegal and/or dangerous markets, such as assasination markets. Since there is by design no administrator that curates the sort of questions possible, such unsavory markets can occur. While this has not been a particularly topical problem to date, it is an inherent problem to the sector that may deter widespread adoption from concerned mainstream users.
When examining summary metrics such as TVL, the DeFi space appears to be an uncomplicated story of growth over the past few quarters. However, this growth has been very narrowly concentrated, with certain applications such as lending and collateralized stablecoins representing the vast majority of this growth, in contrast with areas such as prediction markets and DEXs. Though the space is still in its relative infancy and many unsolved problems remain, the Ethereum DeFi ecosystem stands out as a promising development that has captured the mindshare of much of the developer community: at the recent ETHBerlin hackathon, over 25% of submitted projects were related to DeFi. Thus, despite the variability of demonstrable traction across the DeFi space when examining a more nuanced set of metrics than TVL, there is reason to believe that this emerging set of composable financial primitives presents new possibilities for the sector to expand further with new combinations of functionality, and application-specific growth metrics will be an important method to examine this industry development moving forward.