In reply to Eric Tymoigne: It is, of course, possible both to choose particular definitions and to quibble unnecessarily over their application without changing the substance of the matter. I take "liability" to essentially require repayment of something of value under specified circumstances. Paper money does not fit that requirement in that there are no circumstances under which I, as the holder of a dollar bill, am entitled by right to receive anything else from anybody. So, why then does paper money have a value? In part because of the bootstrap: I value it because you value it. The question then is why such a bootstrap is stable. There is more than one reinforceing mechanism. That payments of taxes are generally receiveable (in principle, not often in practice) in paper currency is one. But it is not the only one. The government's spending practices -- payments are made with checks ultimately convertible only into paper currency -- also help to underwrite the stability, as do legal tender laws (if a debt is expressed in dollars, then this paper dollar can be legally used to satisfy the debt). Mr. Tymoigne's argument that unbacked paper money is a liability of the government relies on treating the levy of taxes as a good that I buy rather than a transfer, since it is then the asset counterpart of my tax liability. When I pay my taxes I exchange my liability to the government for the government's liability to me (my paper money). But this is an odd way to think about taxes and paper money. If we want to stretch the language far enough, then I could regard the paper dollar as the government's liability, but then the counterpart asset is a weirdly contingent one. When I receive a dollar in trade, I have no idea what I am "owed" by the government for it -- that depends on when, whether, and what to what degree it chooses to tax me. The points about coins under a metallic monetary standard are, in contrast, not just matters of words. In Great Britain, gold coin was freely minted. If I brought bullion to the mint, they would turn it into gold coin at the fixed rate of 3� 17s 10 1/2d per ounce. If I owned my ounce of gold outright -- it was, in its bullion form, no one's liability -- and I own the coins once they are minted, how could we possibly regard the coins as a liability of the government? The image that Mr. Tymoigne paints is one of the government deciding that, instead of printing money on cheap paper, it will stamp it on expensive gold. If one is fiat, then so is the other. But that is a poor description of British practice, in which private parties could turn privately owned bullion into coin at a fixed rate. Even if the government bought its own bullion and stamped it, once it had used it to purchase goods, its status could not be any different from an identical coin minted by private demand. What is more, if I melted my British coins and shipped them to, say, France, I could have them reminted into francs. That can't be done with unbacked paper money. I can buy or sell paper pounds on the foreign exchange for francs, but I cannot convert the substance of a 5-pound note into franc notes of equivalent value. Nor does the value of the gold coin hinge on the government levying taxes payable in that coin. Britain could have abolished taxes or demanded that they be paid in kind, and the coins would not have lost all value -- in part because gold has non-monetary uses and in part because it was monetary in other nations and could be converted into their coin (or, as frequently happened, circulate in its original dress). This is not to deny that there are fiat aspects of coined gold. The mediaeval legal debates over whether the nominal or real value of a coin should be respected in contracts -- an interesting part of Sargent and Velde's book, The Big Problem of Small Change -- bear on the process by which a commodity becomes something different as money. And, unlike the British, many other countries charged seigniorage or brassage for minting coins driving a wedge between the value of the coin and the value of the unminted bullion. Since the coin serves as a certification of size, weight, and purity -- at least when newly minted -- it is easy to see why such differences in value could persist in the market. But certification for a fee no more turns the coin into a liability of the government than does the Good-Housekeeping seal of approval turn an oven-mitt into a liability of Good Housekeeping's publisher. And, if the coin is not a liability, then the inconvertible fiat paper money, which occupies the exact same functional space in the monetary system, is not either. Kevin Hoover