Friday, September 16, 2011

What's the Matter with (Quasi-)Monetarism?

Let's start from the top.

What is monetarism? As I see it, it's a set of three claims. (1) There is a stable relationship between base money and the economically-relevant stock of money. [1] That is, there's a stable relationship between outside money and inside money. (2) There is a stable velocity of money, so we can interpret the equation of exchange MV = PY (or MV = PT) as a behavioral relationship and not just an accounting identity. Since the first claim says that M is set exogenously by the monetary authority, causality in the equation runs from left to right. And (3), the LM aggregate supply curve is shaped like a backward L, so that changes in PY show up entire in Y when the economy is below capacity, and entirely in changes in P when it is at capacity.

In other words, (1) the central bank can control the supply of money; (2) the supply of money determines the level of nominal output; and (3) there is a single strictly optimal level of nominal output, without any tradeoffs. The implication is that monetary policy should be guided by a simple rule, that the money supply should grow at a fixed rate equal to (what we think is) the growth rate of potential output. Which is indeed, exactly what Friedman and other monetarists said.

You can relax (3) if you want -- most monetarists would probably agree that in practice, disinflation is going to involve a period of depressed output. (Altho on the other hand, I'm pretty sure that when monetarism was officially adopted as the doctrine of the bank of England under Thatcher, it was claimed that slowing the growth of the money supply would control inflation without affecting growth at all. And the hedge-monetarism you run into today, that insists the huge growth in base money over the past few years could show up as hyperinflation without warning, seems to be implicitly assuming a backward-L shaped LM AS curve as well.) But basically, that's the monetarist package.

So what's wrong with this story? Here's what:


The red line is base money, the blue line is broad money (M2), and the green line is nominal GDP. The monetarist story is that red moves blue, and blue moves green. Between 1990 and 2008, this story isn't glaringly incompatible with the evidence. But since then? It's clear that the money multiplier, as we normally talk about it, no longer has any economic reality. There might still be tools out there to control the money supply. But changing the stock of base money -- the instrument of central banks, at least in theory, since the early 20th century -- is no longer one of them. Monetary policy as we knew it is dead. The divergence between the blue and green lines is less dramatic in this graph, but if anything it's even more damning. While output and prices lurched downward in the great Recession, the money supply just kept chugging along. Milton Friedman's idea that stable growth of the money supply is a sufficient condition for stable growth of nominal GDP looks pretty definitively refuted.

So that's monetarism, and what's the matter with it. How about quasi-monetarism? What's the difference from the unprefixed kind?

Some people would say, There is no difference. Quasi-monetarist is just what we call a New Keynesian who's taken off his Keynes mask and admitted he was a Friedmanite all along. And let's be honest, that's sort of true. But it's like one of those episodes in religious history where at some point the disciples have to acknowledge that, ok, the prophecies don't seem to have exactly worked out. Which means we have to figure out what they really meant.

In this case, the core commitment is the idea that if PY is too low (we're experiencing a recession and/or deflation) that means M is too low; if PY is too high (we're experiencing inflation) that means M is too high. In other words, when we talk about insufficient aggregate demand, what we're really talking about is just excess demand for money. And therefore, when we talk about policies to boost demand, we're really just talking about policies to boost the money stock. (Nick Rowe, as usual, is admirably straightforward on this point.) But how to reconcile this with the graph above? You just have to replace some material entities with spiritual ones: The true M, or V, or both, is not visible to mortal eyes. Let's say that velocity is exogenous but not stable. Then there is still a unique path of M that would guarantee both full employment and stable prices, but it can't be characterized as a simple growth rate as Friedman hoped. Alternatively, maybe the problem is that the monetary authority can only control M clumsily, and can't directly observe how far off it is. (This is the DeLong version of quasi-monetarism. The assets that count as M are always changing.) Then, there may still be the One True Growth Rate of M just as Friedman promised, but the monetary authority can't reliably implement it. Or sublunary M and V could both depart from their platonic ideals. In any case, the answer is clear: Since it's hard to get MV right, your rule should be to target a steady growth rate of PY (nominal GDP). Which is, indeed, exactly what the quasi-monetarists say. [2]

So what's the alternative? I've been arguing that one alternative is to think of recessions as coordination failures, which could happen even in an economy without money. I'm honestly not sure if that's going to turn out to be a productive direction to go in, or not. But in terms of the monetarist framework, the alternative is clear. Say that V is not only unstable, but endogenous. Specifically, say that it varies inversely with M. In this case, it remains true -- as it must; it's an accounting identity -- that MV = PY. But nonetheless there is nothing you can do to M, that will affect P or Y. (This situation, by the way, is what Keynes meant by a liquidity trap. It wasn't about the zero lower bound.)

This, I think, is what we actually observe, not just right now, but in general. "The" interest rate is the price of liquidity, that is, the price of money. [3] And what kinds of activity are sensitive to interest rates? Well, uh ... none of them. None, anyway, except for housing. When an economic unit is deciding on the division of its income between currently-produced goods and services vs. money, the price at which they exchange just doesn't seem to be much of a consideration. (Again, except -- and it's an important exception -- when the decision takes the form of purchasing housing services from either an existing home, or a new one.) Which means that changes in M don't have any good channel to produce changes in P or Y. In general, increases or decreases in M will just result in pro rata decreases or increases in V. Yes, it may be formally true that insufficient demand for goods equals excess demand for money; but it doesn't matter if there's no well-defined money demand function. A traditional Keynesian expenditure function (Z = A + cY) cannot be usefully simplified, as the quasi-monetarists would like, by thinking of it as a problem of maximizing the flow of consumption subject to some real balance constraint.

So, monetarism made some strong predictions. Quasi-monetarism admits that those predictions don't hold up, but argues that the monetarist model is still the right one, we just can't observe the variables in it as directly as early monetarists hoped. On some level, they may be right! But at some point, when the model gets too loosely coupled with reality, you'll want to stop using it. Even if, in some sense, it isn't wrong.

Which is all to say that, even if I can't find a way to disprove it analytically, I just can't accept the idea that the question of aggregate demand can be usefully reduced to the question of the supply of money.



[1] The simplest form of the first claim would be that the money multiplier is equal to one: Outside money is all the money there is. Something like this was supposed to be true under the gold standard, tho as the great Robert Triffin points out, it wasn't really. Over at Windyanabasis, rsj claims that Krugman, a closet quasi-monetarist, implicitly makes this assumption.

[2] In practice, despite the tone of this post, I'm not entirely sure they're wrong. More generally, Nick Rowe's clear and thorough posts on this set of questions are essential reading.

[3] I've learned from  Bob Pollin never to write that phrase without the quotes. There are lots of interest rates, and it matters.

7 comments:

  1. JW: "And (3), the LM curve is shaped like a backward L, so that changes in PY show up entire in Y when the economy is below capacity, and entirely in changes in P when it is at capacity."

    Did you mean "SRAS" rather than "LM" there? (And a bit lower down).

    ReplyDelete
  2. I've always thought of the Quantity Theory as a kind of pedagogic waystation. It is useful, as pedagogy, to loosen the "money illusion", as Irving Fisher called it, and to see how a coordination failure can be laid at a failure to manage the institution of money. And, yes, we are talking about a coordination failure; duh -- money is a coordinating device. But, as soon as I hear someone attributing causality, say, to the velocity of money, I know that he's hopelessly lost. Did you know that Irving Fisher actually built a contraption of pipes and valves, as part of his dissertation? It's all hydraulics and fluid flow. I like to imagine what it would be like to talk about the viscosity of money; I guess that's what's may be meant by waving one's hands at "frictions" and "sticky prices".

    Classically, money has three functions: medium of exchange, unit of account and store of value. A monetarist, or quasi-monetarist, has his thumb on medium of exchange. But, Euler would want us to derive an equilibrium in all three, simultaneously, wouldn't he? And, the whole point of credit and banking is to make medium of exchange trivially unimportant. That's what's wrong with monetarism -- if wrong is the right word for a teacher's attempt to kindle the beginning of insight -- money is just the beginning of finance. The economy is people behaving strategically against randomness and uncertainty; institutions shape behavior by creating games; money is how we keep score and marker our moves and collect our winnings.

    There's no one "interest rate". There's no flow of viscous stuff from ultimate savers to ultimate investors thru the interest rate valve. There's a yield curve, with risk premia ranging above it. Want to cause a recession? Invert the yield curve. Squeeze financial intermediation, the practice of arbitrage between lending long and borrowing short. That's how it is done.

    People always want money. In a sense, there's always excess demand for money. I, personally, can testify to my own excess demand for money. In Nick Rowe's story, there's a hidden switch. Ordinarily, people, who want money, go out and try to make and sell more stuff to get it. Then, one day, they choose instead to try to get more money, by the expedient of not buying stuff. Weird. Curious even, eh?

    ReplyDelete
  3. Thanks for Triffin, very good.

    ReplyDelete
  4. I'd heard of such contraptions but didn't know Fisher built one.

    I'll be honest, Bruce, the more I think about this stuff, the less certain I am about any of it. I feel like I need to spend a solid six months reading before saying anything more. But-

    I think viscosity of money is a good term but one that applies less to sticky prices, than to transactions costs or missing markets in credit markets. As you suggest, that's the key coordination failure. I think part of the problem on the Keynesian side is that the GT largely ignored the banking sector and was written, in places, as if all money is outside money. So we forget that in real capitalist economies, money is created by the banking system, which makes it hard to understand why the supply of it should be so inelastic, let alone exogenously fixed as in the monetarist story.

    Classically, money has three functions: medium of exchange, unit of account and store of value.

    I think the classics, and certainly Marx, added two more: means of payment, and world money. We don't need to worry about the last but as Marx says, means of payment -- i.e. as a vehicle for debts and other forward contracts -- is critical. (im Crotty has a superb essay on Marx's views on money, which really brings out the importance of its role as means of payment.

    the whole point of credit and banking is to make medium of exchange trivially unimportant. That's what's wrong with monetarism -- if wrong is the right word for a teacher's attempt to kindle the beginning of insight -- money is just the beginning of finance.

    Right. Have you read Minsky's John Maynard Keynes? He makes exactly this point, that the transactions motive is a trivial part of liquidity demand, so the income-form of the equation of exchange misses the most important parts of money demand. The transaction form is better but still not enough since money doesn't just finance transactions, but also positions in assets.

    People always want money. In a sense, there's always excess demand for money. I, personally, can testify to my own excess demand for money. In Nick Rowe's story, there's a hidden switch. Ordinarily, people, who want money, go out and try to make and sell more stuff to get it. Then, one day, they choose instead to try to get more money, by the expedient of not buying stuff.

    Yes, it is weird. But, maybe it's not quite as weird as you make it out to be. The idea is that people normally don't want money, but consumption. They hold money for a variety of reasons, but basically to deal with uncertainty, as you suggest earlier. So they normally target a certain ratio of money to income. When they feel more confident about the future, they want a lower ratio of money to income. When they feel less confident, they want a higher ratio. The problem is, since the stock of money is fixed in the short term (or produces offsetting effects on money demand as it changes) the desired ratio can't be achieved by increasing money holdings, only decreasing income.

    This story makes sense. I think you can find something like it in Keynes. And I think it probably does describe some important cases in the real world. But I don't think it exhausts the issue of demand constraints. And in particular, I find it hard to square with the current situation. To explain why low output persists despite all the liquidity that's been injected, you have to either postulate that the "money" in the story is something other than the stuff you we usually regard as money; or that there is some kind of credit market failure so that additions to the stock of money don't find their way to the units with the excess demand for it; or that demand constraints can be understood in some way other than excess demand for money.

    Curious, yes, for sure.

    ReplyDelete
  5. I liked this post up until this point:

    And what kinds of activity are sensitive to interest rates? Well, uh ... none of them. None, anyway, except for housing.

    Having known a fair number of people starting or growing businesses, they were all, in my experience, very sensitive indeed to the interest rates they could get. To be sure, they also cared about the other terms of financing (like collateral), and they all wanted to get to the point where they could pay for growth out of current income, but still. What am I missing here?

    ReplyDelete
  6. Cosma-

    That's very interesting. So you really hear people say something like, "I think I could sell more, so I'd take out a loan and expand if I could get a rate a point or two lower, but at current rates it's not worth it"?

    Because if you look at the empirical literature on the effect of interest rates on investment in the aggregate, the empirical literature is really strikingly negative. When Ben Bernanke writes about how monetary policy affects economic activity, his starting point is that in the real world investment is not responsive to interest rates: "In general, empirical studies of supposedly 'interest-sensitive' components of aggregate spending have in fact had great difficulty in identifying a quantitatively important effect of the ... cost-of-capital variable." Or as Alan Blinder says, "you have to torture the data pretty ruthlessly before they confess to an interest elasticity of investment." There's a really enormous empirical literature on investment functions and while I am no expert I believe the consensus, or at least a respectable plurality, is pretty clearly with Bernanke and Blinder that a strong (or any) response of business investment to interest rates is not really visible in the data.

    Econometrics has all kinds of problems -- as you know as well as anyone -- but in this case I'm inclined to believe it. It seems like a safe assumption that people usually don't make major investments or start new businesses unless they expect a rate of return well in excess of the prevailing interest rate. So the other financing terms you mention are usually more important -- the question is more likely to be how much credit you can get than at what price. And even that is probably less important than demand conditions. I'm sure you've seen that SBET survey of small businesses? Only 4 percent describe credit as their biggest problem.

    So while I definitely don't want to dismiss the views of people with first-hand experience of this stuff, I do think the assumption that most non-housing activity is not very responsive to interest rates is a reasonable one.

    ReplyDelete