Forks and contracts
Most published analysis of the legal consequences of blockchain forks has been underwhelming.
Most published analysis of the legal consequences of blockchain forks has been underwhelming.
January 30, 2019
Discussions often centre around the legal risks to miners and developers, questions of little relevance because of the general absence of contracts between users of public blockchains and the constellation of jurisdictions from which they operate.
In other words, it will freeze in hell before anonymous developers based god-knows-where win a lawsuit against unidentified Chinese miners aggregated in a mining pool. I should add that several articles appeared to be advertorials by law firms looking for new business.
Yet, an ecosystem is building around cryptocurrencies and real contracts are being entered into. As is often the case in our industry, a lot of people have no idea what they are doing and today’s decisions with respect to contracts can have unforeseen or unintended consequences. Having been through a few forks over the past few years, our industry should now be in a position to devise best practices. In this article, I will debunk some unhelpful characterisations of what a fork is and propose simple clarifications to embed in contracts affected by them.
What is a fork, anyway?
Forks happen all the time. Two miners discover a block of the same height at the same time. You have an accidental micro-fork. Both chains are equally valid, but we can’t follow both and keep calling each of them “Bitcoin”: a few more blocks will have to be mined and the results propagated before we figure out what chain the network decides to discard. We never think twice about those forks, but their mechanism is not unlike that of more visible forks.
Contentious forks are intended to supplant the incumbent coin and come in two forms. If a fork implements a new transaction format to ensure that transactions are never valid across the two chains (“replay protection”), then you have a cohabitating contentious fork. That’s the Bitcoin Cash fork. The alternative, which I consider to be the “nasty” kind, is a competing contentious fork. That’s Bitcoin SV. Post-fork, two blockchains coexist and while their consensus rules are slightly different, the common transaction format and UTXO set make it so that, right after the fork, a given transaction is valid on both chains (and can be “replayed” from one chain to the other). For the chains to diverge, miner rewards (by construction specific to one chain only) and other consensus-breaking transactions must propagate through the network, resulting in two incompatible UTXO sets. It’s not a clean split and it can take a long while (or forever) before incompatible transactions “contaminate” the UTXO set and one can transfer old coins without unwittingly moving the new coins, and vice versa.
Lastly, you have spinoff forks. Those are new coins with new consensus rules that don’t intend to compete directly with their original coin, but use the original coin’s existing UTXO set (or its equivalent) to bootstrap the new coin with an initial distribution mirroring its parent’s, presumably supporting adoption.
Note that my examples refer to UTXO and other Bitcoin-specific terms, but the logic above can be applied to other blockchains.
Who can fork?
Anyone can fork. From a major schism in the community (ETH vs. ETC, BTC vs. BCH), to broadly accepted software upgrades, to yours truly forking BTC into MaxCoin in his garage. This last bit will prove to be key to my argument. The original creators of an open-source blockchain cannot prevent anyone from forking. As a result, there likely exist dozens of forks no one has ever heard of outside of their creators. In a parallel universe, perhaps the LDS Church is running a Litecoin spinoff fork called MormonCoin with manual pen-and-paper hashing, and outsiders don’t even suspect of its existence.
Bearer assets out of thin air
“If you don’t hold the keys, you don’t own the coins”. The adage is commonly repeated by crypto enthusiasts. While I disagree with the hyperbole, they have a point when it comes to forks. Any benefit (or cost!) that comes with a fork befalls the actual holder of the private keys, be it the beneficial owner or a third party “custodian” (e.g. an exchange; I use the term loosely). Crypto assets being (mostly) bearer assets, the only surefire way to ensure that you get to enjoy the full benefit of ownership is to secure the private key yourself.
Unfortunately, it is unrealistic to hope that the general population will follow the prescriptions of seasoned crypto geeks and safeguard their own keys. In reality, for the foreseeable future and perhaps forever, most coins will be held by third parties rather than directly by their beneficial owners. As a consequence, our industry needs to put in place best practices for the management of forks.
We will look at a few important types of contractual relationships affected by forks. The issues we need to address are (1) whether contracts should pass benefits through to the end beneficial owner or not and (2) how to best transfer any such benefit.
First, let’s address an analogy that is incorrect but nevertheless expressed frequently.
Forks are not dividends
Forks pop new assets into existence. Pre-fork, a number of coins are linked to your private key. Post-fork, the same private key entitles you to a new coin, transacted on a different network, that might have some value (however defined). This is sometimes called an airdrop.
While that may feel like a dividend, it’s not.
A dividend is a distribution of a legal entity’s past profits to its owners. Public blockchains don’t care about legal entities and the notion of past profit is not applicable. The ICO debacle of 2018 also showed that deeming holders of tokens to be shareholders can be unwise.
It is unhelpful to think of airdrops as dividends, because one might be led to believe that the legal framework(s) surrounding dividends is more relevant to forks than it really is.
Moreover, in the case of competing contentious forks, it can be hard to see how the old and new coin are separate at all. What does it mean to own coins? Is it the ownership of UTXO, or the ownership of a private key? In the absence of replay protection and before the forks have diverged, can it really be said that the two assets coexist? Even with the will to do so, it is not technically trivial to split one’s forked assets. Around the ABC/BSV fork, certain exchangesstarted trading both, but their customers were in reality trading IOUs on future coins in the exchange’s centralised database.
Even then, we’re none the wiser
There are different ways to borrow a thing, but in financial markets it oftens implies a transfer of ownership. Borrowed shares pay dividends to the borrower, not the lender. Operating otherwise would require that the company maintain a register of borrowers (who likely sold the shares anyway) separate from its register of shareholders - this is impossible in practice.
The concept of “a share of a company” does not differentiate between legal owner and beneficial owner, yet it is accepted that the stock lender (i.e. not the legal owner anymore) is entitled to the dividends one way or another. Same goes for repo market participants with interest on repurchased bonds. However, it is an industry standard, not a logical or natural necessity; so why do markets operate like that? It is because the repo and stock lending markets are markets in interest rates.
If the mere fact of borrowing a stock entitled one to dividends, then the expected value of that dividend would need to be reflected in the financing / borrow rates. That would make no sense. For a stock with an announced dividend on 10 January, a one-week stock borrow might trade at an average of 2% in early January, jump to, say, 50% on the 3rddue to the one-off payout, then drop back down to 2% after the 10th. It would be insane because the rates (i.e. prices) would become meaningless. Add the possibility of unannounced or uncertain dividends to the mix and your head will likely explode.
The convention that the “side effects” of legal ownership should flow back up to the beneficial owner is therefore necessary for funding markets to function properly. Similarly, cash bonds trade on clean price but settle on dirty price in order for the trading process to remain intelligible, because the alternative would be that quoted prices would change a little bit every day due to the accrual of interest. Here too, day-to-day price comparisons and comparisons between bonds would become meaningless.
Now a few problems arise if we accept that value should flow up to the beneficial owners. Keeping track of who owns what is already hard enough, what about dividend withholding taxes for instance? The tax treatment of dividends depends both on the country of issuance and the country of the borrower, coupled with a complex network of bilateral tax treaties. It’s no surprise that a “dividend arbitrage” cottage industry developed over time, and the costly blow-up of some strategies in recent years due to over-optimistic interpretations of tax laws tells us one thing: this stuff is complex. Mindlessly copy-pasting practices from established markets is not a good idea, and it’s already hard enough to calculate taxes on basic crypto holdings.
Nevertheless, we do have to take one important page out of the conventional book of financial contracts: because nothing in the nature of a stock can enforce the transfer of the dividend (or its equivalent) back to the beneficial owner, it follows that it must be contracted out. The same goes for the side effects of forks.
Knowing that the devil is in the detail, it will be easier to agree on principles first, and specific cases second. Let’s have a stab at the former, then consider how such principes could translate into actual guidelines for a few types of contracts
Please draft contracts sensibly
1. Liabilities should not arise without informed consent
If you take care of your neighbour’s diabetic cat while they go on holiday, but the poor animal dies because the neighbour forgot to explain that the cat must take pills twice a day, it’s on the neighbour, it’s not on you. But, conversely, if my car is in the garage for repairs and the premises burn down, the garage or its insurance will replace my car.
2. Assets are not liabilities
Remember the borrower who sold the borrowed shares? She’s short now, because she doesn’t have the shares anymore. At the inception of the loan, the borrower had both an asset (shares) and a liability (the obligation to return the shares). Selling the shares short (i.e. getting rid of the asset) is the whole point of borrowing them. Therefore, the idea that the borrower ought to reimburse the dividend does not find its justification in the fact that she benefited from it (she didn’t). The reason lies elsewhere. Liabilities do not imply assets.
3. Value should flow up to beneficial owners... if possible
It seems appropriate that the ultimate owners should enjoy the full benefit of ownership, but it is not always possible. Trivially, if a newly forked coin is not supported by the third party in actual possession of the asset, you’re out of luck. It can be impractical or downright impossible for an exchange to attribute new coins in a contentious competing fork, especially with the use of non-segregated “omnibus” wallets. It can also be illegal; I don’t expect US exchanges to allocate, say, cannabis-related assets anytime soon. Let’s not even consider less obvious benefits such as voting rights.
4. Not all forks are created equal
I looked up Bitcoin’s past forks and found many that I had no idea existed. Why is Bitcoin Cash a well-known fork, but not Bitcoin Diamond? The sad truth is that there is no objective criterion to distinguish material forks from negligible forks. Ideally, the indices against which some derivative contracts settle come with mechanisms and clear rules to determine what happens in the event of a fork. Should Bitcoin fork into 2 coins, each presumed to be worth exactly half of the Old Bitcoin, what price will futures contracts expire against? The sum of the two? The greater of the two? What if the component exchanges that form the index do not trade the new coin? To the detractors of cash-settled contracts: I admire the intellectual purity but physical delivery does not solve all problems.
5. Let’s give the free market a chance
Giving market participants the right incentives to agree between themselves might be a good solution if the framework is simple and standardised. I’m thinking of a parallel with American and European options. Everyone knows what the terms of art mean, the option holder does not need to read the fine print to understand what the best course of action is, at or before expiry.
Let us now consider how some contractual relationships might take the above into account.
Customer satisfaction should be the driving force here. Service providers have an incentive to keep customers happy, but the benefit of convenience can come with the risk of missing out on a new coin. If switching costs are low, we can hope to see standards emerge. It is important that custodians communicate with their clients ahead of potential forks to help them make informed decisions.
For a custodian, the main risk is in not supporting a newly forked coin before reversing course long after the fact, as happened with Coinbase and BCH. After a fork, splitting the coins should be a priority, as it leaves all doors open.
Generally speaking one imagines that it would be wiser for custodians to explicitly state that forks will only be supported on a best-effort basis. Limiting liability in time should also be considered, lest a minor fork unexpectedly becomes significant.
The ability to borrow cryptocurrencies is essential to a healthy market. Forks are very disruptive for funding markets as was seen recently in the ABC/BSV fork when BCH borrow rates shot up on Bitfinex and other markets.
I believe OTC coin lending relationships have the most potential. Taking inspiration for GMSLA/GMRA agreements, the behaviour of contracts around forks must be set in the terms.
My suggestion would be to use puttable and callable funding such that either
(1) if it is agreed that the value of a given forked coin should be transferred to the lender of crypto, the borrower has the right to return the funding with accrued interest ahead of an upcoming fork; or
(2) in the alternative, if it is agreed that no adjustment be made, then the lender of crypto should have the right to call the funding ahead of time.
In scenario 1, the value could be delivered physically or cash settled. One can also imagine that exercising the option would carry a penalty such that “de minimis” forks would not serve as pretexts by unscrupulous parties to tear up contracts.
For the lender, such a penalty could take the form of forfeiture of some interest; for the borrower, an increase in interest or extension beyond the amount accrued.
The fact that one can easily view crypto-currencies as money has interesting consequences for what we consider to be the valid extinguishment of debt. Remember that most people alreadyfind it difficult to understand money as debt. Loans such as mortgages are not repaid in actual legal tender, they are repaid by moving debt around, e.g. my own bank becomes a creditor of my employer’s bank when my salary lands into my account in time for the monthly debt payment. What action validly discharges the borrower from its debt? Crypto-currency transfers are electronic signatures batched in blocks that represent the history of the coin. What happens if I repay my debt but the blockchain is reorganised afterwards?
Chain splits can have a very material impact on derivatives whose term span a fork. The interest rate relationship between spot and future price breaks down as the derivatives starts reflecting the expected future value of the “winning” chain (contrary to popular opinion, the price of futures do not normally represent expectations about future prices).
When it comes to derivatives, certainty is the keyword. What happens post-fork must be made clear. With that in mind, cash settlement against an index might be safer than physical delivery in some respect. . In the vein of crypto transfers as electronic signatures, what constitutes delivery of a crypto asset into a derivative contract? The answer for conventional products such as bond futures is not a trivial one.
I don’t have all the answers, but at a minimum I would expect that most derivatives only ever track one chain in the event of a competing contentious fork with no replay protection. Because no objective criteria can distinguish between such forks, I expect that bodies similar to the ISDA Determinations Committee emerge over time*. Unfortunately, even for benign forks, rules are likely to become more complex as the people and companies behind the main derivatives contracts endeavour to do what’s right (or even what’s good for them), which sometimes conflicts with what is elegant. Expect a juxtaposition of exceptions that, while binary in appearance, give rise to a fuzzier whole.
I hope you can now see that much pain can be avoided with a little bit of preparation. I will leave you with a final plea to blockchain developers: for the love of all that is sacred, use replay protection!
First published in Profit&Loss, 16th January 2019