show/hide this revision's text 1 [made Community Wiki]

Thanks to the comments and answers from Scott Carnahan and Michael Greinecker, I think that I understand it better now. I'm going to write this as a CW summary answer and also accept one of the other answers.

People often talk as if all currency is borrowed from the central bank, but that is not really true. If it were literally true, it wouldn't make sense, because there would be no way to pay the principal and interest back to the central bank. Or otherwise, if all money is borrowed, then an economy's total monetary assets stay at zero, which is not strictly impossible, but doesn't sound right.

What I guess actually happens is that the central bank both buys and sells treasury bonds. Even though this is done for interest rate stability, the central bank is perfectly happy to sell high and buy low, thereby violating conservation of money. It is also counterintuitive in the following respect: Although in the short term a high interest rate contracts the money supply, in the long term the interest paid expands it again. Nonetheless, I guess that the demand to have money to trade sustains the value of the money and keeps everyone from just buying treasury bonds at high interest. I guess here you would point to the money supply equation that Steven Landsburg posted. (It does not leap out at me that it really leads to currency stability, but I can believe it.)

Also, to get a currency started, the central bank can first buy or sell other commodities, for instance gold, so that the private sector then has money to buy treasury bonds. Another counterintuitive point (but one that doesn't bother me) is that if the central bank trades commodities at a monetary "loss", then actually it has gained those commodities. This inverted mode of gain by a bank seems to be one meaning of "seigniorage". Another meaning is any increase of the money supply from the central bank's trades, so at some level seigniorage is the main answer to my question.


Another player is the national government. Unlike the private sector, it is allowed an unlimited amount of debt. So, a second non-conservation of money is deficit spending, if in tandem the central bank keeps lowering the interest rate. Unlike seigniorage, this may be de facto non-conservation of money, but it is not de jure non-conservation of money, if the government keeps an honest account of how much it borrows. (As Deane and Michael discuss, this honesty is only really possible if the central bank is politically independent from the government budget.)

A third type of non-conservation of money is a default by a commercial bank that owes money to the central bank. But this does not look like a natural way to increase the money supply, and I don't think that it is.