The Behavioral Economics of Bitcoin

I'm going to wade into unchartered Slips waters today and head into Bitcoinland. I've been trying to understand Bitcoin from a payment systems perspective, where it has an interesting problem and solution:  double spending.  The lesson in all of this is how Bitcoin has a sort of built in seniorage--payments are never free. Currently Bitcoin builds in its costs through inflation, which is not particularly transparent, but that will ultimately change to being more transparent--and salient-- transaction fees. By disguising its costs through inflation, rather than through direct fees, Bitcoin effectively incentivizes greater consumer use of the system, much as credit card usage is incentivized through no-surcharge rules preventing merchants from passing on the cost of credit card usage to consumers. 

It's hard to describe Bitcoin sufficiently and concisely, but here's an attempt. Cognescenti are invited to correct me; I'm still trying to understand the system. Bitcoin is a completely decentralized peer-to-peer currency and payments network. There is no central Bitcoin authority that verifies transactions, etc. This also means that there is no central for-profit manager that has to be paid to operate the system.  Instead, every Bitcoin user has a Bitcoin wallet on a personal electronic device (or webhosted). This wallet consists of open-source software that interfaces with other Bitcoin users' software.

The core problem in this system is verification of payments so as to prevent a double spend problem. To wit:  Let's say A has 50 satoshi (that's the subunit of a bitcoin). If A pays B 50 Satoshi (that's the subunit of a bitcoin), what prevents A from then paying C with the same 50 Satoshi? How does anyone know who actually has the right to those 50 Satoshi? (Notice that Remote Deposit Capture has a similar problem with repeat deposits of the same check). 

Because Bitcoin lacks a central authority through which all transactions are run, Bitcoin instead has to come up with another solution to the double spend problem. That solution is verification of the transfers through the Bitcoin mining process.

Here's how this works. When A sends bitcoins to B, A creates a message that is broadcast to the entire bitcoin network signed with A's private key that attach's B's public key to that amount of bitcoins. Anyone in the network can verify this transaction, but the verification process is costly and involves the application of a tremendous amount of computing power to check that the keys match. When successful, this verification process produces a "block," which is then distributed within the network as part of an ever-growing "block chain." This verification should show that A sent the bitcoins to B before A sent the same coins to C, so that only B's ownership of that 50 Satoshi is verified. The process of verification through block production is known as bitcoin mining. 

Mining is done by private parties--participants in the Bitcoin network--but not every Bitcoin user is a miner.  Mining takes a lot of computing power, which requires equipment and electricity. In other words, mining has costs. So while there isn't a central verification party that has to be paid, there are decentralized verification parties that have to be incentivized to conduct verification through mining.  

There are two incentives to mine:  successful miners receive a bounty of newly created bitcoins (thereby expanding the bitcoin supply and resulting in bitcoin inflation) and they may also receive a transaction fee for the verification service, if offered by the party whose transaction they are verifying. Bitcoin is designed to making mining difficult and to adjust the difficulty of the mining to the success--mining is supposed to produce roughly one "block" every ten minutes.  

Here's the catch. Right now, no one is offering transaction fees for Bitcoin mining. The sole incentive at the moment is the creation of new bitcoins. But the number of bitcoins in the system is capped at roughly 21 million. Currently there are between 12 and 13 million coins in the system; the rate of new coinage is controlled, such that it is supposed to take around 100 years before the system cap is reached. Yet as the computing demands for successful mining increase, there may well be a point at which mining ceases to be a worthwhile endeavor simply for the coin bounty. Let's call that level "Peak Bitcoinage" (see here for some discussion).  

Whenever we do reach Peak Bitcoinage, be it from hitting the 21 million ceiling or from mining costs outweighing bounty benefits, then the only thing that would incentivize further Bitcoing mining--and transaction verification to deal with the double spend problem--are transaction fees. Once those transaction fees start to appear, the appeal of the Bitcoin system should begin to drop. Right now, one of the attractions of Bitcoin is the absence of transaction fees. There are no swipe fees in today's Bitcoin. But the system is designed in a way that users will have to pay the freight either through the inflation that occurs through the bounties or through transaction fees.

There's a bit of a behavioral economics move going on here that I find interesting. The inflation currently occuring in the system is not very salient--that's the nature of inflation. A Bitcoin user never pays a fee, but the value of the bitcoins simply decreases because of the slow and steady expansion of the bitcoin supply. Moreover, the costs of inflation are borne pro rata by all bitcoin users, not simply those transacting.  This decreases the cost on any particular transactor. Currently this inflation are doubly hidden because it is offset by the deflation caused by growing demand for bitcoins, but it is still built into the system with the mining bounties.

Once bounties cease, then the inflation will cease and the costs of using Bitcoin will be more transparent, as the system cannot work unless there is verification to prevent double spends. That means users will have to pay transaction fees at market rates. As these are user-paid fees paid at the time of a transaction, they will be quite salient. They will not be spread out among all holders of bitcoins, only on those who transact, so the fees will be more concentrated. (It's not quite clear to me how those fees will get set, as transactions will be accompanied by bids without knowing if the bids will be matched to any asks in the market. In other words, A will pay B before A and B know if the transaction fee offered is high enough to verify the transaction. I suppose that they could up their bid subsequently, but in the meantime, A might have also paid C and had the second transaction verified.)  

Post-Peak Bitcoinage bitcoin transaction fees may still be favorable relative to other currency/payment systems, but I suspect that greater transparency of the costs of using Bitcoin will reduce its attraction as a payment system, much the way at-cost credit card surcharging would reduce the attractiveness of credit cards as a payment system. Bitcoin seems to have taken a page out of the credit card playbook for how to get consumers to use a payment system--disguise the costs. I don't write this as a knock on Bitcoin--I'm rather agnostic about the whole endeavor, other than to find it fascinating as an academic who studies payments. Instead, it underscores a key problem of any payment system--the network externality--and how disguising costs (deliberately or just by function of system design) is key to encouraging adoptions to overcome and then leverage the network externality. 

Comments

[Yona]

Hi there, just a clarification that may help with your analysis. There is a "transaction fee" when you transfer bitcoins. You chose that amount yourself (and can make it zero). But this money you are willing to spend will decide the speed at which your transaction is verified by the network.Right now a BTC transfer can take 10min or 10hrs depending on what you pay to get it done. The issue is then how do you expect to run a normal economy when the small transactions (buying a coffee in the morning) are those that you need to do quickly while the large ones (getting your salary paid on your wallet) are those you are not in a hurry to get. Are people really going to spart paying 1% on each transaction, this seems to me close to what they pay Visa now!

[cig]

"the verification process is costly and involves the application of a tremendous amount of computing power to check that the keys match"

This is false, verification and checking the keys is cheap in computing power, the tremendous amount of computing power is just an artificial requirement introduced to produce the "synthetic gold" effect.

The mining calculation has no point other than proving you spent hardware and electricity on doing a pointless calculation.

The arbitrage condition is that 1 BTC = cost of total network hashing power (in the last 10 minutes) divided by mining reward (per block).

You can estimate the dollar price of the total network hashing power from the price of mining microchips and electricity, weighted by how many people get involved, and that gives you the BTC/USD rate. From there you can see how that produces a pyramid-scheme style effect as long as new people join the (mining) party.

Look up "proof of stake" currencies for variants of bitcoin that don't involve doing the pointless mining calculation.

[Marco Crispini]

@Adam: I think you've misunderstood this part as it's incorrect: "Right now, no one is offering transaction fees for Bitcoin mining."

The standard client always forces at least a minimum fee; might have changed recently but I don't think it has. There are clients that allow you to send zero-fee transactions, but most people do use a fee ['most' is anecdotal - I don't have time to pull stats from the blockchain to measure this]

In my own example, I generally DONT include a fee if I pay from one of my own wallets to another one. This isn't just a trust thing, but also a timing thing. My zero-fee transaction often doesn't get included in the next block. Or the next one. Or the next one...I dont care because I figure it'll get into the blockchain at some point. The reason is miners have long been prioritising the highest-fee transactions in any given block they are hashing for. Lower-fee/zero-fee tx's get bumped onwards.

I DO include a fee if I'm doing a 'real' tx, eg buying something. Or especially transferring into my exchange account, because I can't sell those btc until they've had 6 confirmations (6 blocks old).

To summarise, here is *not* the catch; the appeal of Bitcoin is there with the fees; when fees are included they are nominal, nowhere near 1% even on a coffee.

@cig: if you think solving the no-trusted-third-party problem is pointless, then I don't think you understand the social aspects of money.

[cig]

@Marco, yes, solving the no trusted third party problem would be interesting, but Bitcoin's proof of work based system, while a novel attempt, fails here as the effective structure of Bitcoin does still allow for concentrated political power, and indeed one can argue reinforces it, as naive people who have blind faith in a deficient algorithm can easily be mislead into surrendering control without realising it.

[Pete Semile]

You should give proper credit to Matt Levine. You obviously got almost all your ideas from the article he wrote on Bitcoins a week earlier. As an academic and lawyer you should know better.

[Adam]

"Pete": Plagiarism is a hanging offense in my profession. Falsely accusing me of it is defamation. I do not take that lightly.

Your claim is false. Matt Levine wrote an article on Bloomberg on the costs of Bitcoins on Jan. 2. It's a good piece that makes some of the same observations that I do. I only read the Levine article on Jan. 12, a day after my Credit Slips post. (I'm happy to share my browser history with anyone who's interests). I didn't know of Levine's piece when I wrote. That we made the same points only indicates that we were making the points that any reasonably smart person with some payments and behavioral econ background would have if considering Bitcoin.

Moreover, why on earth would I rip off someone else's idea? What possible motive would there be? Plagiarism has no upside for an academic. My publication record shows that I'm hardly short on ideas (whatever you may think of their quality). I've always been happy to give credit when it is due, as shout-outs to folks like Yves Smith and Felix Salmon attest. If I had known of Levine's piece, I would have linked to it, not least because he's got some cool data.

You owe me an apology for assuming a convergence of insight was plagiarism. It was pretty stupid and insulting.