Thursday, March 15, 2012

Money is a transaction good, so expectations rule

(This is an attempt to talk about money somewhat philosophically and avoiding economic jargon while grappling with what monetary policy should be. It is the product of much reading of Scott Sumner, Nick Rowe, Lars Christensen, Marcus Nunes, David Beckworth and other market monetarist bloggers as well as David Glasner.)

Money is a transaction good; the only point in holding pure money is to engage in transactions: i.e. its expected swap value(s) in exchange is its only value.* (By ‘pure money’ I mean money that has no significant value except as a transaction good: paper fiat money for example.) So, use of money is driven by expectations. People take your money because they expect to use it in future transactions. You offer money in transactions because of that expectation. (And, if one money is refused, folk shift, if they can, to another that is accepted.)

There is a range of transactions over which money works as a transaction good. A particular money will only work within a subset of such transactions. For example, local taxes are usually only payable in local money. Indeed, by accepting payment of taxes in its money, a rulership can set a guaranteed range of transactions for its money. In effect, an anchor for its money: but an anchor that determines neither the range of transactions that money is used for nor its value in transactions—i.e. its (other) swap values. The ability to create such a guaranteed range of transactions for its money no doubt explains the origins of coins-as-money, but it determines neither a money’s (other) swap value(s) nor the extent of the range of transactions a particular money was or is used for. (As anyone who has had experiences such as of paying for goods in US$ in Red Square can attest.)

Anything that is used in a transaction purely for its swap value is being used as a medium of exchange. That does not, however, make it money. It is only money if it is also embodies the unit of account. Something that is used as a medium of exchange and embodies the unit of account is a medium of account and so money: something used to both quantify and pay exchange obligations.

(Money is also a store of value, but that is the least distinctive thing about it; many things are stores of value. Its role as a store of value comes from its swap value, which takes us back to it being a medium of account. Yes, we use it because of expectations about its future ability to operate as a medium of account, but that is what is distinctive about it, not being a store of value.)

The subset of transactions for which a particular money can be used will overlap with that for other moneys. The wider the expected range of transactions for which a particular money can be used, the more preferred it is likely to be: preferences which will affect the swap values of that money (either for other moneys or for goods and services).

Shifts in the quantity of money in use, and of the range of transactions over which it is used, will determine its swap values. Note that both of these are driven by expectations. Expectations will affect the gap between what quantity of a money exists and what quantity of that money is in use for transactions: that gap being the demand to hold money (what economists call ‘money demand’). Expectations will also affect the range of transactions over which a money is used: whether by affecting the level of transactions or shifting transactions to another money or to non-use of money. (A commodity money will also have the possibility of shifts to or from non-monetary uses of the commodity.)

Since (pure) money is a transaction good whose (only significant) value is its swap values, expectations about its future swap values affect whether people use it in current transactions or keep it for future use. Ironically, if the demand to hold money increases more than its supply, so it is used less for transactions, then its swap value(s) is likely to rise, increasing the incentive to hold money (particularly if its swap value[s] are expected to continue to rise). If that increases the demand to hold money without a matching increase in its supply, that will further reduce its current transaction use, creating a downward spiral in transactions (to the extent that other moneys do not make up the gap) and so in income (since, in an exchange economy, income is someone else’s spending).**

The less downwardly flexible (“sticky”) prices are, the more the drop in spending (and so income) will lead to drop in output. (Since the less prices adjust to the fall in spending, the more output will.) Conversely, if spending increases, the less upwardly flexible (i.e. constrained) output is, the more any increased spending will register in higher prices.

About credit
Given its swap value(s), money can be used for credit, for loans. In order to make offering money for credit worthwhile, lenders need to be covered for risk: the risk of downward change in swap value(s); the risk of debtor default; and the base risks in not having use of the money for the time of the loan (the time-cost of credit or what is called ‘the risk-free cost of capital’).

If institutions that receive deposits operate on less than a 100% reserve, and so make loans using deposits received, they can then apparently expand the quantity of money (since money which is credited on deposit is also being loaned out). More precisely, it is expanding the use of money (since some of the money credited on deposit is not remaining idle but its being loaned out and used in transactions[: in other words, they are acting as financial intemediaries {pdf}]). This expanded use will affect the range of transactions and how much money is being used in such. It also creates a potential instability, because of the possibility of runs on deposit institutions as people seek to withdraw deposits, some of which has been loaned out, and so is not available for withdrawal.

Clearly, interest rates are not the price of money, they are the price of credit—which includes expectations about the future swap value(s) of money. The nominal price of money is one, it being the medium of account. The price of money in terms of goods and services or other moneys is its swap value(s): the latter being its exchange rates. (Various indices, such as the CPI, aggregate and average movements in swap values for goods and services.) So, interest rates incorporate expectations about the future price(s) of money.

As interest rates are not the price of money, but incorporate expectations about the future price(s) of money, they tend to be inverse indicators of the tightness of money. (‘Tightness’ meaning the extent to which the money supply is sufficient to cover money demand without restricting its use in transactions). If the money supply is not sufficient to cover money demand without leading to its withdrawal from transactions, that will tend put upward pressure on the swap value(s) of money, so creating lower expected risk of loss of swap value(s) and so lower interest rates.

Conversely, if the money supply is more than adequate to cover money demand (and its current level of use in transactions), leading to its expanded use in transactions, that will tend to put downward pressure on the swap value(s) of money, so creating greater expected risk of loss of swap value(s) and so higher interest rates. To put it another way, tight money tends to lead to lower prices and so lower interest rates; loose money tends to lead to higher prices and so higher interest rates.

Since expectations about the future swap value(s) of money are not the only element in interest rates, interest rates are not, however, pure indicators of the tightness or otherwise of money. Which also creates problems in seeing interest rates as “the” indicator of monetary policy.

About capital and assets
Since capital is the produced means of production, it represents resources withheld from present consumption for future (productive) use. The cost of deferring resource use incorporates the risk of failure (such as misallocation or accident), the risk of adverse shifts in the value of the resources used and the base risk in not having immediate access to the resources. Which replicates the elements in interest rates. Capital is often bought via credit, but opportunity cost is enough to make interest rates the cost of capital, given these are all alternate uses of money not used for consumption. Asset prices are inter-connected, because the same deferral considerations operate across all of them, creating a continuum of opportunity costs according to level of expected risk.

This again points to the limitations of interest rates as the central transmission mechanism in monetary policy. If there was no credit, there would still be money and questions for monetary policy. Even with the existence of credit, interest rates are not magically different and separate from other asset prices.

So, the trick with monetary policy is to ensure money supply covers money demand without significantly affecting the transaction use of money. If money supply is insufficient to do so, it will generate downward pressure on spending, and so income, as money is withdrawn from use in transactions. The stickier prices and wages are, the more such fall in spending will lead to a fall in output. If money is in significant excess of money demand (and current use in transactions), it will generate upward pressure on spending which, if output cannot respond sufficiently, will generate upward pressure on prices.

Monopoly provider
Which generates two basic questions: first, should there be a monopoly provider of money? If yes, what should its policy be?

The argument against a monopoly provider is the claim that an open market in money supply will ensure balancing of money supply and demand, as competing suppliers of money will respond to shifts in money demand. This a straightforward competitive-market-is-preferable-to-monopoly argument: though that non-commodity money is a good not constrained by the cost of production provides some interesting complexities in considering an open market in money.

The argument that central banks provide greater stability as monopoly providers than would an open market in money is not a strong one, given that central banks caused the Great Depression, the inflation of the 1970s, the Japanese stagnation, the Great Recession and the Eurozone crisis.

The most obvious appeal of a monopoly provider is the power it provides the ruler; though this is somewhat ameliorated in a world of floating exchange rates. For many, this power is precisely why they are against a monopoly provider of money.

An argument for a monopoly provider is simplicity: it means there is only one (local) money to recognise and assess. This is a weak argument; people cope with assessing different brands in markets all the time. Indeed, in countries where the local currency is weak, people often show great ability to distinguish between moneys in trade.

Another argument for a monopoly provider is externalities. Money is a form of network good, in that the more people who are willing to use a money, the greater its utility. Since the state is both the dominant transactor and gains (via taxes) benefits from transactions, it has the greatest interest and capacity to manage the local money supply. In particular, its taxing power allows it to recover benefits in use of money by third parties that a private issuer of money would not. Indeed, since the state is the dominant user of local money—due to its ability to insist that its taxes and expenditure be paid in a particular money or moneys—it might be better for the state to cover the risks involved in its ability to set a dominant local money.

The preceding is a derivation of arguments for government provision of infrastructure: hardly surprising as infrastructure is typically a network good with significant externalities

Of course, whether the state should be both regulator and provider (given the inherent conflict of interest in being both regulator and regulated) is a question in itself, but the insignificance of the cost of production of non-commodity money removes one of the basic arguments against public provision. Also, the evidence on the outcomes of public provision versus strongly regulated private provision is highly equivocal.

I am agnostic on the question of whether or not there should be a monopoly provider of local money. It is not, however, presently within the realm of practical policy to abolish the various state monopolies over local money.

What policy?
Assuming we have a monopoly provider of local money (aka a central bank), what should be its monetary policy? Presumably we wish neither to strangle transactions for want of money nor flood them with an excess of it. Either way, we don’t want to “surprise” people. Forward-looking agents can manage better if expectations about swap values have some explicit anchor so that they can contract with greater confidence. Sudden shifts in swap values will lead to mismatches between contracts and prices, with potentially unfortunate effects on the level of transactions. Such massive information failure is to be avoided.

An explicit target also allows expectations to do much of the work of monetary policy through the Chuck Norris effect. An explicit target further makes a central bank more accountable (not an attractive quality to central bankers but very much so for the rest of us), and more likely to maintain consistency of policy under stress (which is good or bad depending on the range of circumstances in which the specific policy target is appropriate).

So, an explicit target is (generally) good. But do we target swap value(s) or transactions? If we target transactions, then we are targeting the function of money (to facilitate transactions). We are explicitly aiming to balance money demand and supply to facilitate transactions, accepting whatever shifts in the swap value(s) of money are necessary to do that.

If we are targeting swap value(s), then we are targeting the price(s) of money, operating as if either it is more important than balancing money supply and demand or if stabilising price(s) is the same as balancing money supply and demand. The price/swap value being either for local goods and services or its exchange rate for other moneys: for you have to pick one, you cannot do both. If you target the exchange rate, you import your average change in local swap values (aka inflation) via the set exchange rate; if you target the average change in swap value of your money for local goods and services (aka inflation), then your exchange rate(s) reflects differences between local and trading partner inflation rates.

Either way, you run the risk of sacrificing transactions in order to achieve your target: for you are targeting the average swap value of money, not the level of transactions (or income, remembering that spending=income in an exchange economy). For the policy is not directly matching money supply and demand, it is targeting price. If we target the swap value(s) of money (i.e. inflation), changes in money supply will tend to mirror changes in output, running the risk of exacerbating negative shocks rather than ameliorating them. A danger likely to be worse the more prices are “sticky” and the more people have pre-existing obligations (notably debt). (The current woes of the Eurozone are playing out this scenario; with the most inflexible and debt-ridden economies suffering worst.)

It is possible to target swap value(s) in a way that effectively means targeting income: set an inflation target that is an average over the business cycle. This means that, if output falls, money supply expands to compensate and, if output surges, money supply tightens to compensate. The effect is to stabilise aggregate income. It is clearly superior to just targeting swap values, as it compensates for economic shocks rather than (at least sometimes) exacerbating them.

In summary, money is a medium of account whose function is to facilitate transactions. Its use is dominated by expectations, so managing expectations is central to monetary policy—hence the virtue of an explicit target. It not only minimises the risk of serious information failure but allows expectations to do much of the work (the Chuck Norris effect).

Targeting swap values treat money’s role as a unit of account as dominant: as a means of calculating and quantifying obligations rather than the ability to meet them. Policy that stabilises aggregate income, by contrast, focuses on the full role of money as a medium of account used in obligations across time. Whether it is targeting an average inflation rate over the business cycle or targeting aggregate income (aka Nominal GDP or NGDP), it is the superior policy to having a fixed inflation target.

*A way to put this in economic terms is that money does not appear in any production or utility function. (From here [pdf] [via].)

** A nice expression of this in price terms is in this post.


  1. Replies
    1. Thanks :) Monetary economics does not come naturally to me, so I have to work my way through it. Glad to know I am getting somewhere.