Interesting ideas interspersed with nonsense - RSS - by nikhil bhatla, -
Home Archives May 2016

« Noise reduction on Epson projectors = Motion Blur - The End of LendingClub? »
Can the Bitcoin be Money at a National Scale?
May 25, 2016, 12:50a - Economics

I was at a neuroscience conference (SFN), and instead of absorbing some of the fresh science compressed into the San Diego Convention Center, my mind was elsewhere. By the third day I had had enough, so I just started walking and ended up in Little Italy.

It was November 2013, and the media had declared Bitcoin a Category 2 tropical storm, on its way to becoming a Category 5 hurricane. Bitcoin's price was $370, and within the next month it would rise to more than 3-fold its value, peaking at $1150 in December.

I had read about Bitcoin about a year before, but only with this latest round of press did I become intrigued. Two questions stood front of mind. First, ctould bitcoins actually function as real money, a currency alongside the dollar and euro, as its champions proclaimed? And second, were bitcoins a good investment, especially if they might one day function as national money?

The key breakthrough of Bitcoin, and the linchpin feature in its quest to achieve money-vana, was that it was by implementation a scarce digital good. All of my interactions with digital objects so far went hand-in-hand with copying. Need to make significant changes to an essay? Just copy it and edit the copy. Need to watch a video online? Youtube just copies it to your computer and then you watch it. Sure, some copy restriction tech has been developed (DRM), but a motivated individual can work around these digital shackles, springing the underlying digits.

So a bitcoin was unlike all digital objects that came before. It was a digital object that was not copyable, and its scarcity was enforced by the software that gave it its very existence. Bitcoin account values are tracked using a special digital ledger called a blockchain, and this blockchain is maintained and updated by thousands of computers communicating with each other via the Internet. To transfer bitcoins between addresses (accounts), the transfer must be recorded in the blockchain. To be recorded in the blockchain, computers compete to complete a computationally expensive and unpredictable task, and the first to complete this task gets to record the bitcoin transfer on the blockchain and gets rewarded with fresh, virgin bitcoins. In this way, the system reinforces its very own existence, by incenting compute time with more of the scarce digital good. By being scarce, bitcoins have a shot at being digital money.

So bitcoins seem to resemble conventional money, its advocates claim it is money, but is it actually viable as money? More specifically, is it viable as money on a national scale? It is well-known that bitcoins are used as payment in illicit transactions (guns, drugs, etc.), so it is already used as money in certain circles. But does it have the potential to be money at the level of the dollar or euro, i.e. at a national scale?

To analyze Bitcoin's prospects, the first step is to define the two key terms, "bitcoin" and "money". Bitcoin is a now famous software system that generates and tracks scarce digital tokens, allowing transfers between accounts that people interpret as payments. So that's Bitcoin - but what exactly is "money"?

I have struggled with answering this question for the past 2 years. If you look up "money" in an economics textbook, it will tell you about the 3 functions of money:

(0) Medium of exchange: "Money serves as a medium of exchange in that it is an item that people are willing to accept as payment for what they are selling because they in turn can use it to pay for something else they want." (Taylor 1998)

(1) Store of value: "Money also serves as a store of value from one [time] period to another." (Taylor 1998) This feature means that you don't need to immediately use your money once you get it, as it won't go bad or degrade from just lying around. Money retains its value. This feature implies that the supply of money must remain relatively fixed and low (i.e. scarce), otherwise the value of each money token would drop, resulting in an increase in prices (i.e. inflation). Then the token would no longer be a good store of value.

(2) Unit of account: "Money also serves a third function, providing a unit of account. The prices of goods are usually stated in the units of money." It can be hard to grapple with how this is different than money's function as a medium of exchange, so here is an example: Sometimes when I travel outside the US to another country, goods are listed in the local currency, but the shop owner will happily take dollars as payment (after applying an exchange rate). This is an example of when the unit of account is in one money, while the medium of exchange is in a different money.

Despite the fact that economics portrays these 3 features as being on equal footing, it has become clear to me that a token's function as a generally accepted medium of exchange (#0) is the key feature that transforms a token into money. Exchange is the standard way that we use and think about money (e.g. "How much does this cost?"). Sure, a token of exchange is preferred if its value isn't lost over time (#1). It is also simpler to do transactions and pay back debts if the units of account match the units of exchange (#2). But features #1 (store of value) and #2 (unit of account) are important insofar as they support feature #0, money's primary function as a medium of exchange. If a token was either a store of value (e.g. real estate) or a unit of account (e.g. number of favors owed) without being a medium of exchange, I don't believe we would recognize these tokens (real estate and favor debt) as money. Money's defining characteristic is its function as a medium of exchange, as a generally accepted token used for payments.

The economics textbook cited earlier also makes the same point, though not as unambiguously:
[Money] is the part of a person's wealth that can be readily used for transactions, such as buying a lunch or selling a bicycle. Thus, money is anything that is generally accepted as payment for goods and services, not because it is inherently attractive or appealing, but because it can be used to pay for other goods and services. (Taylor 1998)

So, are these two subsidiary features (store of value and unit of account) sufficient to promote small-scale money to national-scale money? I think the following additional features are important for tokens to become money on a national scale:

(3) Divisible: The token should be as finely divisible as practically needed, to resolve the problem that occurs during uneven exchange. For example, if I give you a $20 bill for something priced at $1, there needs to be smaller denominations of dollars available that I can receive as change, otherwise the transaction will be less likely to occur, making the $20 bill less generally acceptable as a medium of exchange.

(4) Stable value: The token should not only retain value (feature #1), its value should be relatively stable. If the token's value increases too quickly, then owners may not want to give up their tokens in exchange for goods and services, instead opting to trade with a token who's value is not changing. Likewise, if the value of the token is decreasing, sellers may not want to take the depreciating token in exchange for their wares, expecting it to be worth even less later. So for people to use a token as a medium of exchange, its value should be relatively stable. A token with high price volatility is unlikely to succeed as money.

(5) Recourse: The owners of tokens should have recourse, either legally or through another social mechanism (e.g. insurance company, credit card company), that facilitates documentation and attempted recovery of stolen tokens. If your tokens are recovered, then they are returned to you. If they are not recovered, they are lost to you, though your claim of theft remains.

(6) Private: The token should support private exchange, in which the transaction's occurrence and amount, as well as the identities of the object sold and the participants, remain undisclosed to the public.

(7) Simple to use: While the infrastructure supporting a token as money may be extremely complicated and incomprehensible to the user of money (as is the case with any sovereign currency), the actual user of money should find it simple and convenient to use. For example, if I had to do 2-factor authentication whenever I spent a digital token, that increase in complexity relative to swiping a credit card would discourage my use of the token as money. Even today with the advent of chip readers for chip credit cards, the authentication process is so much slower than swiping that I expect that the transaction rate and transaction quantity have decreased significantly.

(8) Reliable: The token system should be resilient to temporary infrastructure failure. The US dollar system is resilient: if the digital system of credit and debit networks goes down, transactions can still be conducted with the physical system of dollar bills and coins. Such a bimodal token system functions more reliably as money than a unimodal system, whether solely digital or solely physical.

(9) Interoperable: The token should be exchangeable for a variety of other national moneys, which internationalizes the token, making it generally acceptable at an international scale by proxy. This supports foreign travel and international commerce.

(10) Advantageous: If the token is introduced into a community that already has a token used as money, the new token should provide substantial advantages to overcome the inertial advantage of the preexisting money. These advantages cannot be small and must be substantially greater than the cost of switching platforms.

One final point that is worth stating is that tokens to be used as money can be created either as a form of debt/liability (e.g. most money today is a liability of the banking system) or as an asset (e.g. gold, bitcoins). World governments have converged on debt-driven moneys, so there may be practical advantages to a money supply that supports debt creation as a monetary policy. This is a more complex topic that may be worth analyzing later. For now, it seems to me that tokens generated either as debts or assets can work as money.

Having established an extended set of criteria that more accurately reflects the modern, practical requirements of money, how does Bitcoin fare?

(1) Store of value: Good, but not in a way that supports its function as a medium of exchange. Unlike sovereign money that increases in supply over time due to inflationary monetary policy, bitcoins were designed to be available, after an initial inflationary period, in a fixed, constant amount of 21 million bitcoins. A fixed supply suggests that its value will not drop, so inflation is unlikely to occur. Instead, the value of a bitcoin is expected to increase and cause deflation, which is a potentially bad property of the bitcoin in the context of exchange. By increasing in value, owners will reduce spending of bitcoin, which handicaps its ability to serve as a medium of exchange. This might make it a good investment, but bad at functioning as money.

(2) Unit of account: Good. It is easy to price goods in bitcoins, since there is a ready market for trading bitcoins for dollars and other sovereign money.

(3) Divisible: Good. A single bitcoin is theoretically divisible into 1/100,000,000 parts, which is a much higher divisibility than sovereign money.

(4) Stable value: Poor. As alluded to at the beginning of this post, the bitcoin suffers from high price volatility. In the last year the price has increased from $236 to $445, an increase of nearly 90%! In the last 2 years, the price is down from $584, a drop of nearly 20%!

(5) Recourse: Poor, though possible in some cases. You can report loss of bitcoins to the government, and in some cases, if enough people are affected, the government may pursue an investigation and recovery process, as Japan did in the case of the abrupt dissolution of the Mt. Gox exchange. In general, however, there isn't an authority or security service within the Bitcoin infrastructure that tracks and returns stolen bitcoins. Unlike credit cards and like cash, once a bitcoin is transferred to another person's address, it cannot be pulled back. Cash theft can be reported to the police and processes exist for investigation and recovery. Such public processes don't exist today for individual theft of bitcoins, as far as I know.

(6) Private: Poor. Truly private exchange is not possible with Bitcoin, as all account (address) transactions are recorded publicly across all servers storing the Bitcoin ledger, also known as the blockchain. Compare this to cash, in which a cash transfer need not be recorded anywhere, and its participants and purchased item can remain unknown. Unlike cash, every bitcoin bears its account transfer history (via the blockchain). Although the transactions are in the clear, the mapping from Bitcoin address to person remains obfuscated, and the object purchased can remain perfectly unknown. In this way, Bitcoin has the benefit of conferring anonymity despite the absence of full privacy. However, if another service manages your Bitcoin addresses, they may require that they know your identity. For example, if you want to exchange bitcoins for dollars, one convenient method would be to provide your bank account for the transfer. In this process, you may be forced to reveal your identity to your wallet service, which might get hacked or otherwise share the information with other parties. If I were the FBI/CIA/NSA, I would definitely have a project that unmasks Bitcoin addresses, linking them to individuals so I could monitor the flow of these tokens to terrorists or other people suspected of significant illegal activities. The bottom line is that if you want privacy using Bitcoin, it isn't simple, you need to be technically savvy, and you need to avoid making casual mistakes, which can be difficult.

(7) Simple to use: OK. As mentioned above, if you want to use Bitcoin as a novice, you will likely trade your privacy in exchange for simplicity. Although the infrastructure is complicated, there are many services available today that hide that complexity and provide a simple user experience for sending and receiving bitcoins. There are even Bitcoin ATMs that accept dollar bills and transfer bitcoins to your Bitcoin address, and, in the opposite direction, dispense dollar bills when you transfer your bitcoins to the ATM's Bitcoin address.

(8) Reliable: OK. If a seller cannot process your Bitcoin sale due to a technical issue (e.g. lost power, phone/computer failure), then no transaction can be completed. Compare this with the dollar, in which a credit card network failure does not prevent one from using physical cash to complete a sale. Some people have implemented physical bitcoins that are real incarnations of digital bitcoin addresses, so this is conceptually a possibility with Bitcoin as well, though as far as I can tell, it hasn't gained much traction among Bitcoin users.

(9) Interoperable: Good. With many markets in existence in many different currencies, it is possible to trade bitcoins in exchange for your local money.

(10) Advantageous: Poor. As far as I can tell, Bitcoin offers 2 distinct advantages over sovereign money. First, bitcoins have proven to be the preferred form of payment in the remote sale of illicit goods, including drugs and weapons. Despite the fact that transactions are not completely private, Bitcoin has succeeded as the platform for illicit exchange that can occur without each party ever physically meeting the other. It seems that people find paying with bitcoins a personally safer and more reliable method than meeting face-to-face. Bitcoin's second advantage is that bitcoins get transferred between accounts (addresses) very quickly, currently within about 10 minutes, 24/7. This is much faster than transferring digital dollars, which can be done by wire (takes hours during the workweek) or ACH transfer (several days). But this doesn't seem to be that big of a need today, though I suppose it could become one in the future. Of course, it is possible that faster methods to transfer digital dollars might also crop up if demanded. Another possible advantage of Bitcoin over digital dollars could be in sending money overseas. With digital dollars, you can do an international wire transfer for about $35 per transaction, and send money to nearly any international bank account. I once sent money to my brother in India this way. Bitcoin is conceivably cheaper, but it remains prone to foreign exchange risk and latency/complexity (e.g. to transfer money from your bank account to a bitcoin wallet will take a few days or require a wire and its associated fee, and getting the money back out from bitcoin to a foreign bank account will again take a few days or require a wire+fee). So it seems that Bitcoin may not be so great for transferring money internationally.

To summarize: of the 10 supporting features a token should have to function as money on a national scale, bitcoins do well on 6 of them, and poorly on the remaining 4: privacy, recourse, stable value, and advantage. Perhaps novel products and features could improve privacy and provide recourse. A stable value seems difficult given the amount of speculation and attention Bitcoin is receiving. But by far the most critical failure is in the advantage category. The need for speed in money transfers doesn't seem critical to most people, and the usage for illicit transactions seems limited to a small subset of individuals within a nation.

Could the bitcoin function in parallel to a domestic money, without the need to supplant it? Countries not in currency crisis seem to rely on only a single currency, not many, due to psychological and infrastructure simplicity. So the likelihood of the bitcoin functioning as a parallel currency seems high only in countries in which there is a currency crisis. However, I suspect that people in such countries generally will not trust bitcoin, which is hard to understand and explain, over the people that they know, who they can establish a new mutual credit system with.

So, in a sense, bitcoins are money, but they are money without a home. Without a home, the bitcoin is not "generally acceptable" anywhere, and therefore fails as practical money, leaving it as a fringe or "theoretical" money. Even if a new country is created and considers using the bitcoin as its national currency, I doubt it will adopt it (or be successful if it does) because of the lack of any control over monetary policy. In existing countries, the network effects of the domestic currency are already too strong to be overcome by or even share significant monetary space with the bitcoin.

To conclude, given the analysis above, I believe that bitcoins don't have much of a chance at becoming money used on a national scale. They are likely to remain a niche medium of exchange, the solace of drug dealers and speculators.

Investing in bitcoins may still make sense, just not because you expect bitcoins to become mainstream money on a national scale. I, for one, am not investing in Bitcoin right now.

Thanks to Rosa Cao for early email discussions that helped clarify my thinking!

  • Taylor, John B. 1998. Economics. 2nd Edition. - This was the econ textbook I used at Stanford in Econ 1, which was also taught by Taylor.

  • Read comments (4) - Comment

    Neha Narula - May 30, 2016, 3:52p
    I'm still digesting what you wrote, but I think you are missing a few things.

    Bitcoin is the first cryptocurrency, not the last, and as such, it's not a good idea to ask the question of whether it *as it exists right now* will be successful, but, whether a later iteration (potentially based on bitcoin, potentially not) will be successful. So, really, we want to know two things 1) are there fundamental problems in cryptocurrency design that cannot be fixed and 2) once its fixable problems are fixed, will it be useful?

    - #6: Have you heard of Zcash? An anonymous cryptocurrency.

    - #5: I think part of the reason cryptocurrencies have the potential to drastically lower transaction costs is exactly because there is no recourse. This increases confidence that a transaction cannot be reversed. An example: let's say I buy a very fancy ming vase from you, and then break it. Now let's say the government decides that the money I used to buy that vase was stolen. Can they reclaim it from you? Who gets made whole, and who doesn't? If there's a risk that a payment won't be final, then that risk will be reflected in the price of the item.

    I think their advantage is that cryptocurrencies will enable software-defined payments, without an intermediary. At the moment, all digital money transfer is owned by a private institution (bank, paypal, credit card company, etc). There is no open platform for the transfer of digital money.

    nikhil - Jun 1, 2016, 4:39p
    Hi Neha,

    I agree that Bitcoin can be improved on. Part of the point of my analysis was to identify areas that should be addressed, *if* someone wants to make a cryptocurrency that displaces a national currency. There certainly can be other utility asides from usage as a national money, but this is the use case I chose to focus on here.

    I think despite the potentially addressable issues with privacy (thanks for the pointer to Zcash) and recourse (perhaps through insurance?), the property of stable value seems hard, in the current speculative climate. Perhaps more importantly, I think more compelling advantages of a distributed, non-sovereign digital currency over a sovereign currency need to be clearly articulated or discovered before non-sovereign currency can displace an established sovereign currency.

    You seem to think that a lack of recourse is a good thing. While your specific scenario of stolen money is one case, I think a more common case is buying something and discovering that it is not as you expected, so you want to get your money back. eBay and Amazon implement recourse in these circumstances. If a product is not as described, then they will cover the cost, even if the seller does not accept returns. People clearly prefer this kind of guarantee over the no-recourse approach you seem to favor, even if it does increase costs somewhat.

    Another use case for recourse that seems to be a more common occurrence with Bitcoin is straight up wallet theft, due to a thief stealing someone's private key. Imagine if you came home one day to discover that your savings had disappeared, and there was no one to call who could investigate the theft and return the money. At the very least there needs to be something like FDIC insurance for Bitcoin addresses, and it would be even better if there was a specific process to investigate such theft, outside of reporting it to the local police.

    Why is it valuable to have "software-defined payments"? What does that even mean? What is the real value in having an open platform for digital money? Just because something is open doesn't mean it will inherently succeed in the market (e.g. Microsoft and Apple OS' are closed systems that dominate the consumer PC market).

    Neha Narula - Jun 5, 2016, 8:13p
    This is a good debate!

    Yeah, I don't think I care that much about Bitcoin displacing a national currency. I'm just more interested in what a widely-used digital currency can enable.

    By programmable currency, here's what I mean: There's no ubiquitous, common API for money. We have tons of digital money (credit cards, paypal, venmo, ACH), but none of it works together, or if it does, exchange between different forms is slow or incurs high fees. For example, within Europe digital payments work very well, but how hard is it to open a European bank account as an American? Why doesn't my bank work with venmo?

    Because there is no cheap, common payments platform, I think there are a lot of applications that have not been built. For example, check out some of the stuff that is pushing, like renting out your dormant resources. We still don't have micropayments. What if I could do things like easily charge people to use my open wifi network by the byte?

    You misunderstand me about recourse -- recourse is incredibly valuable, but the question is at what layer in the money stack does it occur. Ebay and Visa and whatnot make me whole again without necessarily stopping payment or recovering payment from the wrong-doing party -- it's built into their cost structure for them to just occasionally eat these kinds of things. That makes a lot of sense! But if I buy something from a stranger off craigslist with cash and it turns out it wasn't what it seems, I can't go to the federal reserve and demand that they stop the dollar bills I used to pay from circulating. Even if millions of dollars of my cash was stolen I can't do that. Bitcoin seems the same to me -- why should Bitcoin wallets be FDIC insured? Especially if there is no company/bank acting as the entity to be insured? Perhaps we could have individual FDIC insurance but what does that look like?

    nikhil - Jun 17, 2016, 10:38a
    You are making 2 points. The first argues for the value of programmable currency, which can sit on top of any currency, either fiat or crypto. The second argues that recourse isn't always necessary with any currency, as demonstrated when you pay for something using cash and don't receive what you expect to. I will address both points separately (though they are interconnected).

    First, I think the reason we don't have general programmable money is not because of an absence of tech, but because of a rejection of risk. So far, banks and governments (and individuals?) have aired on the side of relatively slow money transfer services which allow time for human intervention. This slowness lets banks satisfy anti-laundering and other regulations meant to reduce criminal activity. Additionally, it lets individuals call their bank to report a theft or accidental transfer, enabling the bank to block pending transactions that would have cleared in 2-3 days.

    In fact, at the central banking level there is programmable money in the form of the SWIFT system. There, significant thefts have already been demonstrated. Most recently, the SWIFT system was hacked at Bangladesh Bank to authorize $1B in unauthorized transactions from the Bank. The theft was limited to $81M because most of the orders were blocked, supposedly because they misspelled the word "Foundation" in the recipient's name. The orders were (likely) intentionally issued at times when Bangladesh Bank was closed for the weekend, while the Fed was still open to effect the transaction. The Fed even tried to contact the Bank on the phone, but since they were closed, they could not be reached and they authorized some of the transfers nonetheless.

    So programmable transfer of fiat currency already exists at the largest scale and carries very high risk, so I suspect that the system will undergo some improvements to reduce that risk, which will invariably cause the slowing of transactions. Cryptocurrencies do not implement any such anti-theft measures, either human review or automated, and I think these would be important to reduce theft and support adoption.

    Furthermore, in addition to monitoring for theft, there is monitoring of money transfers for regulatory reasons, including anti-terrorism and anti-laundering. These probably require review of some transactions, further slowing transfer rates. Cryptocurrencies might legally be required to implement these regulations as well, if they became more widely used. This seems difficult, and perhaps even against the anti-government spirit at the source of cryprocurrencies.

    To conclude, I don't think general programmable money, on fiat or cryptocurrency, without human approval, stands a chance for 2 reasons. First, there is too much risk of unintentional transfer, including theft and accident. Second, and related, there is a significant amount of regulation that money transfer services must comply with.

    To your second point on recourse: if you buy something on craigslist and pay cash, you assume the risk that you may not get what you expect. You can just as easily offer to pay with a credit card via paypal/venmo, and have someone else assume the risk for a fee (aka insurance). The point is that you have flexibility as a user of dollars. With cryptocurrency, you don't yet have that flexibility. And I think this flexibility is yet another feature necessary for wider adoption of cryptocurrency.

    It seems that there might be the hope of recourse with cryptocurrency after all. I'm sure you've seen that the DAO was hacked, after raising the equivalent of $150M in ether. WSJ has a brief article about it here: I thought all was lost, but interestingly the Ethereum Foundation froze the thieving account and is forking the codebase to restore the stolen ether. I'm surprised that they think they can do this, because it means that they can control the distributed network of miners. Perhaps this is true because the network remains small with limited mixing services. I doubt this could be achieved by Bitcoin, which must have many more miners, many of whom are in China and may not care about theft.

    To summarize, I think a single additional feature might be necessary for wider adoption of cryptocurrency. Cryptocurrencies may benefit from a second, "slow" coin transfer process which would occur if the transaction must undergo additional review because it triggers (a) concern of theft or (b) regulatory review. For example, transactions above a certain value would be subject to "slow" transfer, or transactions in which the sender wants a slower transaction, which might be a requirement for insurance.

    « Noise reduction on Epson projectors = Motion Blur - The End of LendingClub? »

    Come back soon! Better yet, stay up-to-date with RSS and an RSS Reader. Creative Commons License