Wednesday, 13 July 2016

I'm still around

You may have noticed that there hasn't been a new blog post for a long time. It's not that I lost interest in Bitcoin or economics, but my goals with respect to its economic research I more-or-less reached. I studied the concept of liquidity and how it relates to Bitcoin. My papers about the topic have been referenced by economists that I admire (Lawrence White, Robert Murphy, Walter Block). I was able to meet many other extraordinary people, in person, including, in no particular order: my dear friend Stephan Kinsella, Jeffrey Tucker, Jörg-Guido Hülsmann, Doug French, George Selgin, Rahim Taghizadegan, Frank Shostak, Philipp Bagus, David Howden, Deidre McCloskey, Nassim Nicholas Taleb, Konrad S. Graf, Roger Ver, Eric Voorhees, Andreas Antonopoulous, Tuur Demeester, Jason King, Meni Rosenfeld, Johann Gevers, Juraj Bednár, Pavol Lupták. If I forgot to mention you, it's all my fault because I have a bad memory for people. This has been an amazing experience and I'm very satisfied about it.

Last year I rediscovered Bitmessage, a project / protocol for anonymous, censorship-resistant, secure and decentralised communication. I had heard about it before but didn't realise all of the possibilities it opens. I became more and more involved in it. I joined the reference client/software (also known as PyBitmessage) and successfully iterated it through a new cycle with the version 0.6 being released in two months ago. At the moment I'm transitioning into the lead position as the original author of Bitmessage, Jonathan Warren, does not have time for it anymore. I hope that I can help to make it into a sustainable and more formally managed open source project with clear goals and path for the future. I would also like to use this opportunity to thank all contributors, translators, and other people who work on a different Bitmessage implementation (like Daniel Krawisz and Justus Ranvier).

Nevertheless, I am still interested in Bitcoin (my Bitmessage-Email gateway,, takes bitcoin payments), and the Austrian School. I think that every now and then, I can probably still make a new post about something interesting. So don't delete me from your RSS feeds yet!

Monday, 20 October 2014

My comments on the BitLicense

I spent a lot of time researching the proposed BitLicense and associated issues, and today I submitted my comments to the NYDFS. Here it is.

Dear Superintendent Lawsky,
dear General Counsel Syracuse,

kindly allow me to add my own comments to the proposed regulation Title 23, Chapter I, Part 200, henceforth “BitLicense”.


I specialise in economic research of cryptocurrencies, with emphasis on the economic theory. My activities involve publications, lectures, reviews and consulting. I have started my research three years ago. Prior to that, my professional focus was in computer networks and security, for about fifteen years, including traditional payment processing, where I was mainly responsible for implementing security policies (PCI-DSS) and disaster recovery. This combination allows me a broad insight into the types of activities and problems of cryptocurrency companies. While my own business is unlikely to require to apply for the BitLicense, several of the companies that I have contractual relationships with might.

Even though it is customary to give recommendations in comments to proposed regulation, I typically try to stay neutral. I strive to help people to understand rather than to tell them what to do. In this spirit, I hope that my comments will cause the NYDFS to become more aware of the consequences of the proposed regulation, which, according to my impression, are not well understood.

I read many of the publicly available comments to the proposal, and used some of them as input for my own comment, in order to make my arguments more complete. Nevertheless, I think that I bring new important insights, and my comment should not be simply be treated as a duplicate of other comments. Links to some of the sources that I used can be found at the end of the comment.

My comment is divided into four sections. The first one lists two issues which, in my opinion, make the BitLicense proposal unworkable. The second one lists issues which, while possible to adapt to, nevertheless cause significant hindrances for cryptocurrency companies. The third one lists issues which are comparably minor, such as omissions and unclarities. The fourth one is an attempt to ascertain the goals of the proposed regulation, its efficacy and is perhaps more “meta” in nature.

Grave issues

Affects unrelated companies

The grave issues follow from the nature of cryptocurrencies. Unlike with traditional monies and financial systems, cryptocurrencies are just numbers. In particular, private keys in the Bitcoin protocol (which I presume is what the “digital unit” in 200.2.m refers to) are 32 bytes long. For a more casual explanation, four of such keys comfortably fit into a single SMS or a tweet. 32 bytes can be stored on any object, digital or analogue, and transferred by a wide variety of means (I explain this in my master’s thesis). Once you realise this, the terms “transmission” (200.2.l) and “storing” (200.2.n.2) gain a whole new meaning. As long as this storage or transfer involves a third party, at least one of the participants is potentially subject to BitLicense. This has the perhaps unexpected consequence of a wide variety of businesses, not merely those who use cryptocurrencies in a non-financial way (as has been pointed out by others, such as Sean King), but who do not even have a cryptocurrency-specific business, being faced with BitLicense requirements. For a better emphasis, let me reformulate that. Anyone storing or transporting data or physical objects, on behalf of their customers, is potentially subject to BitLicense. Some examples of businesses that will unexpectedly be affected:
  • warehouses
  • vault providers
  • physical transport (e.g. trucking companies, car rentals, moving companies)
  • data centers, online hosting (e.g. DropBox)  data processing (e.g. email), or ISPs. If I send an email to the superintendent and attach a private Bitcoin key, Microsoft, who process NYDFS’ email, will become subject to BitLicense. If he views my email on his mobile phone, his mobile phone provider will become subject to BitLicense.
  • decentralised hosting systems like bittorrent (or newer ones like StorJ or MaidSafe). This affects all kinds of non-commercial entities who merely participate in the provision of online storage or data transfer
Companies have no good way to identify whether whatever they store or transport is or isn’t a private key unlocking a positive balance. Even if they realise that they store data that might be a private key, if the key is encrypted, they have no way of knowing the balance or what cryptocurrency it is related to.

What is, to me, surprising, is that this is not an unforeseeable problem. Other types of regulations do contain a variety exemptions, and they actually do exempt at least some of these business types. For example, federal regulation, 31 CFR 1010.100(ff)(5)(ii), has exceptions, among other things, for physical transport of cash, network services, payment processors, and facilitating sale of goods/services. California financial code, division 1.2, chapter 2, section 2010-2011 also has some exemptions. There are no equivalent exemptions in the BitLicense. BitLicense does not even exempt local, state or federal agencies, foreign governments, or the US Postal Service. These might also become subject to BitLicense. The police, if they, during exercising their duties, confiscate physical objects that store private keys (such as computers), will also become subject to BitLicense.

Even prior to cryptocurrencies, money transmitter laws already affected businesses in absurd ways. In “Regulating the New Cashless World”, professor Kevin V. Tu explains some of these problems. The proposed BitLicense makes no use of professor Tu’s analysis and only exacerbates the issue.

No way to comply with BitLicense

Even if a company realises it is subject to BitLicense and attempts to act according to it, they cannot comply with the identification requirements (200.12.a.1 and 200.15.d.1) or avoid “involving New York or a New York Resident” anyway. Once an address has non-zero balance, it is publicly visible on a ledger, and anyone can send transactions to that address, without identifying himself to anyone. The superintendent himself (being a “New York Resident”), if he desired so, could troll and send bitcoins to addresses of companies that try to exclude New York residents, forcing them to qualify their activities as “involving New York or a New York Resident” (200.2.n). The recipient cannot prevent this. If you think that I am exaggerating, similar things already happened in the past. Spammers sent small amounts of bitcoins to random addresses to advertise their products, for example the “Enjoy Sochi” or “Laxo Trade”.

The requirement to identify both of the parties involved in a transaction is akin to requiring a mail server or relay operator to identify the senders and recipients of each email. At least the mail server operator can reject an incoming email. A holder of a private key cannot prevent receiving a transaction, as required by 200.15.i. When Jeremy Allaire argued that the regulation is “technically impossible to comply with”, with other industry leaders (e.g. Wences Cesares) concurring, they were not exaggerating. My conclusion is actually that it is even more problematic than the comments of those gentlemen allege.

Significant hampering

200.8.b requires the BitLicensee to invest retained profits in a few types of US-Dollar denominated investments. It is not clear whether this prohibits retaining profits in other fiat currencies (e.g. Euros or RMB). It however excludes investing into analogous types of investments issued in other countries and denominated in other currencies. Why should BitStamp or Huobi, who are not located in the US, be forced to interact with the US financial markets? Furthermore, here we have a paradoxical situation where most of the BitLicense treats non-financial uses of cryptocurrencies as financial, this restriction treats financial uses of cryptocurrencies as non-financial. Companies that use cryptocurrencies as functional currencies, for example to pay their suppliers or employees, might get cash flow problems due to this restriction. Some companies, such as CoinBase, need stashes of bitcoins to sell to their customers quickly. This could also be potentially hampered by this restriction.

Some companies do not use fiat money at all. In the past, presented itself as having no bank accounts (however, according to Jeremy Liew, who is or soon will be on their board, this is no longer the case)., for example, is another company that, based on their business model, do not need a bank account (I do not personally know whether they do have one). Other types of businesses that do not require a fiat account are mining pools or sellers of physical bitcoin media, such as Casascius coins. Why should they be forced to obtain a bank account and/or services of a broker? What if they cannot find anyone that is willing to provide them such services?

This restriction also creates problems for companies that want to have more than 100% of reserves. According to audits published earlier this year, OKCoin, Kraken and Bitfinex were confirmed to have more than 100% reserves. This can be beneficial, for example, if the company wants to store 100% reserves in cold storage and a small amount in hot wallet. The additional reserves could also be used for other services, such as hedging or facilitating margin trading. If the company needs to liquidate excess reserves according to accounting deadlines rather than business demand, this would have negative impact on security and the provision of variety of business services.

Conversely, the requirement to hold no less than 100% reserves (200.9.a) is in conflict with certain business models (see the paper by Brito, Shadab and Castillo). It is also sometimes in conflict with other regulations, such as CFTC or SEC, as pointed out by Ryan Selkis in “Bitlicense letters #3”.

BitLicense seems to apply to certain type of intermediation services, for example escrow. This would include not only cryptocurrency businesses, but also others like notaries or lawyers. While I assume that in a typical escrow situation notaries and lawyers do identify the parties, why should they be subject to the other restrictions of the BitLicense?

BitLicense also applies to situations where encrypted keys are stored or transmitted by a third party and the holder/transmitter cannot use them in the financial sense (such as the aforementioned Why?

Companies that bring together buyers and sellers are not specifically exempt. While they probably do not qualify as “Virtual Currency Business Activity”, perhaps they should be specifically exempt.

If I travel to New York, say for a conference, companies that I have contractual relationship with might become subject to BitLicense due to my trip. Why? Are these companies supposed to track my movements? My bank does not care whether I travel to US, why should a cryptocurrency company do?

Minor issues and pointless requirements

All BitLicensees are required to have a cyber security program (200.16). This includes companies that do not deal with bitcoin electronically (e.g. sellers of Casascius coins) and in such case is pointless.

In some business models, the identity of the parties is known to another business involved in the transaction. In the case of, Amazon knows the identities of both the buyer and seller of bitcoins (it knows the credit card data of the bitcoin buyer and the shipping address of the bitcoin seller). If NYDFS wished to do so, they can obtain this information from Amazon by a court order. Why does also need to identify these two? This just makes the participants more vulnerable to identity theft.

Some companies act as an agent of the payee (e.g. payment processors). Why do they need to identify the payer? The payee can, with appropriate court order, provide the identity of the payer. During the Senate hearings in November 2013, Tony Gallippi of BitPay said that they do identify the merchant already, but as far as I know, none of the cryptocurrency payment processors identify the payer. The aforementioned professor Tu also uses the example of the agent of the payee, and the California financial code has an exemption in such as case.

It is unclear what happens with the customer's funds after revocation of license (200.6.c) or denial for people already engaged in Virtual Currency Business Activity (200.21). Is the company supposed to return them to the depositors? How much time do they have for it? Will NYDFS confiscate the deposits?

If two BitLicensees facilitate transfers between the two of their respective customers, do they need to identify each others’ customers? E.g. if a payment processor sells bitcoins on an exchange, does the processor need to know the identity of the buyer (of bitcoins) and does the exchange need to know the identity of either the merchant or the buyer of the goods or services?

Storing and transferring the blockchain (as opposed to storing the private key) is not clearly exempted, yet might fall under “Virtual Currency” (200.2.m). This may affect thousands of non-commercial entities and private persons if not rectified.

“Fiat money” (200.2.d.) excludes commercial deposit accounts (only coins and notes are legal tender) and appears to be too narrow. On the other hand, "other value" and "retail conversion" (200.2.n.4) are not defined, can mean anything and appear to be too broad.

“Transmission” (200.2.l) excludes transmission from a person to that same person. I don’t know whether this was intentional, I however think it is interesting.

Exemption 200.3.c.2 does not include the use of Virtual Currency for something else than a payment, i.e. merchants and consumers using of Virtual Currency for non-payment purposes (e.g. document timestamping) are not specifically excluded. Perhaps they should be.

In 200.4.a.13 - "an explanation of the methodologies used to calculate the value of Virtual Currency in Fiat currency" should include "if applicable". 200.19.e.4, for example, does contain "the exchange rate, if applicable". Some businesses do not provide such valuation at all, so they should not be required to explain how they calculate it.

The requirement for a bond or trust account in dollars (200.9) causes a problem for companies that do not operate with fiat money. Perhaps NYDFS should consider signing up with one of the payment processors to alleviate this?

In 200.10 (material change to business), BitLicense does not specify how long the superintendent has to approve or reject it, whereas 200.11 (change of control, mergers & acquisitions) does.

In 200.12.a.1 (books and records), "transaction" is not defined.

In 200.12.c, "non-completed, outstanding or inactive" is not defined.

Achieving goals

We all need to be aware that some of the purported goals of the BitLicense are, to a larger or smaller extent, in conflict with each other. For example, consumer protection and the requirement to conduct an AML/KYC program. If the BitLicensee is required to store personal identification of the customer, this increases the risk of identity theft. NYDFS needs to clarify their priorities. The superintendent’s remarks about not letting “a thousand flowers to bloom on the innovation side” gives us a bit of insight into his personal priorities. However, such attitude is more emotional than rational, and it is very dangerous, as explained by Adam Thierer in “Technopanics”. Jim Harper has been, for a long time, requesting a cost-benefit analysis from NYDFS, and has not received any yet.

NYDFS might consider that certain types of companies, in particular exchanges that deal with fiat, and payment processors, will increasingly tend to do AML/KYC irrespective of regulation. This is because they need good relationships with banks, and the presence or absence of AML/KYC policies at exchanges or payment processors significantly affects banks’ perceived risk.

NYDFS also does not appear to have given much merit to alternative methods to achieve the desired goals. The most obvious method is in my opinion the education of consumers (it is expected that the BitLicensees do this). NYDFS could also perform certification services of public keys or provide APIs for authenticating consumer identities, which would help BitLicensees to identify New York residents without having to store their identities themselves. In “Bitcoin Financial Regulation: Securities, Derivatives, Prediction Markets, and Gambling”,
Brito, Shadab and Castillo attempt to provide examples of many such alternative approaches.

My own impression is that, mirroring the proverb “if you have a hammer, everything looks like a nail”, NYDFS continued in doing what and how it has been doing, the result looking similar to traditional banking and money transmission regulation, and the hearings conducted by NYDFS were moot.


Peter Šurda
Vienna, Austria, October 20th 2014


Jeremy Allaire: Thoughts on the New York BitLicense Proposal,
Jerry Brito and Eli Dourado: Comments to the New York Department of Financial Services on the Proposed Virtual Currency Regulatory Framework,
Jerry Brito, Houman B. Shadab, Andrea Castillo: Bitcoin Financial Regulation: Securities, Derivatives, Prediction Markets and Gambling,
Anthony Gallippi @ Senate Hearing,
Jim Harper (on behalf of Bitcoin Foundation): comments on NYDFS BitLicense Proposal,
Sean King: Here Are My Official Comments on the New York Department of Financial Services' Proposed Bitcoin and Virtual Currency Regulations,
Peter Šurda: Economics of Bitcoin: is Bitcoin an alternative to fiat currencies and gold?,
Adam Thierer - Technopanics, Threat Inflation and the Danger of an Information Technology Precautionary Principle,
Kevin V. Tu: Regulating the New Cashless World,

Saturday, 4 October 2014

Review: BitCon: The Naked Truth About Bitcoin by Jeffrey Robinson

Since I read so much, I thought that maybe I can start posting reviews. I have already been asked by publishers to review other peoples' writings about Bitcoin, and I think I'm getting the hang of it. I take it seriously and I fully read everything that I review, and make highlights and comments. So let's start. Today I'll review BitCon: The Naked Truth About Bitcoin by Jeffrey Robinson.


While the author obviously did a lot of research, my main problem with this books is the author’s bias. The proponents of Bitcoin are, with a handful of exceptions, presented as anonymous, hysterical and associated with ad-hominems, for example: “pretend-currency” or “the Faithful”. Their arguments are ridiculed and derided, for example “I’d been hearing this made-up baloney for months, over and over again, thrown out like absolute fact with nothing to back it up”. The sources of proponents he uses are mainly comments and forum posts. Personally, what I found most outrageous was that instead of calling Jon Matonis (Executive Director of the Bitcoin Foundation, who has been researching money for 30 years and held senior positions in banks and at Visa) by his name, he refers to him as “the Bitcoin Foundation Fountain type” [fixed] and “the same chap”.
The opponents, on the other hand are presented as rational, calm, they are named including their credentials and positions, for example “Yermack's explanation is a reasoned and rational one”. Their arguments are taken as unquestionable truth. He sources them from, among other things, personal interviews and blog posts.

Bias continues

The bias is all over the place. On one hand, he argues that “... all suspects are innocent until proven guilty …”, yet that does not prevent him from claiming that Ross Ulbricht is, as alleged by the US prosecutors, indeed Dread Pirate Roberts, prior to his sentencing. When addressing Bitcoin from the perspective of Austrian school of economics, he quotes three negative opinions, and ignores many others who are either neutral or positive about it, and instead of referencing research publications, he references blog posts.
When describing the position of the Federal Election Commission towards using Bitcoin in political donations, he writes that their decisions “... speak volumes about the confidence the FEC lacks in bitcoin”. I listened to recordings of both of the FEC meetings which resulted in this decision. The debate was mainly regarding applications of internal regulations and how to technically implement them and had nothing to do with confidence.
He says that he is “... skeptical of explanations where money is involved that are too complicated for an 8-year old to understand.”, yet I doubt anyone of that age can understand how the current financial system works. It probably takes at least a 12-year old, like Victoria Grant. Many if not most adults don’t understand it either.
The economic analysis is a mixed bag. On one hand, he describes the omnipresent hype, fraud, bubbles, pump and dump schemes, and so on, and how they are present a problem. I have no issue here and even agree to a large extent, just in one occasion, he describes a pump and dump scheme and incorrectly labels it as a ponzi scheme. Another thing that I agree with him is that Bitcoin is not suitable for money laundering (for similar reasons as he presents).
On the other hand, he lacks understanding of largely uncontroversial concepts like the liquidity premium and transaction costs. Mainly he ignores the hidden costs of a trusted third party and the property rights enforcement of media of exchange. An example would be the costs associated with identity theft, which he does not mention at all. Without these, a lot of economic phenomena cannot be understood, not merely Bitcoin.
He lacks a theory of evolution of media of exchange, and appears to think that one day he wakes up and something that didn’t exist before is now a “currency”. Again, without such a theory, a lot of economic phenomena cannot be understood.
He complains in multiple places that merchants tend to use bitcoin only as a payment mechanism rather than a currency, and does not understand the relevance of this (he writes for buyers, “There is no monetary benefit … [n]or is there any benefit of convenience”). He misses that, among other things, merchant acceptance increases liquidity. When in Can Bitcoin Become a Major Currency? Lawrence White and William Luther explained the economic relevance of the usage as a payment mechanism, I thought they were joking for writing something this obvious, but apparently it’s not as obvious as I thought.
Jeffrey Robinson portrays, M-Pesa or Amazon Payments as potential competitors to Bitcoin, but misses, among other things, that they do not work internationally. Amazon Payments only works in the US, the Kenyan M-Pesa only works in Kenya. International trade is a non-negligible proportion of global trade.
The more controversial economic topics, such as deflation, velocity, intrinsic value are portrayed in a one-sided way and again the statements of the sources are taken as undeniable.
I would also complain about Jeffrey’s statist bias, but I think it probably would be unfair, because this is widely present everywhere and I don’t want to single him out.


In summary, don’t buy this book. There are better explanations of Bitcoin, and better critiques of it as well (indeed, it’s probably better if you read the original critiques he references). If you want to read more about bubbles, I recommend Boombustology, and if you want to read more about taxation and Bitcoin, read Jason Tyra’s blog.

If you like this review, send me some Bitcoins: 1MKkciz5zT4Vg8pxkd3VtAwMMcxyWPiQtQ.

Friday, 14 March 2014

The contradictions of Smiling Dave continue

Smiling Dave continues to post about Bitcoin and making up more and more ridiculous arguments as the gaps in them are exposed, and continues to distance himself from the Austrian approach more and more. However, I recently noticed something more serious: he created a new contradiction.

In All About a Medium of Exchange Having to Be in Wide Use, he writes this:
"It means a medium of exchange has to be more marketable than most things. It has to be in wide demand."
I will divide this into two parts: (1) a medium of exchange has to be more marketable than most things, and (2) it has to be in wide demand.

Marketability of media of exchange

The first part is the one where he introduced a contradiction. This is because when I previously explained to him that the regression theorem is about the emergence of the function of a medium of exchange from liquidity, he disagreed:
"Pete has done this as well in his last post, confusing liquidity with general acceptance as a medium of exchange." [emphasis added]
Maybe he does not understand that liquidity and marketability are referring to the same phenomenon? In that case we can expect his next post to concentrate on explaining the difference between liquidity and marketability.

Now, as I explained several times already, including in my thesis and earlier posts, the connection between liquidity and medium of exchange function is empirically observable with Bitcoin. It was already liquid in early 2010 when the first use as a medium of exchange was observed. It confirms both Menger's and Mises' observations. Furthermore, Menger emphasised the importance of "organised markets" (exchanges) and speculation in playing a role in the concept of liquidity. We observe these two with Bitcoin as well.

Wide acceptance and marketability

Smiling Dave has erroneously argued that somehow wide acceptance is a characteristic of marketability. However, Menger was very clear both in his definition of marketability:
"A commodity is more or less saleable according as we are able, with more or less prospect of success, to dispose of it at prices corresponding to the general economic situation, at economic prices."
There is no sign of "how widely accepted/demanded it is". Furthermore, Menger explains that the factors influencing marketability can be divided into two categories: time (when the commodity can be sold) and place (where the commmodity can be sold). Again, no sign of "width".

Wide demand is not marketability. There are goods which are widely demanded, yet they are illiquid. A very good example are houses: there is a practically universal demand for houses (apart from people who want to live homeless), yet it is difficult to sell a house easily at an economic price. You need to spend money on marketing, you might need to pay an expert for appraisal, you might need to clean it or fix it, there are often special taxes and fees and the necessity to hire a middleman and/or a lawyer to facilitate the sale, and you probably need to wait quite a while in order to get a price corresponding to the general economic situation. If you want to sell it right now and right here, you most likely need to take a huge price cut. This means that houses are illiquid. And illiquid goods cannot be a medium of exchange.

Just like there are widely demanded illiquid goods, there are narrowly demanded liquid goods. Nothing exemplifies it better than early-stage cryptocurrencies. There are very few people demanding them when they emerge. Yet because the internet is accessible anytime and from almost any place, and exchanges allow to place bid and ask offers, the logistical obstacles for the emergence of liquidity are dramatically reduced. If you want to sell a cryptocurrency, you can do it no matter where you are and what the time is, and get a price very close to the current market price. Most exchanges operate 24/7. Then there is, which helps you find people in your area if you want to sell for cash or a national bank transfer (at the time of writing claiming to have offers in 5451 cities in 216 countries). In the last couple of months, Bitcoin  "ATMs" have started being deployed (even though only those that allow to sell, rather than only buy, bitcoins, are rarer and only those count towards liquidity). Furthermore, an increasing number of shops, online and offline, accept Bitcoin as a payment mechanism. This means that holders of Bitcoins can dispose of them in increasingly easier and varied ways at a price corresponding to the general economic situation. Bitcoins are increasingly more liquid, and this means people can use them as a medium of exchange, i.e. hold them for their purchasing power.

The gap between media of exchange and money

Bitcoin shows that the level of demand for liquidity to emerge is lower than many economists have anticipated (and many still don't get). However, we need to be careful when applying this to money. The level of liquidity necessary for money may still lie very high and be unattainable for Bitcoin for all we know. While the implied assumption so far has been that there is a big gap between "nothing" and a "medium of exchange" and a narrow gap between "medium of exchange" and "money", Bitcoin shows that there are goods where the relationship is reversed: there is a narrow gap between "nothing" and "medium of exchange", and a wide gap between "medium of exchange" and "money". This is why it is important to distinguish between money and a medium of exchange, yet another difficulty many economists seem to face: they do not have a theory for media of exchange that are not money. Neither does Smiling Dave.

Tuesday, 11 February 2014

Mt. Gox and fractional reserve banking

While Mt. Gox has been long accused of running a fractional reserve system, the accusations increased during the last couple of days due to an escalation of the transaction malleability aspect of Bitcoin, and the inability of Mt. Gox to handle it on the technical, managerial and PR level. Some of the more elaborate accusals have been done by Dave Howden (firstsecond). While I don't know whether Mt. Gox is or isn't running fractional reserves, I thought I'd bring up a bit of my research on the theory of anti-FRB arguments. In an earlier version of the draft of my master's thesis, dating about two years ago (i.e. long before the situation at Mt. Gox started escalating), I had a section analysing the legitimacy of FRB from Austrian perspective. It didn't make it into the final version because my advisor argued that it is outside of the scope of the topic, so I should consider leaving it out (he was probably right). I think that it might be useful to look at the Mt. Gox situation from this perspective. The next section is quote from the draft (with some minor modifications).

Redeeming deposits on demand

de Soto (2009) argues that a deposit contract requires that the deposited item be redeemable on demand, and because it is impossible to redeem all fractional reserves on demand in all cases, it is illegitimate:

"Its [deposit contract, ed.] fundamental purpose is the custody or safekeeping of the good and it implies, for the duration of the contract, that the complete availability of the good remain in favor of the depositor, who may request its return at any moment. ... The obligation of the depositary is to guard and protect the good with the extreme diligence typical of a good parent, and to return it immediately to the depositor as soon as he asks for it." [emphasis added]
I submit that there are cases where redemption on demand is refused even during a full reserve deposit. Following is a list of some of these cases:

  1. it is outside of the opening hours of the bank
  2. the bank is undergoing an audit during which access to reserves is suspended
  3. the particular branch does not have sufficient reserves1
  4. there are technical difficulties in accessing the reserves, for example a malfunction in the lock in the safe, or the person with the key is indisposed
  5. there is an issue with verifying the authenticity of the withdrawal request2
  6. the deposits were stolen by a third party and the bank was unable to obtain enough reserves from the insurance/bank owners/other financing source on time
In none of these cases is there a fraudulent behaviour by the bank. The last three cases could be argued to be a consequence of negligence and give a rise to liabilities, however the first three are perfectly legitimate occurrences. Therefore, the impossibility to redeem the deposit on demand is not a sufficient criterion to conclude illegitimacy.

There is also the praxeological issue of differentiating withdrawal on demand and after a fixed period. The act of withdrawal is a type of action, and as such takes some time. It begins with the depositor initiating a contact with the bank, continues through the bank obtaining identification of the depositor or at least the bearer instrument he provides, verifying it, accessing their reserves, recording the withdrawal in their accounting systems and presenting the specie to the withdrawer3. During this time, the bank can liquidate some of their investments to compensate for the lack of reserves at the time of initiation of the withdrawal request. So from praxeological point of view, it is impossible to differentiate between on demand and not on demand deposits.

Applying the theory onto Mt. Gox

Based on known facts, Mt. Gox suffered from points 2, 4, 5 and 6. Their automated system for processing withdrawals is broken, they cannot verify if a withdrawal did or didn't occur, and have a mess in their accounting as a result and are re-auditing their systems. We don't know for sure if any of the deposited bitcoins have been stolen as a result of this, but previously, the US Department of Homeland Security and the US Secret Service seized (i.e. stole) about 5 million USD from Mt. Gox's bank accounts in the USA.

I'm not going to speculate whether Mt. Gox can fix this mess, but at least theoretically the problems casued by transaction malleability should be possible to audit, and at least there where account holders themselves successfully performed a double spend attack, Mt. Gox knows who they are, and can attempt legal action against them.

What can we learn from this?

I would like to introduce a saying: "Whatever bad happens with Bitcoin happens to Mt. Gox first". In other words, Mt. Gox is a lucrative target, because of its history, market share, poor management and PR skills. Attacks thus tend to affect Mt. Gox first. This generates knowledge, and entrepreneurs and users who observe these attacks, their consequences and reactions of Mt. Gox can learn from this, and not repeat the mistakes. Maybe we can argue that Mt. Gox should have spent more on management, auditing, PR or lobbying (to avoid the account seizures), but not all of this can be known in advance, and thus some level of experimenting is necessary. I don't want to present myself as Captain Hindsight. Nassim Taleb argues that we should honor failed entrepreneurs for showing us what doesn't work. Maybe we should do the same thing with Mt. Gox. After they repay everyone their deposits back.

1 An analogous situation would be an ATM running out of cash.
2 In a security breach in May 2012, Bitcoinica lost their main trading database and they had to gather the data required for the verification from other sources, which was a time consuming process that took many weeks. Recent announcement (June 13th 2012) indicated that the payouts have started, however at the time of writing (June 17th) there were still unprocessed claims. Update for this blog post: Bitcoinica entered liquidation in November 2012 and still hasn't concluded.
3 The process described is exactly the same irrespective of whether the withdrawal is done through a clerk, through an ATM or online banking.

Saturday, 7 December 2013

"I, Broken Economist": An Analysis of Gary North's economics of Bitcoin

Conceited arrogance

There is one thing that becomes apparent when reading Gary North's articles. He knows a lot. But there is also another thing that becomes apparent sometimes, in particular in his latest article on Bitcoin. He thinks he knows more than he actually knows. I can't address his article in full detail now as I'm at the Latin American Bitcoin Conference, but I thought I'd mention some core issues with his article.

Heterogeneity of a monetary system

He writes that money enables economic calculation, and thus division of labour. He writes that Bitcoin couldn't exist if money did not already exist. He writes that you cannot buy everything with Bitcoin (yet). He writes that without goods being priced in Bitcoin, it can't be money.

I agree with all of this (or let's just assume I do). I also claim that it's irrelevant. Because North does not have a general theory of liquidity, and a general theory of transaction costs. Which is very sad, because Menger was very eloquent on explaining both of these categories and made profound discoveries. People who claim that their arguments are based on Menger, yet do not have either a theory of liquidity or a theory of transaction costs do not really understand Menger.

North only has a partial theory of liquidity (a theory of stable prices) and a partial theory of transaction costs (division of labour). But Menger was very elaborate on explaining why both liquidity and transaction costs are heterogenous and cannot be summed up to a particular activity. The implied error of the homogeneity of a monetary system is visible when North writes:
"You cannot have a monetary system that does not apply across the board, yet still defend the concept of the division of labour through competitive pricing".
It is visible also in other partial implications of the alleged homogeneity, for example assuming that liquidity is a final means of payment. This already has been erroneously claimed by Smiling Dave. Final means of payment is merely one of the factors that influence liquidity. Liquidity is also not a unit of account. Unit of account and liquidity influence each other, but again are merely one of the factors.

In other words, North's critique of Bitcoin misses that there are components of liquidity and transaction costs other than those he mentions. The total mix of all these influences the choice of a medium of exchange. The weight of the result is not only different based on the evolutionary stage of Bitcoin, but also on the particular circumstances of a particular user. This is why some people in some situations will find Bitcoin more advantageous, and other people or even the same people in a different situation disadvantageous. It is also why it cannot be apriori concluded what the future of Bitcoin will be, we can only make educated guesses. The only thing we can do as praxeologists is to conclude that Bitcoin might expand in those areas where its advantages are assessed as subjectively the most important with respect to other media of exchange. It might never develop into "money", but it would be erroneous to conclude that that's the only relevant issue (the good old "money or nothing" fallacy).

Heterogeneity of social interaction

The problem that North thinks he knows more than he actually does is exacerbated by his misapplication of the system of property rights and social frameworks to Bitcoin (or the lack thereof). North does not understand that Bitcoin is a social framework. It is a more efficient social framework. Contractual relationships that are currently expensive or impossible (have high transaction costs) are now profitable and/or possible with Bitcoin. Payments are merely the first, easiest, type of a contract, on which Bitcoin demonstrates its advantages. Rather than being an "implicit denial" of contracts, Bitcoin provides a more efficient framework for them. I think that we can all agree that Bitcoin is not perfect. But there are no perfect goods. There is always the subjective assessment, imperfect information, and opportunity costs.

Bitcoin is at a very early stage, and the basis for the framework is still expecting human actors to fill it with their own activities. Contrary to North's claim that Bitcoin "put the cart before the horse", it's the opposite. Bitcoin first created a framework, and then this was incrementally use for payments.

This is also, paradoxically, why North is clueless. He understands how social institutions evolve in theory, yet he cannot connect empirical data (when it happens right under his nose) with the theory.


North complains that people who criticise his position of Bitcoin do not understand the Austrian school. Well, I know for sure that North does not understand certain aspects of it (in particular Menger's approach to liquidity and transaction costs), and on other aspects he can't connect the theory with empirical data. He's also lazy (because he did not read Austrian literature on Bitcoin and he did not gather empirical data on Bitcoin), and conceited (because he thinks his credentials give him immunity from errors).

As I wrote before, the future of Bitcoin does not depend on the understanding of economists. It depends on human action. I don't care about North's opinion. But as a researcher I see it as important that I address errors. Others than can read both, make up their own mind, and build on top of it. The Austrian school did not end with Menger, it began with him.

Wednesday, 4 December 2013

Gary North is clueless about Bitcoin


In the last couple of days, Gary North has made posts about Bitcoin on his blog:

My major irritation is that North makes two fundamental mistakes that a researcher shouldn't do. He did not review existing Austrian literature on Bitcoin, and he did not collect empirical data. Instead, he repeated long refuted fallacies, and he made up his own fictional history of Bitcoin. This is why I have said in the past that the people associated with the Ludwig von Mises Institute have become lazy and stupid. There has been progress in the meantime, however. Mark Thornton, for example, has become a fan, and the LvMI actually started accepting both donations and payments for their webshop in Bitcoin. But then Gary North pops up.

To a small extent, I am familiar with North's work. I quote him in my master's thesis, and I read his chapter from the Theory of Money and Fiduciary Media (more on that later, when I quote North based on the notes I made when I read it).

Inability to classify Bitcoin

North repeatedly argued that Bitcoin is not used in market exchanges. This is empirically false. I buy goods with Bitcoin all the time. And I was told that even Walter Block (whom I argued earlier to be clueless about Bitcoin) sold his book for bitcoins once. I don't know why North makes claims to the contrary. However, this allows him to perform a methodological trick: because he denies that Bitcoin is used in exchanges, he can avoid having to classify it within the Misesian framework for classification of goods (into consumer goods, producer goods, and media of exchange).

Instead of classifying Bitcoin as a good then, North classifies it as a ponzi scheme.

Ponzi scheme

North provides his own definition of a Ponzi scheme and a reason why it causes problems. I submit my own definition, which is in my opinion more economically useful.

A ponzi scheme is a hierarchical system of fractionally backed claims. By putting an amount of money into the system, the participant gains a claim for a larger amount of money than he put in. In order for the settlement of claims to work at the beginning, the system is built hierarchically, so that earlier participants can get money from later participants. The reason why a system like this collapses is that as the amount of debt increases, so does the risk of triggering a settlement of claims. A full settlement at any particular time is impossible: the system is insolvent. Once there are too many triggered claims, this causes a cascading wave of defaults, and the claims thus become unclaimable and worthless. In a ponzi scheme, the trigger typically happens when not enough new people enter the system on time as the old claims mature, but this is merely a special case of trigger. All kinds of other triggers can hypothetically cause the cascade.

Bitcoin is not a system of claims. Bitcoin is a pseudo-commodity, not a claim. People who purchase Bitcoins do not have a claim on anybody, nor does anybody have a claim on them. There is no debt to settle, and no default that makes Bitcoin unclaimable. The mechanism that makes a ponzi scheme to collapse is absent with Bitcoin. Whatever reasons are there for Bitcoin to succeed or to fail, the analogy to ponzi schemes is invalid.

The necessity to classify Bitcoin as a claim, in the absence of the ability to classify it as a good, follows directly from the Misesian framework for classification of goods. The difference between a good and a claim is explained by Mises several times as relevant for determinant of their price: the price of claims is derived from the price of the underlying good. Sometimes, there are other factors influencing the price of a claim, but the price of the underlying good is a necessary component at least at the beginning.

The connection between the price of a good and a claim has been expanded upon by Malavika Nair, who also has a chapter in Theory of Money and Fiduciary Media, actually precisely about this topic. I exchanged some emails with professor Nair and this is what she wrote in respect to the classification of Bitcoin:
"I agree that Bitcoin is not a money substitute [i.e. not a claim, ed.], I think of it as closer to commodity money, just not a kind of commodity most people are used to. I know Selgin has come up with the term “synthetic money” but I’m not sure if that helps clear things up or confuses them. If anything, it’s a quasi-money or secondary money, which benefits greatly from its liquidity and the ease with which it can be sold for dollars".
In other words, Bitcoin is not a claim, but a good. It is priced for its own sake, not based on a price of another good it refers to. And since it is held in order to buy goods, it is a medium of exchange.

But even if we disregard the methodological nonsense and stick with North's own empirical description of the beginnings of Bitcoin:
"The money was siphoned off from the beginning. Somebody owned a good percentage of the original digits. Then, by telling his story, this individual created demand for all of the digits. The dollar-value of his share of the Bitcoins appreciates with the other digits."
even then the description is false. Almost all of Satoshi's Bitcoin are unspent, still where they were mined. As Bitcoin didn't even have a price for almost nine months, if Satoshi had attempted to sell his coins, he would have made a revenue of ... nothing. If he had tried to sell them when the price first formed on the market, he would have earned ...  932.68 USD. There's no typo, the value of all Bitcoin in existence at the time when the price emerged was less than a grand. If he waited until Mt. Gox started operating, he would have earned ... 171,517.5 USD (assuming he actually had all of the bitcoins himself). These sums are not even profit, only revenue, as they don't consider costs. Sounds like a real opportunity, doesn't it?

The theoretical explanation of the purchasing power of Bitcoin is missing, and the empirical one is contradicted by the empirical data. Now I'll proceed to explain why North cannot provide a valid theoretical explanation of Bitcoin.

Lack of a theory of liquidity

North refers to the chapter "The Regression Theorem As Conjectural History". Luckily, I made some notes when I read the book. North quotes Menger as writing:
"The theory of money necessarily presupposes a theory of the saleableness [nowadays we use the term liquidity, ed.] of goods. If we grasp this, we shall be able to understand how the almost unlimited saleableness of money is only a special case, - presenting only a difference of degree - of a generic phenomenon of economic life - namely, the difference in the saleableness of commodities in general."
This is precisely where and why North fails. He does not have a theory of liquidity. He has a theory of "stability of prices" instead. While Menger did claim that precious metals have a higher price stability than other goods, he argued that this is not a prerequisite for how the market participants treat it. In fact, with respect to the unit of account function, in Principles of Economics, Menger actually wrote this:
"The function of serving as a measure of price is therefore not necessarily an attribute of commodities that have attained money character. And if it is not a necessary consequence of the fact that a commodity has become money, it is still less a prerequisite or cause of a commodity becoming money." [emphasis added]
If the unit of account function is not a prerequisite for a commodity to become money, then it shouldn't matter how stable it is. Liquidity is not price stability. Liquidity is, in Menger's own words:
"A commodity is more or less saleable according as we are able, with more or less prospect of success, to dispose of it at prices corresponding to the general economic situation, at economic prices."
Apples have a relatively stable price. Yet apples are illiquid: they cannot be sold easily at the market price. Bitcoin does not have a stable price, yet Bitcoin is liquid. It's not as liquid as money, and it's probably not as liquid as gold. But it it is liquid and this satisfies the prerequisite for it acting as a medium of exchange. North's grasp of Menger's insights fails.

Speculation with Bitcoin

North argues that Bitcoin is used for speculation. The issue with this argument is that it does not contradict Bitcoin being used as a medium of exchange. The motivations of human actors in this respect are not mutually exclusive. If people hold a good in the expectation of dispensing with it in order to buy something else when they need that something else, it means they are using it as a medium of exchange. If they hold a good with the expectation of dispensing with it at a higher price when the time and place are opportune, they are speculating. But speculation is a normal part of life. A retailer buys, say, clothes wholesale, with the expectation of selling them at a higher price throughout the operating hours of his business, waiting for an opportunity. He speculates on the price of clothes. But that does not mean that clothes are a ponzi scheme or that there's something wrong about it. It does not refute the validity of other reasons for buying clothes.

People in countries with a high rate of inflation tend to increase the proportion of more foreign fiat monies in their liquidity portfolio. They simultaneously hold the foreign money because it is liquid (and can be used for its purchasing power), and because they expect the purchasing power to be higher than when holding their national, faster inflating, money. This proves that the motivations are not mutually exclusive.

The regression theorem

Regrettably, it turned out that a lot of Austrians do not comprehend the regression theorem. I suspect that this is because of the difference in the approaches of Mises and Menger. Mises spends the majority of his Theory of Money and Credit by analysing the mechanism by which prices form, and the factors influencing this. Based on this, he presents a threefold classification system of goods: consumer goods, producer goods and media of exchange. Mises himself did not invent this classification, he got it from an earlier economists but unfortunately I don't have my notes on this in a searchable form so I'll update this post with the proper reference later. Media of exchange, according to Mises, differ from other goods, because they are held for their purchasing power. He then goes on to explain how purchasing power (or "exchange value", another term he uses) emerges from use value through market exchange due to differences in marketability of goods (nowadays we use the term liquidity rather than marketability). He finishes with the conclusion that there is no other way for the exchange value to emerge than through a former use value and market exchange (catallactics), and theories that do not explain the exchange value of goods, "acatallactic monetary doctrines", cannot explain the exchange value of media of exchange.

While Mises viewed exchange value and use value as two components of the final price, Menger viewed the concept of liquidity as orthogonal to the concept of price. Their goals were different. Mises' approach is helpful for the analysis of the economic calculation, and for macroeconomics (business cycle, money supply and so on). Menger's approach on the other hand is helpful of understanding the microeconomic foundations of media of exchange.

Both Menger and Mises argued that the difference between money and a medium of exchange is quantitative rather than qualitative ("presenting only a difference of degree", in fact North himself has the same Menger's quote in his "The Regression Theorem As Conjectural History"). This is why we cannot make praxeological arguments about the origin of money that also do not apply to the origin of a medium of exchange. And this is why it's evident that some Austrians still don't understand the regression theorem.

The Mengerian approach to the origin of media of exchange is that media of exchange emerge out of liquid commodities. People recognise that some commodities are liquid (it is possible to"dispose of it at prices corresponding to the general economic situation, at economic prices"), and assisted by this knowledge, they start to hold them for this purpose (to dispose of them at economic prices). This is the moment when the function of a medium of exchange emerges. The Misesian approach is similar, but he uses the term "exchange value" to explain the liquidity premium of liquid goods (which is a more complex issue). This is why the Misesian approach is more helpful when examining prices, but less helpful when examining the motivations of market participants.

As Bitcoin is a medium of exchange (people hold it in order to purchase goods in the future), it must logically adhere to the regression theorem. It must have been a liquid good before it was used as a medium of exchange. And indeed, empirical analysis shows that this is correct. Before people used Bitcoin as a medium of exchange, it was possible to trade it against the US dollars on bitcoin exchanges, which featured visible order books. This made it easier to sell Bitcoin at economic prices. The fact that the exchanges were founded deliberately does not invalidate their economic function. Menger realised that specialised services that help with sales have a beneficiary effect on liquidity:
"The institution of an organized market for an article makes it possible for the producers, or other economizing individuals trading in it, to sell their commodities at any time at economic prices."
If we go even further chronologically, before Bitcoin exchanges existed, Bitcoin already had a price and was traded sporadically (i.e. was a good). The very early prices appear to have formed based on the variable production costs of Bitcoin at that time. And even further in the past, Bitcoin did not have a price, and while there were signs of using the blockchain, it probably didn't qualify as a good.

I document the process both in my thesis, and in an earlier blog post Professor Walter Block is clueless about Bitcoin. The origin of the function of a medium of exchange for Bitcoin is right out of the book. Menger's book, that is. Not North's book. North needs to deny the purchasing power of Bitcoin, because if he admitted it exists, he'd be left without the ability to explain it, having no catallactic theory of the origin of Bitcoin.

The network effect

North is equally clueless on the network effect (which was brought up by one of his critics). He goes even further:
"I can assure you that Carl Menger, the founder of Austrian school economics, did not use language like this: "money itself replaced non-money as a market network effect good." No Austrian school economist ever has. Austrian school economists do their best to communicate in something other than programmers' professional jargon."
North is wrong yet again. One economist familiar with the Austrian tradition, Mikael Stenukla, wrote a paper Carl Menger and the network theory of money. Many more Austrians are familiar with the network effect and understand that liquidity is a subset of the more generic concept of the network effect. The article was actually pointed out to me by Peter G. Klein, the executive director of the LvMI, who also did research the network effect.

Once you comprehend the network effect, you'll realise that the issue with respect to media of exchange is the empirical question of the critical mass of liquidity: the threshold below which a potential medium of exchange needs other utility in order for the system to be self-sustaining. Economists (including Austrians), who did not comprehend the concept of liquidity, implied that this threshold can only happen at a very high level of demand (or other auxiliary criteria, such as "price stability" proposed by North). But demand is not the same thing as liquidity. Bitcoin simply shows that liquidity can emerge at a lower level of demand than previously thought.

This is why it also cannot be concluded that Bitcoin cannot become money (the same non-sequitur that already was presented by Patrik Korda earlier this year). Of course it can. It just needs to outcompete other media of exchange. And this can't be determined apriori: it's an empirical issue. As I argued in my thesis, transaction costs play a major role in influencing the choice, and Bitcoin has lower transaction costs than anything that existed before that. Even though Austrians typically do not use the term "transaction costs" (I've heard Salerno use the term "transactions costs" once though), Menger used the term "economic sacrifices" and it was evident that it's the same phenomenon. It is also evident from Menger's writing that transaction costs are heterogeneous, with many influencing factors, and that:
"Economic development tends to reduce these economic sacrifices, with the result that even between the most distant lands more and more economic exchanges become possible which previously could not have taken place."
Bitcoin is just such an economic development as Menger mentions. It allows a more efficient conduct of market operations. This is why the claim of North that Bitcoin has no utility is absurd.

The utility of Bitcoin

North's denial of utility is filled with nonsense and lacks fundamental economic analysis. It has been repeated too often by others, but it's strange that reputable researchers like North come up with this too. In order to understand the utility of Bitcoin, I recommend the recent video by Stephan Molyneux. If you don't have the time to watch, I'll just provide some points which I consider important or interesting.

The True Value of Bitcoin by Stephan Molyneux:

Bitcoin can partially or fully replace the following services by more efficient ones:
  1. Proof of ownership (obsoletes notaries)
  2. Dispute resolution (obsoletes mediators)
  3. Record auditing (obsoletes accountants)
  4. Smart property (obsoletes the police)
  5. Decentralised stock exchange (obsoletes centralised stock exchanges)
  6. Highly efficient payments (obsoletes banks, debit cards, money transmitters)
  7. Full control over your money (obsoletes central banks and banking regulators)
  8. Conditional payments (obsoletes lawyers)
  9. No inflation
  10. No business cycle
And now, here's the kicker: points 1-5 does not require that Bitcoin is a medium of exchange (because they do not require Bitcoin to have purchasing power) and points 6-8 do not require that Bitcoin is money (because they do not require that Bitcoin is used as a unit of account). Only points 9 and 10 require that Bitcoin is a unit of account. If that ever happens, that's just the cherry on the top. Some, for example Michael Suede, even come to the conclusion that without the ability to steal money, states couldn't exist.

The claim that Bitcoin has no utility is ridiculous. North really should have done some research.


It looks like I still have a lot of work to do. I thought that the research that Austrians (and semi-Austrians) did with respect to Bitcoin (in particular John Paul Koning, Konrad S. Graf, Daniel Krawisz, and of course me) would be sufficient for the other Austrians to move on and continue contributing more new interesting things. Sadly, this didn't happen yet. There are still many Austrians that are lazy, ignorant, or sadly, outright fraudulent, as they fabricate a fictional history and present it as facts. This is in particular saddening as North is a historian.

I'm not an expert in North's writings. I did find his "Mises on Money" helpful and informative (I highlighted 35 passages into my research catalogue), while his "The Regression Theorem as Conjectural History" somewhat weak. I know he has something against George Selgin, but I don't care about that. However, his attempts to address Bicoin are just annoying. I attempted to address several core problems of his articles. I hope this helps people to understand Bitcoin, and Austrian economists to finally move on beyond long refuted nonsense, and produce something new and helpful.