ADDENDA: I have added in three extra paragraphs, because I came to realise a key part of the argument, and a key relevant distinction, was missing.
As economist Frank Mehrling has observed (pdf):
All monetary theories (at least all those of which I am aware) build from some underlying parable about the nature of money.
One such parable is money as pure creation of the state, or Chartalism. Its original texts are George Knapp’s The State Theory of Money (pdf), which I have waded my way through–I don’t recommend the experience; Knapp takes Germanic confusion of taxonomy with analysis to ludicrous heights–and two articles by Alfred Mitchell Innes, particularly his 1913 What is Money?
Modern Monetary Theory (MMT) is what Chartalism has evolved into. Wikipedia summarises the central claim of MMT as:
money enters circulation through government spending; Taxation is employed to establish the fiat money as currency, giving it value by creating demand for it in the form of a private tax obligation that can only be met using the government’s currency. An ongoing tax obligation, in concert with private confidence and acceptance of the currency, maintains its value.
This expresses very nicely the confusion at the heart of MMT: mistakenly holding that answering the very important question of what is the ambit of money (which transactions it can be used for) will provide some sort of answer to the question of what is the value of money (what are its swap values, its rates of exchange for goods and services).
Crazy as it sounds, taxes do not collect revenue for the government. They can’t because the government has to spend its money first to collect any taxes at all. Taxes ensure that the arbitrary, intrinsically worthless thing it spends is accepted.
This is based on historical record too: many times in ancient and recent history governors of subjugated lands would impose a tax payable in money only they possessed (eg the French in Madagascar – France wasn’t trying to raise francs in Madagascar that had none!) to ensure that local population accepts the thing in exchange for labor and goods.
The commenter is correct in that colonial governors in places which did not already have money did levy taxes in money to get people to start using money. But it is entirely possible for a state to collect taxes without money. It can levy labour-service (as Pharaonic Egypt or the Khmer Empire did–that is how the Pyramids, Angkor Wat, etc got built) or in-kind tribute or some combination thereof.
In fact, that is why, I would argue, that labour-service empires were so addicted to gargantuan building projects. Labour service has a use-it-or-lose-it nature, you cannot store it up for the future. So rulers found things for the labour to be used on; things that they decreed, thereby exercising control in their construction. Otherwise, someone else could have found use for that labour service and the ruler’s own labour-service claims could atrophy. The edifices constructed by said labour-service are statements of power; but far more so was the (continuing) control over the labour that built them in the first place.
If, however, the state wants to collect taxes in money, then the state first has to make sure that there is money in circulation. The above commenter, in citing the colonial cases, is confusing establishing the form in which taxes are levied with the actual extraction of income.
What is money?
Something is money if it is used in transactions, not for its production or consumption utility, but because it can be used in further transactions. So, something is money if it has, and is money to the extent it has, transaction utility. Transaction utility has two parts to it. One is the swap value of the money. How much money has to be used to purchase a given item. The (weighted) average of all such swap values for goods and services is the price level.
Swap values, being the price(s) of money in terms of goods and services (the price of money in money terms is itself), are generated by supply and demand. In the pure fiat money systems we use (by which I mean money in the form of otherwise largely valueless tokens not convertible into any real asset), said supply and demand is how much money is being transacted for how many goods and services. (Not, one notes, the amount of money in existence but the amount of money in circulation, in the sense of being transacted for goods and services).
In a gold standard, where any note can be exchanged for a given amount of gold, the price level is determined by how much monetised gold is backing how much output–the price of gold sets the underlying price level. (Similarly with silver in a silver standard; historically, silver has been a much more widely used monetary metal than gold–though usually as coins rather than as backing for notes.) With commodity money, swap values are determined by supply and demand for the commodity, with some premium for transaction convenience if the commodity has been suitably branded, such as being turned into coins. (A thousand years ago, all money was commodity money; now none of it is.)
Said branding provides a face value or tale, a standard content and a transaction context. In our fiat money systems, the content is about the coin or note being authentic, being issued by who its face claims it to be issued by. In commodity money systems, it is having a certain amount of the underlying commodity. Being not counterfeit means being legal tender, so having a certain guarantee of transaction utility (and convertibility, if that is part of the monetary system). In commodity money systems, having a given amount of the underlying commodity provides its own guarantee of transaction utility, in that you can swap it out at the price of the commodity. If the transaction premium is positive, it will not get swapped out. If it is negative (i.e. the price of the commodity-content is above the price-as-money), then it will. Such as having 50c coins with more than 50c of silver in them.
As for transaction context (who issued the coin and where their writ runs), that is where the question of ambit, of the range of transactions over which a given money is acceptable, comes in. Each money has a currency realm, the range of transactions for which it can be used. The supply and demand setting the swap values of money is a matter of which output operates within which currency realm.
A major complication here is that exchange rates (official or unofficial) connect moneys to each other. If one has a national money–call it US$–which is easily exchanged for local moneys elsewhere, then it may be acceptable in transactions in many countries. Particularly in countries where the local currency is less trusted (such as a store of value) than US$. This more widely acceptable money’s currency realm will therefore extend well beyond its national borders. It has been estimated that somewhere in the vicinity of 40-60% of US currency circulates outside the US.
In that situation, the US$ acts somewhat like gold in a goldzone; as a monetized store of value acceptable across currency realms. The similarity may go deeper; it has been very reasonably argued that a surge in demand for the premier global reserve currency not matched by increased supply was a major cause of the Great Recession.
Explaining why things have production or consumption utility is not difficult. How do we explain the crucial second part of transaction utility; the range of monetised transactions? That money has transaction utility at all, based on the confident expectation of use in future transactions. Particularly for moneys–such as printed notes not convertible into say gold or silver on demand–which have no other significant value. Whose transaction premium is such a large part of their swap value.
Obviously, it is very convenient to have something which can be used across many transactions. In Ancient Mesopotamia, contracts where written that specified payment in shekels of barley, wheat or silver. Units of account, such as the shekel, were often originally weights (such as the pound, originally the pound sterling or a pound of silver) because it is an easy form of standardisation. Cattle were also used as a unit of account while ancient Irish law codes used a slave girl as a unit of account, and continued to do so even when there were no slaves. Units of account make it easier to have transactions on credit. Credit greatly expands the possible range of transactions, since credit transactions do not have to be concluded immediately.
But credit relationships remain of a one-to-one nature (I owe you). Money permits immediate conclusion of transactions–goods and services for money. Things of production or consumption utility swapped for something acceptable because of its expected use in future transactions. There is no need for any ongoing connection, or any past connection, between the transactors. Money expands the range of transactions precisely because it is so anonymous and self-sufficient. Which is also why monetising personal transactions can give grave offence–as the point is precisely that the interaction is not anonymous and is thoroughly embedded in on-going connection; by contrast, a monetary transaction you can have with just anyone. That is its great strength, but makes it lacking in the personal connection stakes. (Hence giving presents rather than money; a present is a much more personal statement of connection–even a gift certificate says you know what the recipient likes.)
Legal tender laws don’t actually get us very far in explaining how modern, non-convertible, non-commodity money has expected transaction utility, as such laws they do not compel private transactors to engage in monetary transactions. Nor do they cover on-the-spot transactions. On the contrary, people are free to insist on a certain currency; or give a premium for preferred currency; or specify some other payment or refuse to contract. Legal tender laws merely specify that, within a given jurisdiction, a specified money must be accepted as payment in obligations already agreed to be monetised. In the words of economist Dror Goldberg:
… sellers are not really forced to accept legal tender money if they are slightly cautious. They only need to state in advance that they want to be paid in a different object, or use a different unit of account. The websites of some central banks are honest about this limited legal status of their money … The role of the state, after declaring what is legal tender, can be described as passive and negative: To dismiss a creditor’s lawsuit if the debtor offers the right quantity of legal tender. A legal tender law never results in the state affirmatively prosecuting a buyer or a seller for using another currency or for rejecting the legal tender in a spot transaction. Other laws might do that, but they mostly exist in totalitarian regimes.
And what is the largest set of transactions in almost any state society? Taxes. Even more to the point, they are involuntary transactions, you cannot opt out of them or out of using the object set as acceptable payment. If a state sets that its taxes must be paid in its money, then that money has a guaranteed transaction utility. Making a money’s use in the payment of taxes compulsory provides an anchor for expectations about its future transaction utility. Which does provide a good answer (pdf) to why otherwise near-valueless tokens have transaction utility–because they can always be used to pay taxes, lots of people have to pay taxes and the state can (and typically does) insist its money be used to pay its taxes.
Even in the case of US$ outside the US–they can be swapped for money that can be used to pay taxes. Why not just stick with the local money then? Well, in places with “hard” (i.e. reliable) currency, including strong property regimes, folk do. US$ are used in ordinary transactions outside the US in place of local currencies due to failings in said local currencies (small matters such as [pdf] hyperinflation and bank confiscations). That US$ can be exchanged for local currency if needed (such as to pay taxes) provides an anchor for their local transaction utility, while their use to pay US taxes is the ultimate anchor for transaction utility.
In money terms, the US economy is about a quarter of world GDP and US taxes are about a quarter of US GDP (or about 6% of world GDP). Even given that maybe half the US currency realm is outside the US, US taxes are enough to anchor transaction utility expectations about US$, but not nearly enough to set its swap values. (Particularly not outside the US.) Add in the countries which use US$ as their official currency or accept it in official transactions to the ability of US$ to be exchanged for local money-you-pay-taxes-in, and the global transaction utility of US$ is well-anchored.
[Use in taxes means that the money is, and will continue to be, swappable. What its current swap values are is a different question. The two are connected by that money's expected future swap values--i.e. its function as a store of value. Hyperinflation is regularly associated with collapsing political authority, as whether the local money will remain swappable at all is increasingly unlikely and the point at which it stops being swappable appears to be getting closer and closer. If quantity was all that mattered, that the point at which production ceases was also approaching would help protect the value of that money (as its supply would be now forever fixed). But price is a matter of supply and demand, and anchoring their transaction utility through use in taxes anchors the demand for the local money.
But it anchors the demand in only a very limited sense. For use to pay taxes grounds a money's use, but not its price. A dollar's worth of taxes is not a set price in the way a given amount of gold or silver is in a gold or silver standard, as tax liability is discharged--once you have paid it, all you have is a release from that obligation. You have no specific item able to be sold, no production or consumption utility to show for it (apart from not being liable for punishment.) Taxes are an extraction, they are not a normal transaction.
Taxes are so much not a set price, that is in part why taxes are typically levied as a percentage of income, as a percentage of a transaction, as a proportion of asset value, etc. Levying tax liability at a set money rate obviously fails to adjust revenue to total output. More to this point, in inflationary periods, it would make revenue worth less and less in goods and services precisely because there is no set price for some asset, good or service involved. The money government gains has to be spent according to market prices, which tax liability sets no automatic connection to because taxes do not set a price, they discharge an extractive liability.]
So, the MMT people are on to something. Unfortunately, they confuse questions of ambit for questions of swap value and write as if the guarantee of transaction utility sets the swap values of money, which it does not. The state may be the largest transactor, but it is not remotely the only transactor and, absent confiscations, has to buy goods and services at going prices. As hyperinflation demonstrates, anchoring a money’s transaction utility is very much not the same as anchoring its swap values, how good a store of value it is. Just as the problems with inflation targeting show that anchoring expectations about money as a store of value is not the same as anchoring expectations about the future level of transactions (i.e. expectations about income).
The taxes-guarantee-transaction-utility is a nice, consistent story. If it wasn’t for the existence of private currencies. It is entirely possible that coins were originally a private invention. During the late C18th and early C19th, privately minted copper coins circulated freely in England; a market response to the Royal Mint’s failure to produce sufficient decent copper coins and the shortage of silver coins. (Copper pennies issued by the Anglesey copper mine were rather charmingly called ‘druids’.) Nowadays, there is even the virtual private money of bitcoins.
Except that the private currencies turn out to be much less of a problem than they appear. The tax-foundation story is about explaining the transaction utility of something otherwise valueless (or, at least, whose transaction premium is hugely dominant in its swap value). Despite much economist mythologising to the contrary, there is no clear case of private fiat currency. Private moneys turn out to be convertible, or failures. Convertible to a monetary metal (typically gold or silver), to a legal tender you-can-pay-taxes-with-it money or redeemable via other assets or goods and services. States have a major advantage in production of money in that they are the largest transactors, the only significant involuntary transaction generators, said transactions operate throughout their jurisdictions and are widely known as such. All of which gives them major “branding” advantages.
The questions of the swap value(s) of money and the ambit of money are related but separate questions. Confusing one with the other is a great way to go badly wrong in monetary analysis.