Fat Tailed Thoughts: Digital Dollars
Hey friends -
We've come (almost) full circle back to government monies, but this time the monies act a bit differently than the cash and other forms we're familiar with today. Why? These monies are digital.
To make the world even weirder, they're not actually issued by governments at all. These are privately issued monies that try to maintain a stable value. We'll explore the multiple competing models of how this can work, an area of active and exciting experimentation.
In this week's letter:
Digitizing the dollar - cash-backed, asset-backed, and algorithmic stablecoins
A day in the life of a port operator, the problems with returning your too-small pants, and more cocktail talk
The MonteNegroni - a fruity, lower alcohol take on the classic Negroni
Total read time: 14 minutes, 52 seconds.
The Rise of Stablecoins
We've discussed how government money does not exist digitally. Innovators around the world built other forms of money that can exist digitally - mobile money for mobile phone networks and virtual currencies in video games among many others.
They also built stablecoins, blockchain-based digital assets that maintain a stable value relative to another asset or assets. Most commonly, stablecoins are pegged to the US dollar such that 1 stablecoin is worth $1 USD.
Stablecoins have proved to be a useful form of money for both new crypto buyers and existing crypto owners. For new buyers, it can be expensive and slow to use cash to purchase cryptocurrencies. Stablecoins are a way to move cash into crypto "accounts" so buyers can then make purchases over time. The buyers quite literally purchase a $1 stablecoin in exchange for $1 of cash and trust that the stablecoin issuer will return $1 of cash when requested. For existing crypto owners, stablecoins are an attractive "store of value" in between trades, a money function that crypto has been particularly poor at serving.
Remember - these exist because there is no digital dollar. Instead, a private entity creates an almost-but-not-quite digital dollar that lives on a blockchain. As cryptocurrencies have grown from almost nothing to over $2 trillion in value, so too have stablecoins gained traction. A lot of traction. $4+ trillion of traction.
$4 trillion is 33% of Visa's last 12 months transaction volume. Not bad for a type of money that was used for just $13 billion in annual transactions four years ago.
Our definition of stablecoin leaves a lot of room for experimentation. As this new type of money has matured, three general models for stablecoins have emerged - cash-backed, asset-backed, and algorithmic. We're going to look at the tradeoffs among all three.
Cash-backed stablecoins are exactly what they sound like - a blockchain-based token that is backed 1-for-1 by cash in an account. "Backed" is a critical concept. As a stablecoin owner, you do not have legal title to specific dollars held in your name by the stablecoin issuer. The dollars are held in an omnibus account that comingles everyone's dollars and the issuer promises to send you your proportional share of those dollars when you redeem the stablecoins. Think of this as a digital IOU.
The omnibus model requires you to put significant trust in the issuer, trust that manifests as four distinct "getting started" challenges for the issuer:
Where do you store the cash?
How do you prove that there is exactly the same value of cash as tokens issued?
Before a buyer engages, how do you prove to the buyer that they will be able to exchange their stablecoins for cash in the future?
How you do make money?
Solving for trust
The answer to the first three questions can be found from traditional finance - go get regulated. We trust banks, brokers, and other financial institutions with our hard-earned money because we believe that they are such to stringent oversight and that we'll be made whole if something bad happens to them. When it turns out that the oversight was too lax (see: 2008) or the institutions were taking absurd risks (see: 1929, savings & loan crisis, 2008, and a lot of other dates), we get rightfully angry.
Nonetheless, getting regulated remains a particularly effective way of signaling that you're a good actor. Signaling is much more effective if it is backed up with operations that further demonstrate your commitment to safety and soundness.
To store the cash - issuers can deposit the cash in FDIC insured accounts or place it in segregated custody. The former is backstopped by the federal government up to $250,000 per depositor, per FDIC-insured bank. The latter is set aside from anything else so that even in the event of the issuer going bankrupt, you retain the legal claim on the underlying cash.
Regular and publicly posted audits by an independent third party can demonstrate that the cash is being managed correctly and that only as many tokens as there is cash have been issued. Publicly posted is critical - it ensures that the independent auditor can also notify the public if something went awry without the issuer interfering.
Proving future exchangeability is greatly eased by the audit reports and publicly posting the costs to convert to/from cash is also helpful. The public blockchain can give us a confidence boost over time as we observe the actual transactions where stablecoin holders redeem for cash.
Making money is a difficult part of this stablecoin model. Fees to exchange cash-for-stablecoins or stablecoins-for-cash are unlikely to be meaningful sources of revenue as both get competed away among issuers. Enabling other businesses to use your stablecoin infrastructure to create their own white-labeled stablecoins is a more promising model to directly create revenue. Alternatively, an issuer can "give away" the stablecoin for free to drive usage of other revenue-generating services, much like how many checking accounts are free.
The most dominant player in this space today is Paxos. They've both issued their own dollar-pegged stablecoin, the Paxos Dollar (USDP), and enabled Binance, a cryptocurrency exchange, to issue a white-labeled stablecoin, the Binance Dollar (BUSD). These two stablecoins are backed one-for-one by $14+ billion in US dollars in FDIC insured deposit accounts, and a combination of US dollars and US government treasury bills in segregated custody backing an equivalent amount of stablecoins.
A larger competitor is muscling into the space - USD Coin (USDC), issued by Centre. I say muscling in because although USDC is three times the size of USDP and BUSD, they're not actually cash-backed. They're transitioning to becoming fully cash-backed but, for the moment, remain backed by "dollar-denominated assets." That's a big difference.
Asset-backed stablecoins deviate from cash-backed by using assets other than cash as the store of value. As might be expected, there are many stablecoins backed by gold and other assets. The gold-backed tokens in particular raise interesting legal questions for the jurisdictions that accept gold as legal tender, including 12 US states. From a stablecoin model perspective, these share more similarities with the cash-backed tokens.
More assets, more risk
Where we get meaningful divergence is with stablecoins that are backed by a heterogeneous mix of assets yet attempt to remain pegged to a single value, such as $1 USD. In a fully cash-backed model, if owners attempted to redeem all of the stablecoins at exactly the same time, the issuer could meet all of the redemptions in full. If there is a mix of assets, it's likely that the issuer could not.
It's also possible that the value of the assets backing the stablecoin falls below the pegged one-dollar value. This isn't just a hypothetical - we saw it happen to a prominent money market mutual fund in 2008, a data point I argued should remove such instruments from consideration as money. I look consider mixed-asset-backed stablecoins to be similarly challenged.
Why would an issuer back a dollar-pegged stablecoin with assets other than cash? There's only one reason - to make more money. By investing the cash in bonds that generate interest, the issuer can make money until the stablecoin owner redeems the stablecoin for cash.
Centre's USDC is the second-largest asset-backed, dollar-pegged stablecoin. It was launched as a joint initiative by Coinbase, a cryptocurrency exchange, and Circle, a cryptocurrency payments and treasury infrastructure company. We can see from the May 2021 transparency report that a large proportion of the assets are not cash.
Tether remains the largest asset-backed, dollar-pegged stablecoin. I'm not going to do Tether the justice of reprinting their "audit" reports. Tether and its executives remain under investigation by the Justice Department, settled a case with the NY Attorney General's office that they misappropriated cash, and are likely tied up in other yet-to-be-announced investigations. While Tether has occasionally published audit reports, their track record of misappropriating funds materially undermines any confidence a reader might take from such a report.
A surprisingly large market share
With a current circulation of $70 billion, Tether is more than twice as large as USDC, which in turn is almost three times as large as cash-backed USDP and BUSD. This is unexpected. The least trustworthy stablecoins are the largest and most trustworthy the smallest. What is going on?
The story is remarkably simple. Tether was first. It retains its prominence and likely will continue to do so until a large event knocks it off its thrown such as a regulatory ban on holding the asset. USDC is the default stablecoin of Coinbase, the largest publicly listed US cryptocurrency exchange. When new crypto investors join Coinbase and convert their dollars into stablecoins, that stablecoin is USDC.
By being first and the default for a major exchange, both have obtained outsized prevalence. Exchanges support trades from Tether and USDC to other cryptocurrencies because buyers and sellers have used those stablecoins in the past. This is classic path dependence and network effects - they're big because they got there first and they've stayed big because they were already big.
From asset-backed to cash-backed
The tides are starting to turn. Centre's USDC in particular has come under attack by cash-backed stablecoin issuers for not being actually cash-backed despite having previously given such impressions, lack of transparency as to the underlying assets, and lack of appropriate regulatory scrutiny. Centre, to their credit, pivoted quickly in recent months to address the well-placed jabs. Below is the August audit report for USDC.
A positive reading of the report is that they are transitioning to becoming a fully cash-backed stablecoin and thus building trust. A negative reading of the report is they continue to have 8% of the assets in not-cash despite attempting to move to fully cash-backed, strong evidence that the assets in question cannot easily be converted to cash which should call the value of the stablecoin into question.
Backing a stablecoin with cash and assets is not the only way to ensure a stable value. The token can also be algorithmically stabilized.
As the name implies, algorithmic stablecoins may not be backed by anything all instead. Instead, a computer program maintains the stable value. There is tremendous experimentation ongoing with different models of stabilization, some of which failed spectacularly resulting in the value of the "stablecoin" going to zero. We're going to focus on the two largest - Dai and TerraUSD - which account for almost $8 billion of value.
Stabilize with already-printed money
A fair warning in advance - the model behind Dai has a lot of moving parts. I'm going to introduce the critical concepts first and then we'll break them down.
Dai is an algorithmic stablecoin that is created when MakerDAO approved Ethereum-based assets, proportional to their risk parameters, are transferred into Maker Vaults as collateralized debt positions.
Talk about a mouthful of jargon. Let's tackle it piece by piece.
"...MakerDAO approved Ethereum-based assets..." - MakerDAO is a program that returns on the Ethereum blockchain. Owners of MKR, the cryptocurrency issued by MakerDAO, vote to determine MakerDAO's rules. Those rules include which other cryptocurrencies to accept to back the collateralized debt positions.
"...proportional to their risk parameters..." - MKR owners also vote on how much of each asset is needed when it is used collateral. Maybe 1 USDP can back 1 DAI, but you'd need $1.50 worth of ether to back the same amount of DAI. The value of the assets is determined by another part of the MarkerDAO program based on the prices at which the asset is trading on various exchanges.
"...are transferred into Maker Vaults..." - these are blockchain-based programs that hold onto your assets until you redeem the Dai.
"...as collateralized debt positions..." - the Dai are actually debt backed by the assets you transferred into the Vault.
Now let's put this all back together. You can create Dai by transferring the required amounts of approved cryptocurrencies into specialized blockchain programs. This system is truly decentralized - anyone can create their own Dai by first launching their own Maker Vault and then transferring in the required collateral. If the value of the collateral falls below the required amounts per the risk parameters, the Vault is automagically liquidated and the contents auctioned off. No centralized parties needed.
While we've successfully created Dai, we haven't addressed how Dai maintains a peg to the US dollar. There are two main mechanisms: market-makers and interest. Market-makers sell Dai when the price goes above $1 USD and buy Dai when it falls below. Most of this is done by bots and MakerDAO actually open-sourced programs so anyone can run their own bot without much effort. Interest is paid via another MakerDAO blockchain program, the Dai Savings Rate. Similar to how central banks set interest rates:
If the price of Dai rises above $1 USD, MKR owners can vote to decrease the savings rate, thus reducing the price of Dai by increasing the supply of Dai (less Dai locked up in savings),
If the price of Dai falls below $1 USD, MKR owners can vote to increase the savings rate and increase the price of Dai.
It's a very cool model. Time will tell if it remains successful. I'm hopeful in part because an emergency shutdown is built directly into the program. If the required quorum of MKR owners votes for a shutdown, the program goes through predetermined steps to freeze the creation of new Dai, allow Dai owners to reclaim collateral, and auction off any remaining collateral. It takes real humility to write your will while still operating successfully, but doing so actually helps instill confidence in the asset.
Dai is just one model for a purely algorithmic stablecoin, a model that enables existing cryptocurrency owners to transform their cryptocurrencies into a dollar-pegged money without selling. An alternative model is to copy the central banks - create a new money and directly govern the money supply.
Why not print your own money instead?
TerraUSD maintains its peg to the US dollar in a different and particularly interesting way - by growing and shrinking a money supply. If this sounds familiar it's because the model is loosely based on how central banks maintain government money value.
To execute the model, the organization behind TerraUSD also created a cryptocurrency, LUNA, which powers the Terra blockchain. The blockchain acts as a market maker that will always exchange 1 TerraUSD for $1 worth of LUNA. Terra can also increase and decrease the supply of LUNA as needed. Let's look at each of these in turn.
The "$1 worth of LUNA" concept introduces a potential problem - how does the Terra blockchain determine the value of LUNA? We're back to the core of how public blockchains work. The Terra blockchain miners hold LUNA in digital escrow and independently submit estimates of what they think LUNA is worth. Those who are within 1 standard deviation of the agreed-upon price are paid in LUNA while those with estimates further afield have LUNA deducted from their digital escrow. It's an elegant solution to the price of LUNA problem as long as there are enough miners and well-intentioned miners have access to similar pricing data.
With the price of LUNA solved, we still need a mechanism to maintain the peg between $1 worth of LUNA and 1 TerraUSD. As LUNA holders sell LUNA to the blockchain in exchange for TerraUSD, the blockchain creates the TerraUSD and removes LUNA. In the parlance of blockchain, it mints TerraUSD and burns LUNA. As TerraUSD holders sell TerraUSD to the blockchain in exchange for LUNA, the blockchain does the reverse - it burns TerraUSD and mints LUNA. By varying the amounts of LUNA minted and burned relative to the LUNA price estimate from the miners, Terra can maintain a stable $1 value for TerraUSD.
When new LUNA are minted, the Terra blockchain has to decide what to do with them. They are sent to two places - firstly to the miners for providing the price input as well as other services and secondly to a Community Treasury. The treasury is used to fund new projects that are built on the Terra blockchain.
It's a very cool model. It remains to be seen whether or not it can withstand exogenous events like a significant crash in the value of LUNA, repeat LUNA price estimate divergences among miners, or massive upticks in the volume of TerraUSD and LUNA being exchanged. Such events have undermined government currencies with similar designs in the past. If the system proves it can withstand such shocks, TerraUSD will likely find a home globally among individuals who want access to dollars but don't have it today.
The Future of Stablecoins
What we've explored is three very different models for stablecoins, all with the intent of maintaining a peg between the value of the blockchain-based digital asset and some external asset like the US dollar. This is exactly what innovation looks like. Some of these models will work, others won't. We'll have competition in the meantime that will create winners and losers.
There is almost certainly a long-term home for algorithmic stablecoins. A money that can mimic other monies like the dollar, but is not subject to the same constraints as the dollar, is enormously powerful. For pornography, cannabis, and other similar companies that are engaged in legal activities but are routinely denied access to the banking sector, such monies are a lifeline. The monies are likely to play a similar role for money laundering, terrorism, and other illegal pursuits similarly cut off from the banking sector.
Asset-backed stablecoins pegged to fiat currencies are likely to die out in the long term. They inherit most of the constraints of cash-backed stablecoins yet instill less trust. That's not a very compelling combination. Stablecoins pegged to gold are more likely to be successful as they allow for fractionalization and transfer of an asset that historically has experienced high transactional friction.
Cash-backed stablecoins are the most fickle of the bunch. I expect they'll thrive in the years to come, but they will first undergo a further evolution in their model, one that will be forced by regulatory change and even newer innovation in the form of government-issued digital currency.
It's where we turn our attention next week to wrap up our exploration of money - central bank digital currencies.
As we continue to work our way out of the pandemic into whatever "new normal" is, one thing remains certain - shipping is still a mess. American manufacturers are waiting a record 92 days to assemble finished goods, in part because of inflation expectations and in part because 13% of the world's cargo is delayed. NY Times profiled the experiences of port employees attempting to manage the chaos. (NY Times)
Free Returns on everything is just one of the drivers of shipping chaos, but it is one of the least studied. Popularized by Zappos, an online footwear company, and now standard for most e-commerce companies, returns are a remarkably poorly studied aspect of supply chains. The Atlantic looks at the difficulties and costs for the companies involved. (The Atlantic)
Turn on the TV and you'll see pundits predicting the future. Never mind that they utterly failed to predict that we'd be here today - inflation, a rapid transition to remote work, a dramatic increase in e-commerce, supply chain problems, and so on. Sure, that's all "unprecedented" but the future - I'm sure! - will be "precedented." At times like this, I find Charlie Munger's commentary particularly helpful and insightful to keep me grounded. From 2015: "How can I be an expert in something I never even thought about that seems so unlikely?" (Forbes)
In an ongoing and increasingly weird battle supposedly over states rights, Texas has yet again thrust itself into headlines by inserting government into private businesses. Via executive order, the governor has banned private entities from requiring COVID-19 vaccines for employees. Put your feelings about vaccine mandates on pause - what we're watching is the continued transformation of the Texas Republican approach to government. The order both breaks the traditional "government overreach" party line and perverts the normal legislative process. (Texas Tribune)
Your Weekly Cocktail
A fruity, low alcohol variant on a classic.
1.5oz Amaro Montenegro
0.75oz London Dry Gin
0.75oz Dolin Sweet Vermouth
1 dash Angostura Bitters
Lemon peel for garnish
Pour all of the ingredients into a mixing glass. Add ice until it comes up over the top of the liquid. Stir for 20 seconds (~50 stirs) until the outside of the glass is frosted. Fill a rocks glass with ice. Strain the cocktail into the glass. Squeeze the lemon peel over the top of the glass to express the oils and drop it in. Enjoy!
Riffing on classic drinks can be a lot of fun. A classic Negroni (equal amounts gin, vermouth, Campari) can be a great canvas. A bit atypically, I wanted something sweeter with a lower alcohol content that still packed a punch of flavor. Amaro Montenegro shares the orange flavor of Campari but is sweet unlike the predominant bitterness of Campari. The juniper and alcohol from the gin offset the sweetness similar to how it offsets the bitterness in a traditional Negroni, and the bitters add a subtle spice that plays wonderfully with the orange. It's an easy-drinking cocktail that could take on entirely different profiles by further altering the ratios.