Thursday, June 2, 2011

Money and expectations

I recently listened to talk by a former Chief Economist for the Reserve Bank of Australia. He was speaking to a lay audience and gave a nice rendition of the basic intuition of monetarism, presenting its central idea as being based on the Fisher equation of MV = PT and, assuming V (number of transactions money goes through in a given time period: often converted into k = 1/V or the proportion of money held as cash) and T (total number of transactions: often simplified to total production of goods and services in a given time period, ‘y’) do not change much, then a rise in M (money supply) will lead to a rise in P (price level).

Basically, in monetarism, it is all about the quantities. Hence monetarism relies on the notion of ‘long and variable lags’ to make this quantity story work: in other words, it invokes a ‘fudge factor’.

Monetarism does accept that a short run effect of rising money supply can be a rise in production (as discussed below) but, in the long run, the rate of inflation is determined by the rate of money expansion in excess of the rise in production, but with "long and variable lags".

The historical evidence on the fundamental claims of monetarism is not as cooperative as it might be. As one economic historian notes:
From my own studies of monetary and price history over the past four decades, I offer these observations, in terms of the modernized version of Fisher's Equation of Exchange, for the history of European prices from ca. 1100 to 1914. An increase in M virtually always resulted in some degree of inflation, but one that was usually offset by some reduction in V (increase in "k") and by some increase in y, especially if and when lower interest rates promoted increased investment. Thus the inflationary consequences of increasing the money supply are historically indeterminate, though usually the price rise was, for these reasons, less than proportional to the increase in the monetary stock, except when excessively severe debasements created a veritable "flight from coinage," when coined money was exchanged for durable goods (i.e., another instance in which an increase in M was accompanied by an increase in V).
The quantity story is not enough for several reasons, starting with the flexibility of supply mattering (as monetarism itself acknowledges). For example, in the C16th and C17th, Asian goods coming into Europe was an upward supply response that would have dampened the inflationary effect of the Central European and American silver flows and the various debasements of coinage (notably the English debasements of 1526, 1542 and 1553 – the total reduction in silver content of coins reaching 83%, though Elizabeth’s recoinage of 1560 reduced the loss to 25%, a further debasement in 1601 bringing the total loss of silver content to 36% – while the silver coinage of the Southern Low Countries was debased 12 times from 1521 to 1644, totalling a 49% reduction in the silver content of coins).

But if the responsiveness of aggregate supply matters, that gets in the way of telling a quantity story. Hence the “long and variable lags”.

The quantity story is, however, also not enough because one has to look at people’s expectations, for that will affect their holding of money (i.e. k) and so how much money is in circulation (i.e. the level of spending: Py). The upward responsiveness of aggregate supply of goods and services to any increase in spending determining how much the response is in prices (P) and the downward responsiveness of prices determining how much any fall in spending is reflected in falling production (y). (Or, to put it another way, whichever of prices or production is more constrained will lead the effect of any change in spending to be greater in the other.)

Institutions also matter: what forms of money there are, how available credit is. So, changes in institutional structures will change the significance of various monetary aggregates. One of the notorious problems of monetarism is trying to work out which monetary aggregates to follow and when. Money is a tool of human action, so what happens to it and what it does is about human behaviour which change as institutional structures change and vice versa: a quantity story is never going to be enough.

Money complexities
It is easy to go wrong in thinking about money as it is so central to so much economic behaviour. One of the basic things money does is that it simplifies. Without money, people are stuck with barter price(s) – price in terms of goods and services. Money prices are much simpler to deal with: everything for sale then has a money price in terms of a unit of account, a single number. (Money has barter price(s) – what it buys in terms of goods and services: what economists call ‘real price’ is some average barter price across a “basket” of goods and services at some point or period in time expressed in numerical – i.e. “money” terms – to make it manageable.)

Money – as the medium of account: that is, a medium of exchange that embodies the unit of account – also connects across time. We accumulate existing obligations from past actions, obligations that are specified to be paid in the medium of account. We also have future spending intentions, making us attentive to money as a store of value. Our expectations about the future path of money as a store of value will affect our current holding and spending patterns; our use of money as something to hold (k) or something to spend (Py) in the current time period. So, if we expect money to seriously lose value, the incentive is to spend it; if we expect it to gain value, the incentive is to hold on to it.

These simplifying and cross-time aspects of money means that money matters in its own right, it is not simply a “transparent” connector to the “real economy” of goods and services. (So monetarism is correct in that.) Particularly as we are not immediately aware of all shifts in the barter price(s) of money: money is a response to a real information problem (keeping track of all those relative prices) yet, while it hugely reduces the information problem, it does not abolish it (since there is still the issue of becoming aware of shifts in the average barter price of money).

(I keep using the term ‘barter price(s)’ because I have come to dislike the terms ‘real price’ or ‘real wages’ as they skate over the information problem that money exists to solve and which is fundamental to how people use and think about money. So using ‘real price’ or ‘real wage’ becomes actively misleading even though talking of the 'barter price(s)' for money sounds a bit odd. Though surely no more odd than calling statistical artifacts 'real prices' and 'real wages'.)

While money does therefore matter in its own right, it is still just a means for human action. The constraints and expectations it is embedded in matter: there is no simple quantity story to tell. So, the level of money offers (i.e. money spent rather than held) matters, but whether any increase in spending will be inflationary or not (and how much) depends also on the responsiveness of aggregate supply.

Hence, as it is the nature of assets that they cannot respond as quickly to increased spending as goods and services can, it is easier to get much bigger inflationary surges in asset prices than in goods and services prices: surges which will be bigger the more constrained supply is. But the shift from using the Fisher ‘T’ to the Friedmanite ‘y’ diverts attention away from spending effects on asset prices.

The level of spending cannot, however, be inferred from M, since people’s expectations matter, as they will affect k. If people have low or negative inflationary expectations, then that is a reason to hold on to money. If they also have negative (i.e. pessimistic) uncertainty, that will be even more reason to hold on to money as a safety measure. If they wish to reduce debt (i.e. are “de-leveraging”), then that will also be a reason to hold on to money. So the combination of debt, negative uncertainty and low or negative inflationary expectations is likely to lead to a dramatic lowering in spending. Even while “base moneysurges: relying on such quantity measures will not merely be uninformative, they will be actively misleading. Hence ‘quasi-monetarism’ [now Market Monetarist (pdf)]: monetarism with the addition that expectations matter.

So, the worst thing a central bank can do in a situation of high debt and negative uncertainty is to reduce liquidity (and so both the availability of money and the expected path of money supply); thereby creating expectations of low or negative inflation, leading to a “flight to cash” and a dramatic drop in spending leading to a dramatic fall in Py (i.e. nominal GDP, GDP in straight money terms), with the drop in y (economic activity/output) being bigger the more downwardly constrained prices are.

This debt-and-deflation story is basically Irving Fisher's story (pdf) about the Great Depression. It is also what happened with the recent Global Financial Crisis and Great Recession (see this chart of changes in nominal GDP and employment).

An example of downwardly constrained prices are wage contracts operating across time incorporating significantly higher inflationary expectations than actually occur, so that the price of wages in terms of goods and services – their barter prices – rise as demand for what they produce is falling. With such wages being downwardly “sticky” since cutting wages means breaking the agreed contract, creating a trust-and-future dealings problem: particularly as people have already acquired obligations to pay in terms of the medium of account, so are resistant to receiving less of such regardless of what is happening to the barter price of money.

There is a long-running critique of central banking with fiat money that fiat money central banks are inflation-addicted. But they seem to do their worst damage when they become inappropriately inflation-phobic. The question is whether these are soluble problems, or they are a manifestation in monetary policy of the classic problem of central planning – the destructive and chaotic combination of poor incentives and information problems.

Still, one can be a quasi-monetaristMarket Monetarist in macroeconomic analysis without thereby being committed to any particular position on fiat money and/or central banking. (For those interested, George Selgin has recently posted a nice defense of fractional reserve banking [via].)

So, yes inflation is “always and everywhere a monetary phenomenon” but it is not a simple matter of measuring (and manipulating) monetary quantities: one has to include constraints and expectations. Indeed, expectations are where the real game is. So targeting stable growth in Py (nominal GDP) incorporates responsiveness of aggregate supply to money offers (i.e. output to spending), works to both stabilise and respond to expectations while not being constrained by institutional changes in money. It also optimises use of a policy instrument (monetary policy) which is a lot quicker in responding than fiscal policy without the nasty debt consequences (which is to say, fiscal policy is more rigid in both operation and consequences).

So, quasi-monetarists Market Monetarists of the world unite! There is nothing to lose but misleading obsessions with monetary aggregates!

ADDENDA I have amended the post to use the term 'spending' and 'output' more and 'money offers' less and incorporate the new name of 'Market Monetarists'.


  1. I was just told that I am a quasi-monetarist - good to hear. Your text indicates that it is not an insult.

    See my texts on the financial issues - e.g. the recent one about the Swedish policy - to see what I think.

  2. Yes, I thought that post was very good. As I commented on your post, you should read Scott Sumner.