Welcome to another issue of Net Interest, my newsletter on financial sector themes. If you’re reading this but haven’t yet signed up, you can join 18,000 others and get Net Interest delivered to your inbox each Friday by subscribing here:
My Adventures in CryptoLand
I’ve been meaning to write about decentralised finance for a while. For those who don’t know, decentralised finance (or DeFi) is an emerging ecosystem of applications seeking to rebuild financial services from the ground up. Its roots lie in crypto, where it sits on top of Ethereum blockchain technology.
Truth is though, I’ve never really understood it.
All my career, I’ve been integrated into the centralised financial system. Straight from university I went to work at an investment bank where our job was to sit between buyers and sellers as middlemen in the stock market. After that I worked as a partner in a hedge fund for ten years. It wasn’t a central node in the financial system (although it was 20% owned by Morgan Stanley) but the financial stocks I looked at typically were, their value deriving from their role as institutional intermediaries in the global financial system. From a personal perspective, I’ve never been excluded from the banking system. I got a mortgage when I needed one and my network of contacts across the industry gives me access to all sorts of exciting financial products.
These experiences inform a view that centralisation works, as it has for many years. Before Visa, if you wanted to pay for something by card, the merchant’s bank would need to ask your bank directly for payment. “Back rooms filled with unprocessed transactions,” remembers the founder of Visa. By forming an institution to sit between the merchant’s bank and the cardholder’s bank, Visa fixed that. In the stock market, the DTCC plays a similar role. By netting transactions, these organisations inject more efficiency into the process and they also solve the problem of trust that impedes free flowing financial activity. Visa spends over $1 billion per year on advertising to engender trust among its users; DTCC has $12 billion of cash on its balance sheet.
However, centralised financial systems have several drawbacks:
They cede control to a small number of organisations. I happen to trust these organisations, but I’m aware others don’t.
The system limits access. Around 1.7 billion people around the world remain unbanked; they don’t have access to the financial system.
The system can be expensive, partly because of the legacy infrastructure it’s built on; remittance fees of 5-7% for example seem high when information can be moved around the world much more cheaply.
There’s a lack of interoperability between financial institutions globally. Many fintech companies have emerged to solve this problem – companies like Plaid, which allows customers to transfer financial information between providers; and dLocal, which allows multinational companies to take payment from customers in emerging markets using their own niche payment systems.
The financial system is seen as opaque. Many people don’t have a good sense of the financial health of the financial institutions they use and place their faith in deposit protection schemes. Having dedicated my career to analysing the health of financial institutions, this is one drawback I don’t get, but I guess not everyone enjoys interrogating bank balance sheets!
Ethereum blockchain technology paves the way for an alternative. It allows ‘smart contracts’ to be embedded in code on an open network. Such contracts can control assets and data and define interactions between assets, data and network participants. Over the past few years a number of protocols have emerged that deploy this technology in financial services. They cover trading, lending, borrowing, derivatives and more. Right now, there’s around $65 billion locked up in these protocols. It’s a small number – about the size of People’s United Bank of Connecticut – but a year ago that number was less than $1 billion.
The best way to learn is to do, so I’ve spent part of the past week attempting to ‘do’ decentralised finance. My goal was to recreate a decentralised bank of my own in an effort to earn some literal Net Interest.
Here’s how I got on.
1. Establishing a footprint
The thing about decentralised finance is that it lives in the world of crypto rather than in the world of traditional assets. So in order to begin I had to establish a footprint in crypto. Fortunately I already held a number of crypto brokerage accounts. I’ve had a Gemini account since 2017 which I’ve used to trade small amounts of Bitcoin. More recently I opened a Coinbase account to see what the experience was like ahead of the company’s listing, and I’ve also done some crypto trading on Revolut.
2. Friction at the border
I should say that my passage into the world of crypto wasn’t smooth. For a start, the traditional banking system didn’t make it easy to transfer funds across. The UK has a pretty good faster payments service and so, having tapped the payment instruction into my online banking app, I anticipated that I’d be ready to go in minutes. To my surprise, not only was the payment refused but my entire online banking app was disabled and I was asked to call a number to reinstate it.
I called the number and, after a frustrating hold, a friendly operator, Rob, explained that they needed to “make checks”. Rob did concede that Coinbase was a legitimate account and that he’d taken multiple similar calls over the past week, which made me wonder if the account was flagged simply to slow down the flow of funds, but who knows. Rob left me with a caution to be careful about being scammed. Ha! He doesn’t know that I’m a financially literate professional investor with 25+ years of trading and investing experience. I hung up and moved the funds across.
3. Getting scammed
And then I got scammed. I was scrolling through Twitter one evening and Tyler Winklevoss (or it may have been Cameron) who co-owns Gemini with his twin brother, tweeted that they were giving away free Bitcoin. All you had to do was transfer some of your Bitcoin to a special account and Gemini would send back double. The offer only lasted for a short while until the funds they’d allocated to the campaign were exhausted. I can see now of course that this makes absolutely no sense but they give away free coins all the time – Vitalik Buterin the founder of Ethereum was gifted trillions of Shiba coins, ultimately worth over $1 billion. Turns out it wasn’t Tyler (or Cameron). You live and learn.
4. Choosing a stablecoin
Fast forward a while and I’m now an experienced HODLer of Bitcoin. It’s time to do some decentralised banking. First I need to fund my bank. I don’t want to do that with Bitcoin because no sane banker can deal with that kind of volatility in his funding base, so I look around for a stablecoin. We’ve talked about stablecoins a few times here before – in this piece on central bank digital currencies and again in this piece about Facebook’s digital currency, Diem. Stablecoins are crypto assets that are tied to the value of other assets such as US Dollars. Unfortunately Diem isn’t up and running yet so I need to look elsewhere.
The largest stablecoin in circulation is Tether ($61 billion). Launched in 2015, “each Tether unit issued into circulation is backed in a one-to-one ratio… by the corresponding fiat currency unit held in deposit by Hong Kong based Tether Limited.” At least according to its whitepaper. But the New York Attorney General found that the company was overstating its reserves and as part of a settlement Tether was required to provide quarterly breakdowns of the assets backing its tokens. Tether’s recent disclosure for end March reveals that only 3% of its assets actually consist of cash, with much of its asset base exposed to risk through commercial paper (50%), corporate bonds (13%) and secured loans (10%). Meanwhile, its capital buffer is very thin: the gap between assets and liabilities is only 0.36% of total assets on a balance sheet of $41 billion (as at end March). So it would only take a small impairment of assets to erode some of the value of my coins which, unlike bank deposits, are not insured. No thanks!
The next largest stablecoin is USDC ($22 billion). It was co-founded in 2018 by Circle and Coinbase and is managed by a consortium called Centre. The consortium holds US$1.00 for every USDC in circulation in bank accounts, checked monthly by Grant Thornton. USDC has been endorsed by Visa which announced that it will settle the coin over its network. Purists don’t like it because of its centralised nature but hey, that’s the price for stability.
Unfortunately, you can’t exchange British Pounds (GBP) for USDC on the main crypto exchanges. I could have exchanged GBP into Tether and Tether quickly into USDC but that seemed like too much of a hassle, so I went for a third option: Dai ($4.6 billion).
Dai is different from the other two stablecoins because rather than being backed directly by US Dollar denominated assets, it is soft pegged using economic mechanisms that incentivise supply and demand to drive its price to US$1.00. Thinking about it, if I’m setting up a decentralised bank I may as well go all-in. According to its whitepaper, “unlike other Stablecoins, Dai is completely decentralized… [it] exists without the authority of a centralized entity.” (Although last March, when the price of crypto assets tumbled, USDC was introduced into its collateral pool to help maintain the peg and now makes up over a third of the pool, so maybe it’s not as decentralised as all that.)
The crypto crash this month provided a good test of the peg. On Wednesday 19 May, the value of one Dai fell to a low of $0.9958 and rose to a high of $1.0146 but broadly it performed as expected.
So I bought some Dai, which wasn’t cheap – the fee was 1.5% on Gemini (USDC would have been free on Coinbase). Nevertheless, funding secured.
5. Transferring custody
Next step was to get my Dai out of custody at Gemini and into my own hands. To do this I needed my own wallet. The one most people use is called MetaMask, which sits as an extension on your Chrome browser. Security is obviously a big issue here; banks employ thousands to oversee security, here it’s just me and I’ve already shown a track record of being breached.
MetaMask furnishes you with a seed phrase of twelve randomly generated words to remember and warns: “if you lose your seed phrase, MetaMask can’t help you recover your wallet.” They recommend that you write it down. “The most common method is to write your 12-word phrase on a piece of paper and store it safely in a place where only you have access.” Mine’s on a post-it note stuck to the back of my laptop (jk).
With my wallet installed, it was time to transport my coins from Gemini and Coinbase into my custody. I submitted a transfer request to Gemini and was shown some more warnings: “Please ensure the accuracy of this Ethereum address since transfers are irreversible. Never type an Ethereum address by hand!” I placed my trust in the efficacy of my Mac’s copy-and-paste feature and, after anxiously hitting refresh on my browser furiously for a few minutes, saw the coins appear in my wallet.
Coinbase offers deeper functionality. There are fees involved in executing transactions on the Ethereum blockchain – so-called gas fees – and transferring funds incurs them. Fees vary depending on how much work is involved (not much in the case of transfers) and how much capacity there is in the system. Coinbase warned that “fees are high right now” but offered different prices for different transfer speeds. I thought 45 seconds was about as long as I could hold my breath so went with that.
6. Becoming a bank
My means of deploying funds in the market is a protocol called Compound. Developed by economist Robert Leshner in 2019, it allows borrowers to take out loans and investors to provide loans at rates that are determined algorithmically based on the supply and demand of funds. On the borrowing side, collateral needs to be placed in the system so users can borrow one asset against another. In principle, this is no different from borrowing on margin, but this is a decentralised version. Compound looks down on the centralised alternative; according to its whitepaper, margin loans from brokers, “are trust-based systems (you have to trust that the exchange won’t get hacked, abscond with your assets, or incorrectly close out your position), are limited to certain customer groups, and limited to a small number of (the most mainstream) assets.”
As in other areas of crypto, decentralisation doesn’t mean control is evenly distributed. Compound is governed based on ownership of its token, COMP, which is issued to users as an incentive to use the platform. A third of the ownership is in the hands of three entities (Andreessen Horowitz, Polychain Capital and Bain Capital). But as a user, I now have some of those tokens. Compound has $13.2 billion of funds on its platform and some of those are mine.
To get started, I first connected my wallet to Compound. I then clicked a few buttons to place my Dai on the market. None of this was cheap. Every time I moved some funds around I incurred a transaction fee. The whole effort cost around $50. And it was a bit slower than I had expected, taking between 6 and 10 minutes for my instructions to be processed. But the rates currently available on Dai are quite attractive, so I should be able to earn back my fees quickly. They change dynamically, but right now Compound offers a 2.91% interest rate on Dai plus 2.03% in COMP tokens for a net rate of 4.94%.
According to Compound, there are 18,567 people like me currently supplying Dai into the market for a total amount of $3.1 billion. On the other side there are 2,907 borrowers, paying a rate of 4.49% (plus receiving 2.64% in COMP tokens, so their net rate is only 1.85%). How much they can borrow depends on the quality of their collateral but it can be in the range 60-75% of the value of the collateral.
The spread between the cash rate borrowers pay and the rate I receive is 1.58% which is lower than the customer spread a bank typically generates. As a comparison, the spread between what People’s United Bank earns on loans and what it pays on deposits is 3.27% as at the first quarter (although its loans carry different credit risk). But then, People’s United Bank has shareholders to remunerate and a headcount of 5,640; Compound has a mutual governance structure and a headcount of 19.
Having placed my funds in the market I can log on to Compound and see interest accrue to my balance every time a new Ethereum block is added – at the current rate, every thirteen seconds. There’s something hypnotic about seeing funds compound continuously in real time; now I get how bankers feel. In addition, The COMP tokens I’ve been ‘comped’ accrue as a separate holding, although their value goes up and down with fluctuations in the token’s price.
My only problem is that I’m not entirely clear what borrowers are doing with my Dai. Most of the activity on the platform – and across DeFi generally – appears to revolve around speculation; others are borrowing my coins so they can buy more coins. Speculation accompanies real world activity everywhere, but it rarely leads it. For now at least, the impact my new decentralised bank has on the real economy appears limited.
Part of the issue is that blockchains are largely self-encapsulated and have no knowledge of the outside world other than the transactions added to the native blockchain. Lending against assets that live outside the blockchain (e.g. houses) or against character (e.g. unsecured) therefore becomes difficult. The particular design of this protocol also runs into real world problems: prices that move up and down with supply and demand are fine in theory but can cause distrust in practice, as Uber surge pricing has demonstrated.
One thing’s for sure, though: DeFi isn’t going away. It survived the crypto collapse of the past few weeks without damage. As Raoul Pal tweeted, “This was a big two weeks for crypto and for the future financial system.” The issues of speed and cost will be addressed with the launch of Ethereum 2.0 which will allow much higher scaling.
So where does that leave me? I guess I’m more comfortable analysing lines of a bank’s balance sheet than lines of a Solidity program, so despite my early exploration I’m unlikely to be an early adopter. But I’ll carry on watching and learning and I’ll report back.
More Net Interest
The IPO of dLocal, an emerging markets payment company, launched this week. The company presents itself as the “Adyen of emerging markets”. In both cases payment volume comes from large global merchants, although dLocal has a narrower focus. Its business – which began in Brazil, even though the company is based in Uruguay – allows global merchants to tap into the distinct payment systems that operate in emerging markets all over the world. It operates in 29 countries where it facilitates payments via 600 different methods. That makes it pretty entrenched. Revenue yields are very high at 2.9% of payment volume, but a large chunk of that comes from FX. dLocal’s biggest customers include Microsoft, Amazon, Spotify, Kuaishou and Didi; its success hinges on how rapidly these companies scale up in emerging markets.
Caixin reports that Tencent is being asked by Chinese regulators to put its financial operations into a new financial holding company, just like Ant. Its financial activities span payments (WeChat Pay), banking, fund sales, microlending and insurance. According to the report, “China’s top financial authorities summoned 13 tech companies including Hong Kong-listed Tencent to a meeting on April 29 to tell them they will be subject to stricter compliance requirements on their financial operations. Participants from Tencent were asked to stay behind after the main gathering ended and given further ‘window guidance,’ the sources said.”
The move suggests that China is ahead of the curve globally in applying financial regulatory standards to non-banks. It’s dealing with a bigger issue, but as the financial activity of non-banks outside China grows, they have their own ‘window guidance’ to look forward to.
Insurance is a peculiar business because customers don’t really want it, hence the adage, “insurance is sold, not bought.” As much as she’s a customer, she’s also a counterparty: what’s good for her (a claim) is not good for the company. There’s a zero-sum dynamic to the relationship, which means that the classic Amazon flywheel around customer experience and lower pricing doesn’t work.
This concept got Lemonade tied up in knots this week. In a series of tweets, the company told of how its platform is getting better at “delighting customers”. One way it does this is, “when a user files a claim, they record a video on their phone and explain what happened. Our AI carefully analyses these videos for signs of fraud. It can pick up non-verbal cues that traditional insurers can’t, since they don’t use a digital claims process.”
It seems a strange way to “delight” customers by allowing AI to auto-reject their claims based on how their face looks or their accent sounds. The company realized its (PR) error, deleted the tweets and issued a denial. But this is what happens when your customers and your shareholders start mixing in an industry that doesn’t lend itself very well to that.