Friday, August 17, 2018

Two notions of fungibility

A few centuries ago, lack of fungibility used to be a big weakness of monetary systems. But technological and legal developments eventually solved the problem. Nascent systems like bitcoin are finding that they must wrestle all over again with fungibility issues.

Fungibility exists when one member of a population of items is perfectly interchangeable with another. So for instance, because your grain of wheat can be swapped out with my grain without causing any sort of change to our relative status, we would say that wheat grains are fungible. Fungibility is a desirable property of a monetary system. If all monetary items are interchangeable, then trade can proceed relatively smoothly. If monetary items are not fungible, then sellers cannot accept the monetary item without pausing for a few moments to verify and assay it, and this imposes frictions on trade.   

In this post I argue that there are two ways for something to be fungible. They can be fungible for physical reasons or for legal reasons.

By physical fungibility, I mean that members of a group are objectively indistinguishable from each other. In the previous example, our wheat grains are physically fungible because a cursory inspection shows that they look, feel, and smell exactly the same. Now, a deeper analysis might reveal that the two grains are not in fact perfectly fungible. For instance, it may be the case that your grain of wheat is the hard red winter variety and mine is durum, in which case they are not substitutes, durum being better for making pasta. Or perhaps we each have durum grains, but yours enjoyed an excellent growing season—plenty of sun and sufficient rain—whereas mine isn't so healthy. And so your grains can produce more pasta per than mine. And thus they aren't exact substitutes.

We could even go down to the molecular level and determine that the grains are not perfectly equal and thus not quite interchangeable. But for commercial purposes, there is typically some sufficiently-deep level of analysis at which fungibility between types of wheat grains can be established by an experienced grain inspector and accepted by the market. 

Among commodities, gold and silver achieve a notably high level of physical fungibility. As long as a gram of gold is pure, it is perfectly exchangeable with any other gram of pure gold. Gold's fungibility doesn't necessarily carry over to gold coins, however. Earlier processes used to make coins, in particular hand striking, were not very effective at creating perfectly equal specimens. The edges of coins were often irregular, leaving coins vulnerable to clippers who would safely cut off some gold (or silver) without fear of being detected. Thanks to natural wear and tear, coins that had been in circulation for a few years would contain less precious metals than new coins. Both clipping and natural wear & tear meant that the metal content of coins was not uniform.

New technologies helped increase the physical fungibility of coins. For instance, reeded edges—those little lines on the edge of a coin—prevented people from clipping off bits without detection.  It was now obvious to the eye if someone had attacked the coin. Likewise, shifting from hand-hammered coinage to mechanical screw presses allowed for a more circular final product, one less susceptible to clippers (see comparison below). The invention of restraining collars—which prevented metal disks from shifting around while they were being stamped—also helped. With clipping much reduced, coins that had been in circulation for a while were more likely to be equal in weight to new ones.

These two photos compare hammered coins to milled ones (source)

In addition to physical improvements, an attempt was also made to buttress the fungibility of coins with laws. There are two types of laws that achieve this: legal tender and the so-called "currency rule." Legal tender laws required debtors and creditors to accept all coins deemed legal tender by the authorities at their stipulated face value. So even if two different shillings were not physically fungible--say one was clipped and worn and thus contained far less silver than the second newer one--those participating in trade were obligated to treat them as if they were perfectly interchangeable.

Legally-enforced fungibility was no panacea. In the absence of physical fungibility, the imposition of legal tender laws often had  perverse effects. If two coins were not exact physical substitutes because their metal content differed, but law required them to be treated as interchangeable tender, then the owner would always spend away the lighter one while hoarding the heavier one. Legal tender laws, after all, had artificially granted the "bad" coin the same purchasing power as the "good" coin. Thus the good money is chased out by the bad, which is known as Gresham's law.
The second set of rules that courts formulated in order to help fungibility, the currency rule, requires us to shift our attention to banknotes. Like coins, banknotes are not particularly fungible in the physical sense, but for a different reason. Banknotes have historically carried a unique identifier, a serial number—coins haven't. An owner of a banknote can carefully jot down the serial number of each note and, if it is stolen, use that number to help track it down.

In 1748, Hew Crawfurd did exactly this. Before sending two Bank of Scotland £20 notes by the mail, Crawfurd not only recorded their numbers but also signed the back of each one with his name, thus further breaking down their physical fungibility. When they went missing, Crawfurd was able to use this lack of fungibility to his advantage by advertising in the newspapers the numbers of the two stolen notes and the fact that they had been signed by him. One of the notes was eventually identified after it had been deposited at a competing bank, presumably long after the robber had spent it. The bank, however, refused to return the stolen property to Crawfurd.

In the resulting court case, the judge ruled in favor of the bank. Crawfurd would not have his stolen property returned to him. The court reasoned that if the note was returned to Crawfurd, then no merchant would ever risk accepting a banknote unless they knew its full history. This would damage the "currency" of money. After all, requiring merchants to pour through newspaper after newspaper to verify that no one was advertising a particular serial number as lost or stolen would be prohibitively expensive. Banknotes would be rendered useless, depriving the Scottish economy of much of its circulating medium. By allowing merchants to ignore the lack of physical fungibility of banknotes, i.e. the unique marks on each banknote, the court recreated fungibility by legal means. To this day, the currency rule that was first established in Scottish courts in the 18th century continues to apply to banknotes in most legal systems. (Kenneth's Reid's full account of this case is available here).

Bitcoin, a purported monetary system, is interesting because it: 1) lacks physically fungible and 2) is unlikely to ever be granted legally fungibility in the form of legal tender status or via an extension of the currency rule.

Bitcoin's lack of physical fungibility is more similar to that of banknotes than coins. It arises from the fact that all bitcoin transactions are publicly recorded. This means that it is possible to trace the history of a given bitcoin. If the token has been stolen, say in a highly-visible exchange hack, then said token may not be as valuable as a bitcoin that has a clean history. In theory, a forward-thinking actor will only accept a tainted coin at a discount because there is always a risk that the original owner will be able to reclaim his or her stolen property.

There seems little likelihood that the courts will solve bitcoin's lack of physical fungibility by fashioning a form of legal fungibility for it. The state will probably never be friendly enough toward bitcoin to grant it legal tender status. Nor do I think it is advisable that courts extend the currency rule to bitcoin by granting merchants the right to ignore the trail left by a given bitcoin, as they do with banknotes. As I pointed out here, to do so would violates the property rights of the original owner of the stolen objects. Only a select few instruments, those that have already proven themselves to be vital to facilitating society's trade, should be protected in this way.

With no legal route to establish fungibility, the only path remaining for bitcoin's architects is to go back to square one and try to improve the physical equivalency between bitcoins. One way they can do so is by anonymizing the blockchain. If transactions can no longer be traced, than clean and dirty bitcoins all look exactly the same. Full anonymity is easy to implement in new cryptocurrencies. Monero and Zcash, for instance, have gone this route.

In the case of a legacy cryptocurrencies like bitcoin, this functionality would have to be added on to its existing codebase. I have heard rumours that bitcoin developers like Adam Back and Greg Maxwell are working on developing code for anonymizing the bitcoin blockchain. But even if the technology is up to snuff, given the difficulties of achieving sufficient consensus for upgrading bitcoin, it remains to be seen if a fungibility-restoring technology could ever get off the ground.

In my view, the idea that bitcoin developers must try to achieve the same level of fungibility as coins and banknotes is misguided. Proponents of this idea are operating on the assumption that bitcoin is, like coins and banknotes, a payments medium or monetary system. But this is wrong. Whatever its original purpose might have been, bitcoin's first and foremost role is as a new type of gambling machine, a global and decentralized financial game. Like lotteries, casinos, and poker tournaments, and other types of zero-sum games, the main service that bitcoin provides its users is the fun of gambling and the allure of becoming very rich. If they want to benefit their users, Bitcoin developers should be working towards furthering its role as a gambling machine rather than mistakenly pursuing the dream of becoming the next monetary system.

People who play financial games such as lotteries benefit from the unique serial number on lottery tickets. If their tickets are stolen from them, this identifier may allow the original owner to get their ticket back. And that way they can still potentially win the big pot.

The same applies to bitcoin. Most people who hold bitcoins are doing so because they expect its price to hit $1 million. At least if their coins can be traced, a bitcoin owner who has been robbed may still have a chance to win that jackpot (and buy that Lamborghini they've been dreaming about). Removing the very feature that makes bitcoin non-fungible—and thus potentially traceable in the case of theft—would only do harm to the average bitcoin user. Anonymizing the bitcoin blockchain would make about as much sense as removing the serial numbers on lottery tickets.

Bitcoin's lack of fungibility isn't a bug, it's a nice feature.


  1. Are any of the other crypto currencies anonymous?

    1. Maybe Dash? Probably others, but along with Zcash and Monero those are the top three.

  2. After reading your last two posts, it seems to me, that you base your analogy between banknotes and bitcoins on a wrong assumption.
    You seem to think (but perhaps I'm wrong), that like with banknotes, a bitcoin transaction tranfers a bitcoin from one owner to another, so that in the end you a able to see a "transfer path" of every bitcoin in the blockchain.
    That is not true however. In reality, a bitcoin transaction, unlike a transaction banknotes, always destroys (spends in the bitcon talk) some bitcoins and creates some new bitcoins of the same value. Because auf that, it is not possible "to trace the history of a given bitcoin", this "history" always consists of only two transaction, the first one creating the bitcoin and the second one destroying it.If there is an analogy here, it's rather not to banknotes, but to wire transfers, though this analogy wouldn't be perfect either.
    It is of course possible (like with wire transfers) to track a payment history, something like:

    A got in an illicit way 1 bitcoin from B and paid 0,5 bitcoin to C and 0,5 bitcoin to D. Now both C and D could be forced to give 0,5 bitcoin to B.

    1. The analogy of banknote transactions to bitcoin transactions may be a bit sloppy due to technical reasons.

      But I think the following element of the analogy is valid: the UTXOs in your digital wallet may not be fungible with the UTXOs in mine because mine can be linked to a theft like the Mt-Gox hacking incident whereas yours are untainted. The banknotes in our wallets will always be perfectly fungible, even if mine were stolen from a safe in Mt-Gox's office.

  3. Excellent

    I look forwards the the bitcoin that was used to pay for the last load of drugs a v. famous popstar used to be auctioned at $1m

    ... or the BTC banksy used to fly to china.



  4. Fungibility is just another name for the liquidity test of an asset. Bankrupt-u-Bernanke provided that failed test with the remuneration of IBDDs. Setting up an interest rate differential in favor of the DFIs, is in direct contrast to the administration of Reg. Q ceilings where the differential was in favor of the thrifts, NBFIs. Thus MBS were negatively impacted on both sides of the ledger domain.

  5. May I suggest a different starting point to your story.

    You begin with barter examples (wheat, bullion etc.) and then move from there to paper money and on to crypto 'currencies'.

    I would start with a simple ledger, run and kept by some trusted central authority, say a central bank (OK, no surprise there).

    From there, you can move straight to cryptos. They are, after all, also defined as decentralised ledgers.

    In each case it is transactions, i.e. change of ownership of goods or services, that are recorded.

    In the first case, the central authority will want some see sort of proof that ownership of a good has been transferred, say a deed that has been signed over or a receipt. Then, the buyers' / sellers' accounts will be debited / credited according to the agreed price in the chosen unit of account. Done, no monetary medium required. Nor any computer.

    To facilitate commerce, the authority may issue its own credit tokens that can be used to pay for goods and services in decentralised transactions. Those transactions must not be officially recorded on any ledger nor do they require proof of transfer of ownership etc.. Restrictions may apply if larger sums or incorporated entities etc. are involed. Why was Trump advised not to pay off the porn star in cash?

    Anyway, it is to protect this decentralised arrangement, and thus facilitate trade, that laws protecting the owners of tokens make sense (to me).

    The emergence of modern, digital money, debit cards, e-banking etc. has basically married the security of the first arrangement with the convenience of the second. A win-win situation except maybe in places where the authorities cannot (as a matter of fact, not imagination) be trusted. But then, even decentralised token money is useless. Venezuelan cash is no better than a Venezuelan bank account.

    In come crypto-currencies as an attempt to have it both ways. The security of a ledger, at least as a means to prevent counterfeiting and theft, but without the need for a central authority. In fact, the role of the authority is transferred to the user(s). For that to work, the ledger changes ownership with every transaction. The new owner inherits, and thus becomes responsible for, the history of all previous owners just as a bank is responsible for the legality of all transactions recorded on its books.

    Now, one thing you seem to be implying is, that if the laws protecting the ownership of cash were applied to crypto-currencies, that they would become more like money in the sense that they would be more widely accepted / fungible because less risky to hold. But you then argue that cryptos do not deserve that privilege because they are mainly used for gambling, and rarley for commercial purposes. Firstly, that sounds a bit like a chicken-egg conundrum. Secondly, the other condition for fungibility, being identical, is prohibited by the fact they have ledgers built in. This might be comparable to the era you called wild-cat banking. Each bank note had the particular risk profile of its issuing authority embedded in it, no matter what the face value said.

    But thirdly and most importantly, none of this addresses the fact that token money of any kind, even wild-cat notes, and whether officially exempt from nemo dat or not (after all, the law was imposed after the introduction of money), is fairly stable in value vs. goods, whereas cryptos, unlike money, but very much like other useless commodities, are extremely volatile. Are you suggesting that with exemption from nemo dat and anonymity of the ledger, that would change? I seriously doubt that. And what is an anonymous ledger, anyway? A book that no one can read?

    1. " Firstly, that sounds a bit like a chicken-egg conundrum."

      Yes, there is a bit of that going on. If bitcoin isn't granted the privilege of currency status, it will be less likely to used as a general medium of exchange, but if it needs to have a certain level of usage as a general medium of exchange before it is granted status, it seems impossible for it to ever break into the circle.

      That being said, banknotes were already very popular before they had the privilege granted to them.

      "Are you suggesting that with exemption from nemo dat and anonymity of the ledger, that would change? I seriously doubt that."

      Yeah, I seriously doubt that too. Cryptocurrencies like bitcoin are volatile by nature, nothing will change this.

    2. Bit of a ramble there, on my part.

      I tend to agree with andrew above. A $ is a $ is a $, just as a bitcoin is a bitcoin is a bicoin. Until it isn't, because it is 'tainted', either literally or metaphorically. After that, you're dealing with stolen goods.

      That's fundamentally different in the case of barter goods such as bullion coin, where each will be judged on its specific, physical merits.

      So again, as in previous discussions, there is a clear line between barter and a token system. In the latter, fungibility, in the sense of being physially indistinguishable, is not actually a question. Or only to the extent that it applies to specific, institutional risk associated with an issuer, but never to the unit of account.

      Legal fungibility, firstly, only applies to token money and secondly, is an after the fact recognition of an established, common practice.

  6. Good post. I didn't know about that Scottish case.

    I think there might be a third type (or cause of) of fungibility: if the issuer is willing to swap a dirty old $10 note for a clean new $10 note (or swap either for two $5 notes) that will make them fungible.

    1. Yep, that makes sense to me. And as long as there is a big actor (doesn't always have to be the issuer, I suppose) who's willing to do that swap, smaller actors will replicate the swap.

  7. JP, given your hypothesis that cryptocurrencies get all their value from lottery- or Ponzi-like features, do you expect the following things to come true in the next decade?
    - Relatively more volatile cryptocurrencies will have taken away market share from relatively less volatile ones on net.
    - The volatility of a market-cap-weighted index of cryptocurrencies will have trended up or at worst stayed flat over that time, as falling volatility wouldn't satisfy the gambling appetites of market participants.

    1. "...given your hypothesis that cryptocurrencies get all their value from lottery- or Ponzi-like features"

      I guess I can see bitcoin getting at least a small bit of its value from those who value it intrinsically, say hardcore supporters who hold it as a symbol of being anti-establishment or whatnot. Some of the incessant tribalism we see in the various cryptocurrencies may be indicative of some fundamental demand too.

      "Relatively more volatile cryptocurrencies will have taken away market share from relatively less volatile ones on net."

      Dunno. Maybe. I think it would be reasonable to say that people with a gambling mentality don't usually migrate to utility stocks, they prefer to trade penny stocks and biotechs. So maybe the same applies to cryptocurrencies?

  8. JP,

    A great set of posts on fungibility. Thanks! As I've said earlier, Bitcoin is not money, but that is exactly why it helps us understand money better.

    Oliver said: "Legal fungibility, firstly, only applies to token money and secondly, is an after the fact recognition of an established, common practice."

    True! The courts protect what money does, because that's also what money is (according to a popular definition among economists). Can we do better and base legal fungibility on what money is? I think so.

    You have already touched this in your "rant" above. If banknotes, and token money in general, are (first and foremost) tokens used for recording transactions in goods and services, then their changing of hands is best viewed not as an exchange of property but as a message, or signal. We see why when we look at today's mainly accounts-based monetary system with its more centralised ledger(s). Notes being handed over from one user to another is comparable to two (or 4, 6, 8...) new entries being made on one or more ledgers. There's no change of ownership, as legal scholars have pointed out, because there's no object which could be owned.

    Individual holdings of tokens at any point in time are comparable to the balance of one's account. It's the record we're interested in, not the actual tokens.

    Casino tokens usually remain at all times property of the casino, the issuer. So it could -- and in theory should, I argue -- be in the case of (other) token money. A banknote thief should be charged with trying to cause a monetary loss to the victim, or even falsifying public records, but not with property theft. You can grab a fire extinguisher to put out a fire in a government office, but you don't own that extinguisher, nor can someone be charged with stealing it from you, as if it was your property.

    If we make it explicit that token money is not property of its holder, but part of a public recordkeeping system, then we have based legal fungibility on what money is. This would also explain why it is in many countries forbidden to burn paper money. And, for that matter, also why the issuer stands ready to replace worn out notes with new ones free of charge, as Nick Rowe points ut. The issuer is just maintaining public infrastructure.

    1. Ramble, not rant. Sorry, Oliver! :-)

    2. Who actually does hold legal title to money? I've found little evidence. In the US, it's often said that the Fed lends its notes into circulation? There's also legal disagreement on whether the Takings Clause applies to money, see for instance this:

      "...the Supreme Court has never actually spelled out that money is property..."

      Do you find anything more conclusive?

      As I suggested above, it would make a lot of sense to declare money not property of its holder but property of its issuer.

    3. I think one should distinguish between the rights / entitlements and corresponding obligations inherent in the ledger entries and the rights to the (physical) money things by the holders. The latter can easily be dismissed as you argue, but the former are quite rightly subject to public and legal debate. It is a debate about the rights and obligations of individuals vs. those of the community. Cato is typically at one end of the spectrum in that debate. They like to frame it in terms of private property rights (maybe private entitlements would be more precise) and confiscation thereof. I think it is more a question of conflicting private and common interests the solution to which is subject to the ongoing political decision making process. It's magical thinking to believe there can be a rule or law that will settle that conflict once and for all, or that there can be a state of the world in which no such conflict exists. Only hard headed deontoligists of a very selfish bent (or communists) would want to go there.

      But that debate should be separated from the question of what money is, I believe. A more fruitful line of argument might be to ask what it is that money entitles one to and why each unit of entitlement (even if limited in terms of enforcibility) must correspond to a unit of obligation. And that if one treats the money thing itself as property, on top of the entitlement inherent in the ledger entry, one is actually double counting. Furthermore, since the entitlement is often to something that doesn't yet exist (future output), it cannot really be considered property. Not even intangible property.

      That also leads to the question of what it is that holders of crypto currencies are entitled to. And whether that can pan out if no one on the other side is obliged to deliver. And, depending on the answers to those questions, whether cryptos should therefore be considered 'property proper'.

      Even more off topic. Nick Rowe has apparently retired from teaching and there is a tribute to him in the economist. The comments on WCI include a link to an older post of his named 'financial assets > liabilities'. Apart from lots of smart comments by a famous Finnish blogger, it ties in somewhat with the discussion above, I find. Also, green and red money is mentioned. I think I am slowly coming to (my own) terms with the idea. I'll try and write them down and post them somewhere, maybe in the discussion forum, if JP doesn't mind.

    4. Antti: some governments make it illegal to deface or destroy your money (IIRC). Which is certainly an attenuation of property rights, rather like owning a listed historical building. And the fact that (most?) governments will replace your notes if you can prove they were accidentally destroyed shows that physical notes are just an alternative way of keeping records. But then the same is true (I think) with bonds and stocks.

    5. Nick, thanks for taking time from canoeing and gardening to answer me ;-) And congrats on your retirement! You deserve all the praise I've seen being sent your way.

      I, too, think that what applies to physical money applies to paper bonds and stocks too. They are records, and the form is not, well... material. Proof of you having a certain (immaterial, and in many cases vaguely defined) right is what matters, and possessing a physical token is just one way to prove it.

      In your answer you assume it's the holder's money, and that makes you talk about attenuation of property rights. But a piece of paper is far from a historical building, isn't it?

      As a starting point, I think we must allow the courts and lawmakers to fail to define money in a clear and consistent way, because even economists don't agree on what money is. Perhaps many have taken it as a starting point that physical money is property of its holder. But we need to look at what they do, not what they say. And to me, our laws and rulings related to currency speak of public property, part of the economic infrastructure, which a private person is free to use and possess, but not destroy. Consider this:

      1. "Do not deface or destroy public property, even if it is of insignificant value" vs. "Do not deface or destroy certain items of your private property even if they are of insignificant value" -- which makes more sense?

      2. Wouldn't it make more sense for the Treasury/CB to replace worn and torn public property free of charge than to do the same for private property? Then it would be like a nice auto leasing deal, where the driver doesn't need to pay for maintenance.

      3. Let's say you find a nice piece of wood in a national park which you use as a walking stick. Your misbehaving friend takes it from you and throws it down the hill. I pick it up, unaware of you having held it before. You want it back, but I won't give it. It's public property, so there's no need to talk about legal fungibility. You can sue your friend, not me.

      4. As I said earlier, casino tokens usually remain the property of the casino even when they are in circulation. And those are used much in the same way as currency is, at least in Vegas?

      5. The Cato piece I linked to said that the Supreme Court has never spelled out that money is property.

      Don't you agree that it would be easier to make sense of all this if currency in circulation was considered public property?

    6. Antti: Thanks!

      If it really is public property (more generally, property of the entity that issued it), what am I doing selfishly selling it for whatever it is that *I* want, rather than using it for the public interest?

      By the way, any idea why it would be a good law to say that casino chips are the property of the casino? And that does not (presumably) mean that the casino can take them away from me, uncompensated, when I walk out of the casino doors?

    7. Nick, don't we all use public property for our own interest? Last time I walked in a park was because I like it.

      When it comes to selfishly *selling* money, there's Clower again. (I expected him to intervene!) You don't sell it. You pass it on, so that the decentralised records are updated.

      Take red money. Is it ever sold? I don't think so. Red money is incompatible with Clower.

      A buyer of goods is selling a service if he accepts red money from the seller of the goods (who is thus buying a service, like he does when he gets garbage picked up). In an accounts-based system a goods-seller with a negative balance is thus buying a service. The goods-buyer with a positive balance might mistakenly think he's selling green money, when he really is selling a garbage-disposal service. But he cannot know which way it is. Luckily it doesn't matter, because both selling green money and selling red money disposal services is *imaginary*. What's going on in reality is recordkeeping related to the sale and purchase of goods.

      (I was happy to read above that Oliver is finally warming to the idea of red money, so I included it here.)

    8. The Bundesbank and the SNB both consider notes and coins property of the holder. One can do with them as one likes. In Austria, apparently, you can burn notes up to a value of 15'000.00 euros. Above that, it's an offence. In Thailand, it's even an offence to step on a note.

      If you think of electronic money in a bank account, how does any of that translate? What do I own, if anything? The paper my statement is printed on? Does it even matter? Isn't the important feature the rights that come with the bonafide possession of whatever money thing I happen to be holding whether I'm the poprietor of it or not?

      Andy Murray has won Wimbledon twice. He doesn't own the actual trophy, though, he owns a 3/4 sized replica for his cabinet at home. That doesn't mean he doesn't own the title or that it's only worth 3/4 of a real title.

      P.S., I've posted the comment I promised above on red & green money over at the discussion board.

    9. Antti, it's an interesting question whether the government is the ultimate owner of a banknote. I don't have any answers. It's a complicated topic since some of the legal points you bring up, like defacement of coins and notes, are rooted very deep in history.

      But even if a banknote is ultimately the property of the government--i.e. citizens are granted permission to hold those notes in their wallets--we still need some rule for determining who is the official user of record when a thief steals a note and spends it. That rule can be the same rule that applies to almost all property, namely that the note belongs to the user of record just prior to the theft. Or it can be a special rule that says that the owner of record is whoever receives a note in good faith.

    10. JP, yes, it's an interesting question. I approach it from a theoretical perspective (what is money), so it isn't decisive to me whether under the current law the holder or the state is considered the owner.

      You're right about the rules around theft. There's not that big a difference between ownership and right to use. As to your 'special rule', I would say it isn't that special if this is about records, and not property. The seller sold goods and this sale was recorded. He is not in possession of stolen property. The question is: Can we amend the records so that it is as if the sale never took place? Only if the seller must have known there was crime involved on the buyer's side.

      I think it's also important to have consistency between the treatment of currency and 'book money'. In the latter there's nothing that could be considered property of the account holder (this is especially clear when one considers overdrafts; legal scholars all agree on this). A credit entry on the seller's account could never be cancelled on the basis of him having received stolen property. It could, though, be cancelled on the basis of him having taken part in a transaction he must have known to be fraudulent from the buyer's side. Thus I suggest the same treatment in the case of currency.