You cannot make loans denominated in Bitcoin (because the price is going up too fast to ever be paid back, and if it ever stops going up, people will be dumping it too fast for an interest rate) so there's no basic demand for the currency at a given price level (to pay back loans). People who want to use the beautiful transaction-settlement mechanism do not have any motive to thereby hold coins, and that means everyone holding the coins is doing so in an expectation that somebody else will buy them at a higher price. Bitcoin is pure instability, pure feedback (in either direction) of the velocity of money into the inverse price of money. I wish I had time to write an article about what an awful, terrible, no good, very bad idea it would be to try running a planet on Bitcoin.
If anyone does have a good idea how to implement NGDP level targeting in an e-currency, please say so or write me at yudkowsky at gmail. In fact, just write me if you can figure out how to measure NGDP in an e-currency in a way that can't be hacked and can distinguish more widespread adoption from increasing RGDP.
Then there's the problem that bitcoin is an international currency and GDP is a measure of national economies. As far as I know, Sumner does not have a theory for optimal monetary policy for a currency shared by a coalition of nations (i.e. the EU). The management of the Euro roughly stabilizes NGDP in Germany, but not in Italy and Spain which is the source of the current crisis in Europe.
On the other hand, an old-fashioned monetarist coin (constant rate of increase - "Friedman Coin") would be easy to implement and a lot better than a fixed supply coin. I'll leave others to figure out the "Sumner Coin".
Who gets the extra Friedman coins? Also if you don't have a central bank stabilizing the supply of money via interest rates in response to velocity changes things can go to crap very quickly a la the Great Depression.
The problem with NGDP is the D. As far as I know, there is nobody (including Sumner) that believes the total nominal transaction volume of an arbitrary subgroup of international users should grow at some low constant rate over time.
Remember that MV = PY, Money Supply times Velocity equals Price Level times Real GDP. PY is nominal GDP, Y is real GDP. AFAIK, Sumner's followers argue that M should grow so that PY grows at some constant rate. V is assumed to be endogenous.
Now think of two countries with similar sized economies that share a currency. Imagine country 1 had Y growth of 3% and country 2 had Y growth of -1%. How should M change? As far as I know, this is virgin territory. Also, country 1 and 2 are basically Germany and Greece.
NGDP is easy to calculate for the users of bitcoin as every transaction is encoded in the blockchain. Of course, you are right that there is a difficulty because of a rapid unstable increase in V due to adoption, the NGDP of the Coin economy is measuring the economic activity of its users + some amount of canibalization of the legacy dollar economy. If you are looking for the "right" answer, I think it will elude you. But I have a "good enough" proposal for a Sumner Coin - start with the BTC reward scheme, and after PY stabilizes (i.e. changes less than 10% per month), use that as the base for PY and then set block rewards to grow PY by a constant amount. V is public so it is easy to calculate the appropriate M.
The Friedman coin is much simpler, since you are pegging M to a growth rate and not PY. Of course, the miners get the extra coins.
Both models decrease first mover advantage and relieve the need for transaction fees for stability of long-term mining. And for both models I would recommend an initial deflationary period to sweeten the pot for early adopters.
But it's more interesting how the Bitcoin monetary policy compares to the current real-world human-controlled monetary policies of government money - which in the cases of most countries is definitely pure instability and filled with error.
The Bitcoin monetary policy doesn't have to be perfect to be an improvement, because the status quo is very imperfect.
I'll give you 1 BTC on April 1st, 2013. You can repay me 1.5 BTC on April 1st, 2014, adjusted for deflation or inflation according to mtgoxUSD × CPI, as the case may be.
Problem solved. I give you spending power today, you give me more spending power later on, and we conduct the transaction in Bitcoins.
In the future, we will have our choice of price indexes that are denominated in Bitcoins directly.
On a planet with a growing economy and a growing population, Bitcoins will always be appreciating faster than material investments can grow. If you can plant 100 seeds to get back 103 seeds (3% material interest) and also the planetary population is growing at 2%, that's 5% interest on Bitcoins. Except then people try to hold more Bitcoins since they grow faster than seeds, which makes Bitcoins appreciate even faster... but even before taking that into account, you still wouldn't be able to grow enough seeds fast enough to pay back a loan in BTC, if the planetary economy used Bitcoin.
The deflationary process makes things much worse; the more people try to hold Bitcoins, the fewer circulate and the lower the velocity, increasing the apparent price of Bitcoins. This process continues until instability in the price of Bitcoins as people occasionally dump their hoards is so great that Bitcoins stop appreciating. Then fewer people want to hold Bitcoins and they dump their Bitcoins and the velocity goes up and the price goes down and the bubble pops.
(The reason this doesn't happen with existing central bank currencies is that central banks regulate interest rates to stabilize the supply of money and hence its price in response to velocity changes, or at least they're supposed to. The reason this doesn't happen with equities is that they have an intrinsic rate of return due to dividends or stock buybacks; real estate has intrinsic rent; bonds have returns denominated in stabilized money.)
So like I said, Bitcoin is a horrible crackpot theory of money as a store of value, married to a beautiful transaction settlement mechanism.
I am happy to offer you the loan, because Bitcoin prices are volatile and there is no guarantee if I keep my Bitcoin in storage that it will still be worth much one year from now. There have been many crashes. But if you agree to repay me an amount equivalent to 1.05x my current spending power then I know whatever you pay me I'll be in a better position than I am today in terms of my ability to pay my rent.
Such a loan offers me a drastically different risk profile than holding Bitcoins. If BTC were as guaranteed to increase as you say then they would cost more.
This is a loan denominated in dollars, it's just being transaction-settled via Bitcoin. The medium of account is dollars and the medium of exchange is Bitcoin. This is pretty much exactly the original point I was talking about. :)
People earn to consume. The bitcoin rich, if they apply the Kelly criterion to their own portfolios, will diversify into whatever assets will provide them a regular real dividend. That is, farms yielding grain, orchard yielding fruit, NG wells yielding NG, wind farms yielding electricity. They will not be buying any investment for the purpose of money, but for the purpose of CONSUMPTION.
There lies the negative feedback that will stabilize any new currency as it conquers the world.
I'm actually not totally sure if NGDP-LT is optimal in a free-banking scenario though. But certainly some sort of value stabilization is.
Is it implausible that these two forces would reach a relatively stable equilibrium, like stocks?
What ultimately sets the price of stocks (and prevents undue speculation) is the profit earned by a company.
This mechanism doesn't exist in currencies, where the value comes only from what people think it's worth.
Nobody does. Now why is that?
Friedman understood perfectly well that failure to increase the money supply in response to decreasing velocity was responsible for the Great Depression. He's sort of famous for that part, if I understand economic history correctly, it's why Bernanke apologized to Friedman's memory on behalf of the Fed. It seems odd to call them 'Friedman coins' when they depreciate automatically, but don't adjust to velocity and especially to changes in fractional reserve lending.
Probably executive order 6102 in 1933. By the time the government allowed private ownership of monetary gold again, it had fallen out of use as currency.
This solves the problem of decreasing mining incentives and the possibility of a too-big windfall for early adopters. It's not unrealistic that Satoshi could own 2% or 3% of all the money that will ever be made. That's a problem.
Not really. It might offend someone's sense of fairness, but it hardly has an impact on whether the currency will achieve adoption.
It's really only behavior on the margin that matters.
Such a coin has been created. Freicoin is a bitcoin-variant that devalues at a constant rate. It hasn't caught on, probably because people prefer currency that doesn't lose its value.
"It's not unrealistic that Satoshi could own 2% or 3% of all the money that will ever be made. That's a problem."
He owns 2-3% of all the money at one point in time. He can only spend that money once. Money circulates, and can only give value to its user when they circulate it.
Given a given unit of money will circulate indefinitely, the relative economic power of a holder of a particular percentage of the money supply at a particular point of time will decrease over time as a percentage of total economic power exercised throughout the currency's existence.
When that happens, then Bitcoin will maintain a steady price and it will serve its purpose as an efficient medium of exchange on the Internet without the need of a central authority. Until, of course, something better comes and replaces it.
Deflation is easily negated by the natural rate of inflation, which will always occur simply because of supply/demand economics. Natural inflation is actually healthy for the economy. Artificial inflation forces people, businesses and governments to depend on lending a lot more than on the money they earned. Note that lending will still continue with natural inflation, but the interest rates won't have to be kept artificially low (like they are now).
This being the theory that "no one is going to spend Bitcoin if it keeps increasing in value, since they will sit on it instead, for the future gains. Therefore the currency is deflationary."
Amazing how this rationalization, given to slaves to justify their slavery, ends up being touted by the slaves themselves.
Technically the same is true of the U.S. dollar: If you put your dollar in a savings account, instead of spending it, you will earn interest. Therefore you will have more money in the future, from the interest earned, if you don't spend that dollar now. Therefore the dollar is also "deflationary."
The current US inflation rate is around 2%, so unless you're making more than 2% interest on your money every year, your purchasing power is going down. For point of reference, yield on a 1yr treasury bill is around 0.14%.
Bitcoin is deflationary because, once the coin limit is hit, the number of available coins will decrease over time simply due to coins being lost. As the currency contracts, the purchasing power of a coin goes up over time.
In a world where BTC had an inflation rate of -2%, 1 BTC now would buy what 0.98BTC will buy in a year. Any investments you make now need to yield in excess of 2% per annum for you to be better off. In an inflationary world, an investment only needs to beat the T-Bill rate for you to gain (or least be less worse off), which allows for less risky investments.
Very few people forgo buying something they need because they could get 2% one year later if they hold on to the currency.
I think deflation IS economically inefficient, but the negative effect of slight deflation has been exaggerated in the popular economic literature.
For a peer-to-peer currency network, a fixed money supply could even be beneficial by creating an incentive for bitcoin holders to invest in the technology.
Bitcoin is working, has created the most powerful distributed supercomputer in the world, and is processing an increasing number of instant global transactions every month.
It works in practice, as a medium of exchange, and that is a better argument for it working as a currency than theories about how deflation affects economies.
But let's grant that the Great Depression, along with the Long Depression, was during the gold standard's reign. That's two serious economic contractions interspersing mostly long periods of expansion. Compared to the last several decades, which has been marked by several 'lost decades' for the Middle Class, the gold standard era doesn't look bad.
I'm a lot more intrigued by Ripple I feel they try to solve the correct problem, though there are some issues left. The most glaring one being that it lacks a unit of account of it's own but instead pickyback on a different unit of account. It is also a harder to explain, especially to people that believe in hard money since it is basicly a stripped down fiat system.