Wednesday, December 28, 2022

Let it burn


I sympathize with parts of the let it burn thesis. The thesis goes a bit like this:

Crypto is mostly gambling. It provides very little of social value, and may even be a net negative. The recent collapse of the FTX crypto exchange is illustrative of this. It would be a travesty for us to wade in after the fact and lever public resources to regulate crypto. To do so would grant undeserved credibility to the stuff. Thankfully, the collapse of FTX didn't spread into the real economy. Let's keep crypto unregulated and isolated. Leave it to die of its own accord.

Four quick push backs:

1. Crypto is not going to burn down because of FTX. We know this because it has already collapsed multiple times (2011, 2014, 2018) yet each time people come back and want to play the Dogecoin or Shiba Inu or Bitcoin games. Dozens of crypto trading venues have collapsed over the last decade due to fraud and incompetence, yet burnt customers keep coming back to the table to play. 

Crypto is fun and exciting. It's here to stay. Maybe it's time to set up some guard rails.

2. Yes, crypto is mostly gambling. But we already allow all sorts of gambling activities, including sports betting, online casinos, and speculation on double-leveraged VIX ETFs. We set requirements on providers of these activities in order to protect users from fraud and wrongdoing. Let's do the same for crypto.

Start with the venues that facilitate crypto gambling: so-called "exchanges" like Kraken, Binance, Crypto.com, and Coinbase. These platforms provide both brokerage and exchange services, combining two functions that regular finance has traditionally separated. Require these crypto broker-exchanges to comply with the same basic consumer financial protections that currently apply to non-crypto brokers and exchanges. This includes segregation of customer funds, third-party custody, regular auditing, and insider trading prevention.

The idea is to let people engage in risky gambling, but to do so as safely as possible.

Canada and Japan have already taken these steps. That's why Canadians and Japanese are much less likely to be on the list of those hurt by FTX's collapse than folks in Australia and U.S., which haven't yet gone down the road to regulating crypto broker-exchanges.

3. The failure of a casino or sport-betting site is rarely systemic. Likewise, regulated crypto venues will probably never pose significant systemic risk. Even if a crypto venue were to somehow became so integral to finance that its failure would be catastrophic, we have tools for this, like designating venues as systemically important financial institutions.

4. It's possible that the value that crypto provides to society one day transcends gambling. Crypto could  become a way to get a consumer loan or finance a startup. If so, better to hold a given type of crypto venue to whatever set of regulatory standards are the most appropriate, and do so now rather than later. If a crypto platform quacks like a bank, for instance, then regulate it as a bank, perhaps tailoring the rules a bit here and there to account for the peculiarities of crypto. If it quacks like a broker-dealer, then regulated it as a broker-dealer.

Tuesday, December 20, 2022

Six reasons why FTX Japan survived while the rest of FTX burned


[This is a reposting of my latest article for CoinDesk. Since writing the article, FTX has announced that it will be selling the few solvent subsidiaries it owns, one of which is FTX Japan. Meanwhile, other FTX spot exchanges, including FTX US, remain mired in bankruptcy. The potential for a quick transition to new management is just one additional data point that illustrates the relative advantages of Japanese crypto regulation.]

Japan Was the Safest Place to Be an FTX Customer

As regulators look to regulate exchanges in light of the FTX's collapse, they would do well to look to Japan, which has some of the most mature rules in the world. 

FTX was a massive hydra with subsidiaries across the globe. Amid FTX’s failure and entrance into bankruptcy court, one of these subsidiaries appears to be relatively unscathed: FTX Japan. Assuming FTX Japan makes it through, here are some things that other nations can learn from Japan’s experience. 

FTX Japan is a Japanese-based crypto exchange, formerly known as Liquid, that Bahamas-based FTX purchased in early 2022. Whereas the customers of most FTX entities are in limbo, FTX Japan says that it is close to paying out its customers in full:

"We have put together a plan for the resumption of withdrawal service, which has been shared with and approved by the new FTX Trading management team. Development work for this plan has already started and our engineering teams are working to allow FTX Japan users to withdraw their funds."

Japanese customers' cash and crypto will not be bogged down in U.S. bankruptcy proceedings given "how these assets are held and property interests under Japanese law," the exchange says. Meanwhile, the funds of customers of the flagship Bahamas-based exchange, FTX International; Chicago-based FTX US; and FTX Australia remain stuck in bankruptcy limbo.

What is it about Japan that may end up allowing Japanese customers of FTX to get their money before anyone else?

In brief, careful regulation of crypto exchanges.

Spurred by the failure of Mt. Gox in 2014 and the 2017 hacking of Coincheck, both Tokyo-based exchanges, Japan's Financial Services Agency (FSA) established a broad set of standards for crypto exchanges, or what it defines as Crypto Asset Exchange Service Providers (CAESP). The FSA is also responsible for overseeing banking, securities and exchanges, and insurance sectors.

Here are six key elements of the FSA's framework for overseeing crypto exchanges:

1. Japanese crypto exchanges must segregate customer fiat and crypto from the exchange's own crypto. That is, they can't deposit the exchange's own operating funds into the same account, or wallet, as their customers' funds.

A separation of funds reduces the scope for fraud. For example, it would have been easier for FTX executives based in the Bahamas to raid customer funds held at their Japanese subsidiary if those funds were mingled together with FTX's corporate money.

2. Going beyond segregation, Japanese exchanges must entrust customers' fiat money balances to a third-party Japanese institution – a trust company or bank trust – where they are managed by a trustee with customers designated as the beneficiaries.

By interposing a third-party trustee between FTX Japan and its customers, regulators would have reduced the latitude for FTX insiders to tamper with Japanese customers' cash.

Another advantage of a trust requirement is that it adds a layer of protection in the event of bankruptcy. Storing customers' funds with a third-party trustee prevents them from being diverted into a general pot where they can be claimed by an exchange's other competing creditors.

Other countries are less stringent. Take the U.S., for instance. U.S. exchanges, including FTX US, operate under state money transmitter law. While some states do require money transmitters to keep customer funds in a trust but many don't, including Florida, Pennsylvania and Georgia. This lack of a trust company layer may be one reason why FTX US customers haven’t heard a peep about getting their money back.

FTX Japan claims to be holding 6.03 billion yen worth of customer fiat in trust, or US$44 million.

3. A more explicit bankruptcy protection stipulates that customers of Japanese exchanges are entitled to receive payment in priority to general creditors in the case of bankruptcy.

Customers are creditors of an exchange. They own an exchange-issued IOU. But a crypto exchange may have other creditors including bond holders, bank lenders, suppliers or other subsidiaries holding inter-company debts. When an exchange goes under, all of these IOU owners are desperate to get some of the remaining crumbs. Putting customers at the very front of the line of creditors is a way to protect them.

Compare the luxury of being a Japanese customer of FTX to the plight of Australian customers of FTX. To their horror, they recently found themselves competing with the parent company, FTX Trading, for part of the Australian bankruptcy estate.

4. The FSA requires Japanese exchanges to keep at least 95% of customers' crypto in cold wallets. Because cold wallets are not connected to the internet, they are more secure against hacking and internal fraudsters.

FTX Japan claims it currently holds 3,194 bitcoin (BTC) in cold wallets, as well as 16,418 in ether (ETH), 64.1 million XRP and a handful of other assets.

Many exchanges in unregulated jurisdictions already use cold wallets (although probably not for 95% of their customers' funds). However, smaller exchanges may use other exchanges such as FTX to store customer funds rather than their own cold wallets.

Australian exchange Digital Surge, with around 30,000 customers, recently entered into voluntary administration because it kept a significant amount of money on FTX. Huobi lost $13.2 million worth of customer funds that it had stored on FTX, while Crypto.com had $10 million in exposure.

Japan’s 95% cold wallet rule helps protect against such losses, as does the following 5% rule:

5. For the 5% of customer's crypto that can be kept in a less-secure hot [internet connected] wallet, Japanese exchanges must "back" each unit of hot-walleted crypto with exchange-owned crypto held in a segregated cold wallet. So, for example, if an exchange holds 5 BTC of customer funds in a hot wallet, it must hold another 5 BTC of its own personal coins in reserve, for a total of 10 BTC.

The FSA refers to these reserves as an exchange's performance-guarantee assets. If there are any inappropriate leakages from hot wallets, the exchange's reserve must be used to make customers whole.

6. Lastly, all of these rigid requirements must be verified by an external watchdog.

Each Japanese exchange must undergo a yearly "audit of separate management" whereby a public accountant examines that each of the above requirements for holding assets are abided by. That is, the auditor verifies that all customer fiat money is being held in trust, that customer funds are segregated from exchange funds, that at least 95% of all crypto is held in a cold wallet and that the exchange is holding an appropriate amount of performance guarantee assets.

FTX Japan customers haven't received their funds back yet. So we don't know for sure if they’ll be made whole. But initial indications suggest they will be. If so, credit goes to the six preceding protections afforded to customers of Japanese exchanges.

In response to FTX's failure, many jurisdictions are already scrambling to fashion their own regulations for crypto exchanges. They should be watching Japan closely.

Saturday, December 17, 2022

How cryptocurrency exchanges peg stablecoin prices

An example of a stablecoin peg from the now defunct FTX US [source]

This post is for anyone who is curious how cryptocurrency exchanges and stablecoins work behind the curtains. It's common knowledge that stablecoin issuers like Tether, Paxos, and Circle run pegs. What isn't commonly known is that crypto exchanges like Binance (and the now-failed FTX) also run their own versions of stablecoin pegs.

Stablecoin issuers like Tether anchor, or peg, the value of their tokens to $1 in fiat dollars, and use their dollar reserves to enforce that peg. Another way to think of the process of pegging is to take two heterogeneous things and use economic resources to make them homogeneous, or fungible, with each other in terms of price.

Exchanges such as Binance peg stablecoins in two ways:

1) Pegging multiple stablecoins to each other

Most crypto exchanges do not peg stablecoins to other stablecoins. They let the price of stablecoins float, or fluctuate, against each other. That is, if you deposit 1 unit of USD Coin and 1 unit of Binance USD to an exchange, you don't get credited with $2. You get credited for a single unit of each heterogeneous stablecoin. The exchange rate between these two coins fluctuates on the exchange according to supply and demand.

FTX was the first exchange to move from floating to pegging stablecoins. Binance followed FTX when it adopted its own pegging mechanism this fall. Basically, Binance promises to treat heterogeneous stablecoins in a homogeneous manner, by allowing customers to deposit any amount of approved stablecoins at the exact same price; $1. Customers can also withdraw whatever stablecoin they wish from Binance at $1, in any amount.

The approved basket for both Binance and FTX includes USDP, USD Coin, Binance USD, TrueUSD, but not Tether.*

By promising to process all stablecoin transactions at a uniform rate, the former-FTX and Binance shifted from the traditionally passive let 'em float practice of dealing in stablecoins to setting, or administering, stablecoin prices.

Pegging a basket of stablecoins is more complicated than letting them float. It requires having sufficient reserves of each stablecoin in order to defend the fixed price. If Binance users all want to suddenly withdraw a certain brand of stablecoin, but Binance runs out of reserves of that type, then it'll have to temporarily suspend its peg, at least until it can acquire more of the in-demand stablecoin.

It appears that this was exactly what happened to Binance earlier this week. When customers wanted to withdraw large amounts of USD Coin, Binance ran out and had to temporarily suspend USD Coin withdrawals. "In the meantime, feel free to withdraw any other stable coin, BUSD, USDT, etc." wrote the exchange's owner, Changpeng Zhao. 

Later, Binance sent a massive chunk of its own Binance USD hoard to the issuer, Paxos Trust, for redemption into fiat US dollars, before turning those dollars back into USD Coin (by going through Circle, USD Coin's issuer). Binance's coffers refilled, it could thus reestablish its peg.

Let's move onto the second type of peg that exchanges set.

2) Pegging the same stablecoin on different chains to each other

Stablecoins of the same brand exist on different blockchains. Tether, for instance, exists on both the Tron and Ethereum blockchains, as well as a host of other chains. The same goes for USD Coin.

Exchanges always peg a given stablecoin across its multiple instances.

What I mean by that is exchanges allows customers to deposit any amount of Tether (on Tron) or Tether (on Ethereum), and the exchange will treat those heterogeneous deposits as a single homogeneous Tether unit. And when customers want to withdraw, they can withdraw any amount of TethersTron or Ethereumfrom that pot at the same fixed price.

But Tether-TRX and Tether-ERC are not homogeneous tokens. They are very different beasts, with different characteristics, use cases, and demographics. Exchanges could in principal treat each instance of Tether separately, letting them float against each other. So in a given minute a single Tether-on-Tron token might be worth 1.001 Tether-on-Ethereum token, and the next 0.998 according to supply and demand.  

Exchanges don't do this. They peg the two instances of Tether. Customers take this for granted, but it's thanks to these exchanges' pegs that a customer can deposit 1 million Ethereum-based Tether onto an exchange and three seconds later withdraw 1 million Tron-based Tether, all at a convenient fixed price rather than a floating one.

To maintain these intra-stablecoins pegs, exchanges must have sufficient reserves of all blockchain flavors of Tether. (And all flavors of USD Coin and Binance USD, too.) Sometimes you'll see an exchange accumulating too much of one type of Tether while running out of the other type, and it'll engage in a swap with in order to rebalance its reserves.

This is likely what happened to Binance this week, when it swapped a massive 3 billion Tether-on-Tron into 3 billion Tether-on-Ethereum. Too many customers we're withdrawing Ethereum-based Tether, and so it had to rebalance its reserves:

In the next section, I'm going to sketch out the bigger picture.

Crypto exchanges as stablecoin watchdogs

I generally think it's a good idea for exchanges to treat stablecoins homogeneously, both in the first way (pegging stablecoins to other stablecoins) and the second way (pegging a stablecoins across its multiple instance). Doing so makes things easier for customers. Could you imagine, for instance, if exchanges didn't peg the different flavors of Tether, USDC, and BUSD? You'd end up with dozens of different stablecoin exchange rates:


But creating a stablecoin standard isn't costless. Exchanges need to devote resources to constant management of their reserves. If they make a mistake, as the case with Binance this week, they end up looking bad.

Pegging stablecoin to other stablecoins opens exchanges up to credit risk, too. If a given stablecoin suddenly collapses, exchanges that let stablecoins float needn't worry about a thing. They can continue accepting deposits of the failed coin at its market price. 

Not so exchanges that administer stablecoin prices. Traders will rapidly send the now worthless stablecoin to any exchange that is still pegging it at $1. To prevent the danger of becoming a sop for failed stablecoins, exchanges like Binance have to constantly surveil the stablecoins in their basket for credit risk.

In the grand scheme of things, this is probably a good thing. It deputizes exchanges as stablecoin watchdogs. Since exchanges have significant resources and insider knowledge, they are probably better at analyzing stablecoins for credit risk than outsiders like myself. Binance's basket of fungible stablecoins becomes a signal to the market of what stablecoins are safe.

We already saw an example of this stablecoin watchdog role in action not too long ago. A stablecoin called HUSD began to wobble in August:

After regaining its peg, HUSD outright failed in October, collapsing from $1 to a few pennies. 

The collapse seemed to come out of the blue. No so. FTX, the first exchange to peg stablecoins, had quietly removed HUSD from its stablecoin basket in early August, tipping anyone who was observing that something was up. 

However, if there are ecosystem-wide benefits to exchange stablecoin pegs, there are also drawbacks. Exchanges that treat stablecoins homogeneously (and thus take on credit risk) may do a poor job of it, and thus the fallout from a major stablecoin failure could spread to exchanges, an isolated failure becoming a systemic one. The benefit of the traditional practice of letting stablecoins float is that it renders the crypto exchange system more immune to the systemic risk stemming from the failure of a stablecoin.


*Why is Tether not included in these stablecoin baskets? One theory is that exchanges like Binance and FTX are acting as watchdogs and don't want to include Tether because of its unique credit risk. That's possible, but I think it's more likely that they don't want to include Tether because managing Tether reserves is too costly. This cost arises from the fact that Tether charges a 0.1% fee on all withdrawals and redemptions of Tether tokens. Other stablecoins provide this service for free. As long as this fee exists, it's just not worth it for exchanges to include Tether in their basket.

Monday, December 12, 2022

 Are U.S. banks more competitive than Canadian banks?

Over the years I've had a lot of connections to the Bank of Montreal. I'm a disgruntled ex-customer, a fairly happy shareholder, and a former employee. I stopped being a customer after the Bank of Montreal began charging me monthly fees in the middle of the pandemic without telling me, and I didn't notice for over a year. They refused to refund the fees, so I walked.

In any case, given my multiple interactions with the Bank of Montreal, I try to keep tabs on what it is doing. I was glancing through the bank's 2022 annual financial statement and stumbled on the following notable table:


Source: BMO. P&C refers to personal & commercial banking


The bits that struck me are in yellow. Bank of Montreal's net interest margin is much higher in the U.S. than Canada. By way of background, Bank of Montreal is fairly unique in that it operates as a sizable commercial bank on both sides of the U.S.-Canada border. So its data, including its margins, provides some interesting insights into the fundamental differences between U.S. and Canadian banking.

Net interest margin is a measure of how much a bank is squeezing out of its customers. To calculate it, start by counting up how much money a bank makes in interest on its loans. Then subtract from that its interest costs: all the money it pays out to depositors in the form of interest. That difference is the bank's net interest. Divide net interest by all of the money it makes on its loans to get net interest margin

Banks want higher margins. Their customers don't. The higher the net interest margin, after all, the more interest the bank is extracting from its customers.

In Bank of Montreal's case, its margin in the fourth quarter is 3.88% in the U.S. and 2.66% in Canada. So for every $100 it lent, the bank collected net interest of $3.88 in the U.S. but just $2.66 north of the border. In short, Bank of Montreal was much better at squeezing Americans than Canadians in 2022. That difference in margins doesn't sound like much, but repeated over billions of dollars it comes to quite a gap.

This isn't a fleeting phenomenon. I glanced over the last 10-years of Bank of Montreal financial data, and its U.S. net interest margin has been consistently superior to its Canadian margin over that entire period.

This goes against my long-standing stereotype of Canadian vs U.S. banking, which goes a bit like this:

I've always thought that it was better to be a U.S. banking customer than a Canadian one. Canada once had a fairly vibrant banking sector, but after many waves of mergers and acquisitions it has consolidated to the point that we've really only got five big bank. Everyone refers to them as an oligopoly. Everyone. I recall that even the Bank of Montreal's in-house bank equity analyst routinely referred to Canada's big 5 as an oligopoly in his research reports.

To make matters worse, Canada prevents foreign competitors from entering and stirring up the pot.

But America is huge and thus capable of supporting a much richer range of banks. For instance, the big 5 Canadian banks hold assets equal to 2.5 times Canada’s gross domestic product, but the assets of the five largest U.S. banks amount to just 0.4 times of that country’s GDP. See the chart below:

This lack of concentration means that U.S. banks don't have the same oligopolistic stranglehold over Americans that Canadian banks do.

On top of that, U.S. commercial culture is more cutthroat than Canada. Whereas foreign banks are locked out of Canada, they can freely enter the U.S. market. And so I saw the U.S. as an arena for ferocious bank competition, with customers benefiting in the form of better services and higher interest rates. Meanwhile, we Canadians are getting stiffed by our banks.

But after looking Bank of Montreal's net interest margins, I'm not so sure about my stereotype. A lower net interest margin in Canada means that the bank is extracting a smaller pound of flesh from its Canadian customers, which suggests more banking competition up here, not less.

Incidentally, net interest margin doesn't include those pesky user fees we all hate, or what Bank of Montreal calls non-interest revenue. And we know that the Bank of Montreal ruthlessly skins its customers for fees; after all, that's why I closed my account. However, even after adding Bank of Montreal's non-interest revenues to its net interest income on both sides of the border, its Canadian banking business still only sports a margin of 3.5% in fiscal year 2022 compared to 4.5% for its American business.

That is, even after accounting for pesky user fees, Bank of Montreal is still gouging its American customers more than it gouges its Canadian ones.

Admittedly, Bank of Montreal provides just a single data point. So I cast around for more data, and stumbled upon a database called Bankscope, hosted on the Federal Reserve's FRED. Bankscope is a popular source of bank balance sheet information among banking economists.

Here is what U.S. and Canadian net interest margins from Bankscope look like:

Chart source: FRED

It confirms my Bank of Montreal anecdote. Going back to 2000, banking net interest margins in the U.S. have been consistently higher than in Canada, and by quite a large amount.

To sum up, given the preceding data I may have to revamp my conceptions of Canadian and U.S. banking. It's true that we have an incredibly concentrated banking sector up here in Canada, with the big 5 controlling an outsized chunk of the market. Paradoxically, this "oligopoly" doesn't translate into higher net interest margins for Canadian banks. Margins are actually more elevated in the the hotbed of capitalism, the U.S., even though its banks are far more diffused. This margin difference suggests that competition among banks is more strident north of the border than south of it. 

In short, although the bastards at the Bank of Montreal skinned me for a bunch of fees during the pandemic, the bigger picture is that it's better to be a customer of a Canadian bank than a U.S. one.