This post is part of CoinDesk's 2019 Year in Review, a collection of 100 op-eds, interviews and takes on the state of blockchain and the world. Haseeb Qureshi is a managing partner at Dragonfly Capital, a cross-border crypto venture fund.
2019 was a watershed year for DeFi. We saw the number of DeFi users increase by 30x, while the total loans originated via DeFi locked in DeFi more than tripledhttps://loanscan.io/loans?interval=max. MakerDAO, the first and most dominant DeFi player has clinched market share, while platforms like Compound, dYdX, and Uniswap have had their breakout moments. Wallets like Coinbase and Argent are now offering the effective equivalent of dollar-denominated savings accounts, all settled on public blockchains.
Just a year ago, DeFi was an obscure and somewhat speculative concept. Now nobody doubts that DeFi will be integral to public blockchains.
And yet, DeFi is still in its infancy. Most of the products are relatively simple: overcollateralized loans (known traditionally as secured lending) and simple on-chain settled exchanges, like 0x and Uniswap. The collateralization rates are in the neighborhood of 150%, meaning the value of the collateral the borrower needs to put up is 50% larger than the value of the loan.
This is a good start, but we’ve got a long way to go.
Bloomberg columnist Matt Levine often claims that crypto is replaying the history of modern finance, but in an alternate and sped-up timeline. If that’s true, then DeFi today is analogous to Venetian banking in the middle ages, when banking as we think of it today was just getting started. In a bustling port city like Venice, there were few long-standing reputations, and many interactions were one-shot games. There wasn’t a lot of room for identity and building up credit. In that context, you generally needed collateral and wide margins of security for a loan.
That’s where we sit today in DeFi: everything is overcollateralized, and your lenders know absolutely nothing about you. Any data, reputation, or iterated games are completely ignored by these protocols.
So what comes next? DeFi has a lot to improve on, but here are top three priorities for 2020.
The first is to increase efficiency by driving down overcollateralization margins. Most consumer mortgages operate on a 5-20% downpayment, making the total loan collateralization at 105-120% (since the house itself is already 100% collateral). 150% overcollateralization would be unheard of anywhere in normal finance.
Why are collateralization rates so large? There are two main reasons for this: volatility and latency.
Because volatile crypto is used as collateral, lenders need to protect themselves against a sudden drop in collateral value. It’s possible that crypto matures and becomes less volatile over time. But even without that, you can decrease volatility by onboarding more stable collateral, such as stablecoins, gold-pegged coins, or digital securities.
The more worrying issue is latency: if you see the collateral drop in value, how quickly can you liquidate the loan to ensure you don’t end up in the red against the borrower? After all, if a DeFi borrower’s equity has gone negative, you can’t claw back money from them—on the blockchain, they’d just abandon their address.
In this case, the latency of the underlying blockchain matters quite a lot. If you can decrease block times in Proof of Stake (or better yet, push DeFi applications to layer 2), and use a more professionalized network of keepers or market makers, these systems would be able to liquidate collateral much faster. This would allow them to bring their collateralization ratios closer to rates we see in traditional finance. That 50% margin is so large mainly because it’s currently plausible to have your collateral drop by 33% in value before the system can liquidate your position! With faster liquidations and more solid collateral, that overcollateralization ratio can drop by quite a bit.
After driving down collateralization rates, the second priority for DeFi is greater diversity of synthetic assets. Look at MakerDAO: MakerDAO produces dai, a synthetic dollar-pegged stablecoin, with no actual dollars being held anywhere in the system. All requires is an ETH/USD price feed. If we just swap out that price feed for ETH/gold instead, suddenly we can produce a synthetic gold-pegged token through the exact same mechanics. Maker and UMA are two protocols that are pursuing this, and I expect we’ll soon see a wide array of synthetic financial assets. In the future, anyone with a mobile phone will be able to buy whatever financial asset they want from anywhere in the world, all mediated by crypto.
Finally, DeFi will need to migrate toward iterated games. Instead of treating every user as a blank slate, protocols will take account of your long-standing on-chain cash-flows and revenue-producing assets. If your address has behaved well on other DeFi platforms, you should be eligible for lower rates and better credit.
Eventually, we should be able to import robust forms of identity onto the blockchain by building bridges with the real world.
Identity is a prerequisite here. After all, when it comes to lending, blockchains are not quite like the real world. If you default on your debts in real life, you must declare bankruptcy and surrender your credit for at least seven years. But on the blockchain, you can just abandon any address that’s defaulted and start over with a new address, no harm no foul. It’s impossible to force somebody into the red when they can just generate a new address for free. It’d be like letting people change their names and start over on their credit scores after declaring bankruptcy.
So we need stable, costly identity. But if we wait for governments and other identity brokers to provide this service, we might be waiting forever. More likely, as Balaji Srinivasan advocates, we’ll build bridges between a person’s blockchain and Web2 identity, which can be done without any governments getting involved. If I can use my Twitter or Facebook reputation to backstop my credit history on the blockchain, then we can build a purely digital credit system. There are a lot of unsolved problems here when it comes to privacy and k-anonymity, but in principle it should be doable.
The Internet disrupted monopolies in almost every industry, but somehow it left finance mostly untouched. DeFi introduces what was missing: a truly internet-native form of finance. Eventually, it will do to finance what the Internet did to every other industry. But it’s still early, all of these products are experimental, and there will undoubtedly be failures. But the power of permissionless innovation makes DeFi all but inevitable.