New Keynesian vs Neo-Wicksellian macroeconomics

I am not going to disagree with Simon Wren-Lewis. There is a lot of merit to his perspective on what New Keynesian macroeconomics is really about. But I want to put forward a very different perspective that I think also has a lot of merit.

Here's Simon: "If I had to characterise New Keynesian theory by one thing, it would be the analysis of economies where the real rate of interest differed from the ‘natural’ real rate."

I would say he is talking about a theory that would better be described as "Neo-Wicksellian". Because Wicksell is the economist we associate with the analysis of economies where the rate of interest differs from the natural rate of interest. And when you do that analysis today, with explicit intertemporal optimisation, you might want to stick "Neo-" in front of "Wicksellian".

And yes, that Neo-Wicksellian strand is very much part of modern New Keynesian macroeconomics. Michael Woodford best exemplifies that strand. But New Keynesian macroeconomics long pre-dates Woodford's 2002/3 Interest and Prices (massive pdf).

I see New Keynesian macroeconomics as taking shape in the 1970's, in the debates between New Classical and New Keynesian macroeconomists over the Lucas Sargent Wallace/"Policy ineffectiveness Proposition" -- was activist monetary policy possible and desirable in a model with an expectations-augmented Phillips Curve and rational expectations? The New Keynesian macroeconomists said activist monetary policy, seen as a feedback rule, was possible and desirable.

And this debate in turn hinged on two different interpretations of the expectations-augmented Phillips curve, both of which went back to Milton Friedman 1968 (pdf), and the "Phelps Volume" of 1970. There was the "Lucasian" interpretation, which saw it as a supply curve under imperfect contemporaneous information. And there was the "Phelpsian" interpretation, which saw it as representing the behaviour of price-setters under imperfect information about the future. (I owe this distinction, and the idea that both are found in Friedman, to David Laidler, who has been saying this at least since 1977.) New Classicals followed the Lucasian interpretation; New Keynesians followed the Phelpsian interpretation.

Now Simon took heat for saying that maybe Leijionhufvud was the first New Keynesian macroeconomist. But funnily enough, if you accept my very different perspective on what New Keynesian macroeconomics really is, you could argue that Leijionhufvud played a role there too.

Here's David Laidler (pdf), (HT Mark Thoma) writing about Milton Friedman and the origins of the expectations-augmented Phillips curve:

"In 1953, however, Friedman treated perfect competition as a Marshallian tool, useful for empirical analysis, and classified monopolistic competition as empirically vacuous and therefore Walrasian. At that time the awkward question of who sets prices in an economy in which everyone is a price-taker was not on the agenda, and the necessity of postulating an “as if” auctioneer who continuously keeps prices at their market-clearing values as a necessary component of the microeconomic foundations of the perfectly competitive model was not understood. It was, above all Axel Leijonhufvud (1968) who forced macroeconomists to pay attention to this question." (my emphasis).

And if you are going to model an expectations-augmented Phillips curve as representing the behaviour of those who set prices under imperfect information, and quantities traded being determined by the demand-side given those prices, which is what the Phelpsian New Keynesians did, it makes a lot of sense to do that in a model where those price setters have, at least temporarily, some sort of monopolistic power. And so imperfect competition also became central to New Keynesian macroeconomics.

Phelps is the father of New Keynesian macroeconomics. But maybe Leijonhufvud had a role too?

Just another way of looking at things.

(P.S. I don't know how old Simon is (other than his hair looks less grey than mine), and I haven't tried to find out. But I would say that his is a "young guy's perspective" on New Keynesian macro, and mine is an "old guy's perspective" on New Keynesian macro. For us old guys, all that Neo-Wicksellian stuff came later. You can't do Wicksellian analysis without sticky prices anyhow, because otherwise the economy will explode instantly if the actual rate is ever different from the natural rate.)

Comments

[Matias Vernengo]

History of ideas is a difficult subject, and does not receive as much attention as it should. The central concept in Wicksell is the natural rate of interest, and the whole dynamics comes from its difference with the bank or monetary rate. Friedman (1968), which brought back the concept (which Keynes explicitly rejects in the General Theory), as the natural rate of unemployment is the one that made all modern macroeconomics (including New Keynesians, New Classicals and RBC) post-Wicksellian in the appropriate term used by Edward Amadeo in his book "Keynes' Principle of Effective Demand." In that sense, Friedman, not Phelps or Leijonhufvud for that matter, is the forefather of New Keynesian economics. And that is why Mankiw or John B. Taylor, for example, are New Keynesians in spite of their very anti-Keynesian policy positions.

[Simon Wren-Lewis]

Nick

You are absolutely right that I'm talking about a Neo-Wicksellian version of New Keynesian theory, as laid down in that seminal work by Michael Woodford. And I do not think I have any problem with the other things you say. I guess what partly motivated me in writing my post was an earlier post where I argue that price flexibility would not be sufficient to restore demand at the zero lower bound under inflation targeting, precisely because it fails to get us to the natural rate of interest. And why is that important? Only because there are still plenty of people who refuse to acknowledge that prices are sticky, and so exhibit demand denial.

And yes Leijonhufvud has been influential in many ways, even for those of us who refuse to come over from the dark side of mainstream orthodoxy.

However unfortunately you are completely wrong about one thing. Admittedly you were not to know your evidence base was tainted - its a very old photograph. The problem is that it reflects how I think I look rather than reality, so I never think it needs replacing. But I liked your theory, because it suggests I'm still young at heart!

[Nick Rowe]

Matias: "In that sense, Friedman, not Phelps or Leijonhufvud for that matter, is the forefather of New Keynesian economics."

Curses! You have just reminded me of something I had forgotten! I argued very much the same thing just recently!.

There is an awful lot of truth in what you say. In fact, if I wanted to distinguish New Keynesians from Old Keynesians, I would say, like you, that Friedman is the father of NK. But if I wanted to distinguish New Keynesians from New Classicals, I would say Phelps is the father.

[marris]

Lots of references here that I should read. Thanks!

> You can't do Wicksellian analysis without sticky prices anyhow, because otherwise the economy will explode instantly if the actual rate is ever different from the natural rate.

This part struck me as very odd. I think the classical Austrian analysis is very Wicksellian... the CB sets the interest rate below the natural rate, causes too many (or the wrong kind of) capital goods to be produced, and everyone "realizes the mistake" later, then the first non-CB banks start calling in bad loans.

In this example, you have "Wicksellian analysis," but no sticky prices per se (at least, not of the form where quantity adjustments are faster than price adjustments). And yet, the economy does not "explode" due to the incorrect rate. What is the missing piece? Is the Austrian business cycle theory incoherent? Who says that the economy *has* to explode?

Or do capital goods investment provide a new source of "stickiness" since bad investments can be "hidden on the books" for a longer period of time?

Either idea would be worth thinking about.

[Nick Rowe]

Simon: Thanks. I very much agree with what you say.

But damn, damn, damn! I wish my memory weren't so bad, and I had remembered what I had written earlier, or had seen Matias' comment *before* hitting publish (impossible, of course)!

Which just proves I'm an old guy!

"I guess what partly motivated me in writing my post was an earlier post where I argue that price flexibility would not be sufficient to restore demand at the zero lower bound under inflation targeting, precisely because it fails to get us to the natural rate of interest."

Understood. But I've never been able to get too excited by the question of what *would* happen if prices were perfectly flexible and the actual rate were set above the natural rate. Because it's such a weird hypothetical. Not just that prices aren't perfectly flexible, but that the economy would explode. And the central bank wouldn't be able to set interest rates anyway, if prices were perfectly flexible.

[Nick Rowe]

marris: "This part struck me as very odd."

Yep. And that is precisely why I have never been able to get my head around the Austrian vision. Because if prices are sticky, you are going to get Keynesian effects (and possibly Austrian effects on the capital structure too). But if prices are perfectly flexible, you don't get any Austrian effects, because there's just a big bang! In short, and oversimplifying: because the vertical AD curve doesn't intersect the vertical AS curve at any price level.

[Jon]

The Austrian view was always that no one knows what the actual real rate is. When the banks pushed the market rate below the natural this sets off a dis-coordination. At the first, there is no sense of the mismatch but stocks of intermediate goods begin to drain down. Meanwhile the shift in the interest rate causes projects with distant returns to look profitable. This diverts factors of production away from the very intermediate goods whose stocks are draining. This turn cases factors of production to be bid up and more labor to mobilize due to what appears to be a higher real rate which in turns puts pressure on stocks of consumer goods.

Importantly this system does not eqilibrate because if market loan rate is pegged nominally, the rising price level makes long term projects seem yet still desirable despite the cost increases...

Because all of the excess deman is driven from the banking sector, th effect starts off in proportion to net volume of new lending vs the size of the economy.

There are only two sticky behaviors in this model: the market rate of Interest which is pegged by the banking cartel and the price level expectations which remain anchored.

[marris]

> Yep... Because if prices are sticky, you are going to get Keynesian effects (and possibly Austrian effects on the capital structure too).

Well, what's so special about the *interest rate* in the Keynesian view? Would the economy explode when the CB fixes *any* price (e.g the minimum wage, or the price of oil)? Why do Friedman et all think the interest rate is so important when there are many other prices which may twist the Phillips curve?

[Jon]

Btw, viewing e Austrian critique in a historical context.... Banking instability had been of interest for hundreds of years. This led to all manner of laws to restrict the supply of bank notes, although financial product innovation kept winning. The Fed was just one of the supposed solutions. Abandoning gold was another...

The Austrians argued was that the problem was endemic to the banking system--that any inflation rate and any price level path would suffer these fluctuations: The boom bust cycle merely required that the banking system make forecasting errors.

So in the 20th century the Austrian solution was to keep the size of banks small relative to e economy. Thus you had mises and rothbard pushing 100% gold reserves. The big Austrian mistake was to blame the depression on the damage due to miscoordination of inter temporal preferences. They had a nice hammer but not everything is a nail.

Pretty much everything you call wicksellian I would call Austrian, and the Austrians got to expectations first. What they didn't understand in the early 20th century was what we now call AS/AD. The concept of national income appears no where. Hayek of course internalized these and started pushing ngdp targetting eventually.

Internet Austrianism won't die because most of Austrian thought is right. The Austrians led the marginal revolution and cannot be viewd distinct from wicksell who was their contemporary. It's time for the tribalism of Hayek vs Keynes to die. The path forward is to set the ducks in the right positions and start getting the historical narrative organized...

[K]

I'm with the Austrians here (Lord help me :-).

The first point is that even for economies with very simple, but stochastic dynamics, it is impossible for Bayesian agents who know the form of the dynamics but not the parameters to actually discover those parameters, *even given an unlimited quantity of observations*. Anybody who hasn't read Weitzman's Subjective Expectations and Asset-Return Puzzles really needs to do so and think about it *very hard*:

"More technically, this paper shows that when agents are experiencing a dynamically evolving stochastic process that is relevant to asset pricing, the subjective probability measures from Bayesian learning stay uniformly bounded away from the actual data-generating process—even with asymptotically infinite past observations."

The rational expectations equilibrium is merely one of a continuum of possible equilibria: the one in which the agent somehow knows the structural parameters, presumable as handed down by God. One consequence of this is that the Bayesian has no idea what the "correct" natural rate is. The best estimate is highly sensitive to that Bayesian's prior. It therefore seems unlikely that in a properly specified NK model (I.e. one where the rep agent is not privy to God-given parameters), that the dynamic would be hypersensitive to misspecification of the real rate. There is, after all, a very wide distribution of plausible natural rates, including presumably, the market rate.

I think, perhaps, that this points us towards a modern (post-Lucas) specification of the Keynesian "animal spirits" (vague prior), which I think is at the core of the Leijonhufvud critique of the Neoclassicals. We don't even need irrational behaviour (though there may be plenty of that too). Just rational econometricians faced with the limits of mathematical inference even under infinite observations. Sticky prices may be a distraction, or at least, a lot less central than imagined.

[Ritwik]

Nick

I think you're right that Ned Phelps is really the grand-daddy of New Keynesian Economics - this holds true even if you consider the Neo-Wicksellian perspective. Leijonhufvud, in a paper criticising the natural rate of unemployment as a theory fundamentally incompatible with Keynes - {link here pdf NR] (page 3), partially absolves Phelps for his realisation that unemployment over the natural rate is isomorphic with the market rate being higher than the natural rate.

And yes, if imperfect information about the future was really the key, then Leijonhufvud does deserve a mention - his UCLA micro-economics is fundamentally of the imperfect information nature. But in that sense, New Keynesians are often removed from both Phelps and Leijonhufvud - nominal rigidities/ real rigidities are about exchange in the present, not inter-temporal trade-offs. In this sense, it would be Friedman, rather than Leijonhufvud, who would be the real complement to Phelps as the fountainhead of New Keynesian macro.

In fact, one can even argue that the god-father of most things New Keynesian is actually Modigliani, who admitted in his debate with Patinkin that it is nominal wage rigidity, and not liquidity preference, that is the most crucial link in explaining cyclical unemployment.

But I disagree when you say this - 'You can't do Wicksellian analysis without sticky prices anyhow, because otherwise the economy will explode instantly if the actual rate is ever different from the natural rate'. This is you saying that the only way to be a Wicksellian/ neo-Wicksellian is to be a neo-Walrasian equilibrium theorist, ala Woodford 2003. From there, it is easy to conclude that nominal rigidities/ money illusion are essential to any New Keynesian theory of business cycles, and money as a hot potato is essential to any non-Walrasian properties that NK business cycles may exhibit. So that NK is either internally contradictory, unless subsumed under a properly defined monetary theory where interest rates may not be as special as they are in Woodford.

This is true, if neo-Wicksellians are limited to New Keynesians. But this definition of would exclude Keynes, and indeed even Wicksell, from being neo-Wicksellian. And I believe both are.

There are two other ways to be a neo-Wicksellian. One is the Marshallian disequilibrium value theory way, which is Keynes of the Treatise of Money (and indeed, Axel Leijonhufvud himself). The economy does not explode when the natural rate is different from the market rate, because there is no Walrasian simultaneity. Yes, everything causes everything else, but what appears simultaneous is instead a set of quasi-linear adjustments enacted by heterogeneous market participants acting on imperfect and heterogeneous views of the future. This leads to a market state which is in equilibrium in the stability/ market clearing sense, but in disequilibrium in the original sense of ex-ante plans not being realized ex-post. In the Keynesian scheme, this happens through liquidity preference, of bond market speculators (in the Treatise) or the more naive/ extreme money demand of all agents (in the GT).

I get my insight form Leijonhufvud - here [link here pdf NR] and here [link here pdf NR]

(As an aside, you will find page 21 of the first pdf quite interesting. In 'Monetarist counter no. 2' you will see Leijonhufvud lay out the Chuck Norris conception of monetary policy, subsumed under rational expectations. This is Scott Sumner. I believe that the sequence also applies to monetary disequilibrium and an upward sloping IS curve. This is you and the hot potato, without requiring ratex. I plan to write about it sometime.)

The other way to be a neo-Wicksellian without being neo-Walrasian is the way of Wicksell himself, from the capital theoretic standpoint of the hypothetical pure credit economy. In this view, you could have a pure loanable funds theory of interest, and still generate a market rate different from the natural rate, because through actions of the banking system, loanable funds become distinct from time preference and savings, so that the economy does not quickly adjust to the natural rate, nor explode/implode when the market rate differs from the natural rate. You will find the distinction between loanable funds and time preference, between saving and financing, being made on page 27 of the second pdf by Leijonhufvud, but also through-out this recent work by Claudio Borio - [link here pdf NR]

The common factor of both ways is that they're essentially about inter-temporal disequilibrium through finance, adapted to the financial systems of their times. The Wicksell way is of an unfettered banking system. The Keynesian way is of an unfettered bond market. Neither has much use of either nominal rigidities or imperfect competition, which are both about atemporal exchange. But both are concerned about the information, institutions and prices coordinating inter-temporal decisions.

In this sense, Wren-Lewis is correct, but incomplete, in describing New Keynesian economics as one where the market rate differs from the natural rate. NKE, Woodford style, only presumes this about the short rate. But the short rate is not the cost of capital. It is not the market rate of either Keynes or Wicksell. The reason NKE is able to get around this is because

1) Nominal stickiness. This is your preferred explanation.
2) Presuming Fischer Black style financial market equilibrium about the rest of the term structure. This is the Woodford-ian way.

I submit that to be a true neo-Wicksellian is to posit that the bond/credit market has a life of its own, and rather than being bounded by time preference and factor price equalization neo-classical style, it may in some circumstances create the bounds on those processes. The price theory underlying this is Marshallian, and the GE benchmark underlying this is basically a CAPM, ala Jim Tobin and Fischer Black, with a variety of Keynesian and non-Keynesian deviations possible from this Walrasian ideal.

This is Axel Leijonhufvud, Perry Mehrling. It is Ned Phelps + finance. It could be Woodford, if he makes substantial revisions to his magnum opus. It is different from monetary disequilibrium/ market monetarism but it is also different from Woodford 2003.

[Brito]

Nick, do you ever get the feeling that New Keynesian economics is complete? What I mean is, there's nothing wrong about it, but I feel as if we've stretched how far nominal rigidity, imperfect competition and other frictions can explain deviations of output from its potential. I can't see much more interesting insights to come out of it anymore.

[Ritwik]

K

What you're saying is right, but not necessarily Austrian. Consider the Fischer Black challenge, which I interpret as saying that *the natural rate* is whatever the market says it is, and the various natural rates can all be priced in through CAPM.

In fact, it is difficult to imagine how Lucas style ratex of representative agents has become so popular without incorporating finance, because Black style CAPM is logically the sine qua non of any well-founded ratex, and the arbitrage pricing logic of finance is so much more appealing (and compatible with Marshallian price theory) than constrained optimization, Arrow-Debreu style.

I disagree with Black, of course - one of the reasons is the entire literature on the various reasons for allocative inefficiency of financial markets, of which you quote one important paper - but to the extent that there exists a neo-classical GE from which one may wish to explain departures, it's often struck me that the right GE to start with is that of Black rather than Arrow-Debreu.

[K]

Ritwik,

I think the problem of incorporating asset pricing is that it's quite tough to extend the single period CAPM to the basic DSGE framework. If the moments are bounded and the increments are independent then the central limit theorem applies in the long time limit, and it's possible to get arbitrarily small risk exposure in the limit, which precludes the existence of risk premium. So you need finite horizons (OLG) and the model ends up depending critically on the exact parameters of the time horizon, and the assumptions regarding frequency of tail events (Peso problem). This is the Rietz/Barro approach: Assume ratex and explore the limits of busting the model out in the regions where your hypotheses are untestable anyways. Entirely "sound and fury, signifying nothing," in my opinion. The Geweke/Weitzman approach is beautiful by comparison. Assume Bayesians in a simple stochastic (continuous with finite moments) economy. Endow them with powers of perfect observation and unbounded rationality. It turns out that with very non-vague (normal) priors, and normal returns (with unknown parameters), the agents posterior distributions may not have finite moments and are therefore not subject to the central limit theorem. So we get risk premium and lots of it. So I think the answer to your question about why macro hasn't incorporated asset pricing is that it's *really* hard but that maybe it's about to happen in a big way over the next few years.

When you refer to "Black style CAPM" are you thinking about the zero-beta CAPM (Black 72)? Is that where the "Black challenge" comes from?

Also, I don't think of Weitzman as a challenge to allocative efficiency. Am I wrong?

As far a complete markets being a waste of time... I couldn't agree more. You can't even hedge an equity option in the underlying stock. The idea that you can hedge, e.g., your human capital (or that it doesn't matter that you can't) is a joke. Incorporating agent *assessment* of risk premium into macro models is critical and, from what I can tell, the need to do is very poorly understood in the macro mainstream.

[rsj]

K,

That is a great paper! What type of reception has it received?

[K]

rsj,

"That is a great paper!"

It really is. You should also check out Jobert, Platania, Rogers (2005), which is similar but technically clearer, in my opinion. But Weitzman is full of great insights.

"What type of reception has it received?"

It still seems to be hotly contested in the asset pricing literature, but if it's gotten much attention in the macro literature I'm not aware of it. But I'm an asset pricing/derivatives person, not a macro person. I've seen a fair bit of "we don't need your weird model to explain asset pricing puzzles" criticism from the ratex crowd, which I consider to be a weird criticism, since it's not a "model." In my opinion it's a deep and inescapable criticism of ratex. Other criticisms are of the kind: parameter uncertainty is insufficiently great to account for the premium puzzles. Also seems to be a lousy criticism since parameter uncertainty doesn't preclude some intrinsic kurtosis (Peso problem) in the underlying dynamics or the parameter prior. It just means we don't need to postulate very much of it (if any).

I'd love to know why it hasn't gotten more attention in macro (or maybe it has) since the implications ought to be significant and relevant to stabilization policy and public debt dynamics (the relationship between growth, risk premium and the risk free rate is endogenous). Obviously it's mathematically much tougher than REE models so perhaps that's the problem. No excuse for not trying numerical techniques though. But as I said, I *really* don't know the macro literature, so I could be off base.

What *has* gotten lots of attention is Weitzman's papers (e.g. this one, also a great read) on the impact of parameter uncertainty in discounting the expected payoff of climate change under concave utility. Similar ideas as the above, used to provide a mathematically coherent statement of the precautionary principle.

[K]

K: "I don't think of Weitzman as a challenge to allocative efficiency. Am I wrong?"

I probably am. Again, I'm a finance person, not an economist. But I am somewhat aware of complete market versions of proofs of the welfare theorems. I assume other proofs depend on ratex, restricted forms of utility and unlimited, equal access to borrowing? I assume rational *Bayesian* expectations are insufficient, but that seems like a good question. Maybe it depends on choice of prior, or perhaps it's totally hopeless. ?

[rsj]

because otherwise the economy will explode instantly if the actual rate is ever different from the natural rate

I don't understand why. Can you spell it out a bit more? IIRC, Wicksell did not believe in infinite prices as the consequence of a deviation of the financial rate from the natural rate.

For example, compare mortgage bond rates with the price/rent ratio of houses. I am thinking, if the bond rates are lower than the rental rate, then there will be a lot of residential investment, at which point the increase in supply of houses is so large that the rent paid falls to the bond rate.

[Nick Rowe]

Jon: "There are only two sticky behaviors in this model: the market rate of Interest which is pegged by the banking cartel and the price level expectations which remain anchored." [italics mine]

That was potentially a very useful comment. One I'm going to think about, slowly. I'm not sure if it solves the problem or not.

[Nick Rowe]

marris: "Well, what's so special about the *interest rate* in the Keynesian view? Would the economy explode when the CB fixes *any* price (e.g the minimum wage, or the price of oil)?"

A possible answer. A rate of interest has the units 1/time. All those other prices have $ in the units. The nominal wage is $/time. The price of oil is $/barrel. Etc.

[Nick Rowe]

Brito: "Nick, do you ever get the feeling that New Keynesian economics is complete?"

I don't know. I expect I do get that feeling, even though I can't really justify it. Sure, lots of people will add lots of little bits and pieces to NK macro, all useful. But I really think there's a fundamental problem with the structure, in how it treats money. I've attempted a few posts on this in the past, trying to put my finger on it.

But from the policy angle, the fact that it all falls apart at the ZLB is a biggie.

[Nick Rowe]

rsj: "I don't understand why. Can you spell it out a bit more?"

Simple version: take a standard ISLM/AD/AS model. Make the LM horizontal. The AD curve now becomes vertical in {P,Y} space. Under perfectly flexible prices the AS curve is also vertical. The AD and AS curves don't cross. If the CB sets a rate of interest above/below the natural rate, the vertical AD is to the left/right of the vertical AS, and the economy goes "Bang!"

[Nick Rowe]

Very good comments. Some will take a lot of thought. My brain hurts.

[Ritwik]

K

I inferred the Weitzman paper as a Godel type undecidability proof. While assuming that there is indeed a deep parameter of the economic process (the natural rate) that the agents are trying to hit through Bayesian updating - it shows how this process is extremely dependent on the subjective priors, and may never converge. If the market rate differs from the natural rate, allocative efficiency is hampered by definition in some conceptions, and this is surely the impression I got from Weitzman, but maybe I was wrong.

I once read (from Tyler Cowen, perhaps) about someone who had proved that the informational requirements for a GE/ CAPM GE are such that GE exists iff. P=NP. Both this and Weitzman are important papers, but I see them as belonging to the assumption - proof type existence theorems, in which agents are still trying to solve constrained optimization problems.

These are challenges to the Walrasian ideal from within the Walrasian framework.

My interpretation of the Black challenge is my own, and may be incorrect. It does not follow from any specific paper, but rather form Perry Mehrling's overall interpretation of him, esp. here - http://economics.barnard.edu/sites/default/files/inline/understanding_fischer_black.pdf . It is to do with arbitrage pricing, and to begin with the given market price, rather than any preferences or constraints, as the origin. Say 'vol' is a parameter. What is the 'true' vol? Weitzman says that agents who don't know at the beginning may never know in the future. Black would agree, but would ask how we know if there's a true vol that is independent of the vol implied in the market price? In any case, I see CAPM as the more logically satisfying version of GE. Here is John Geneakopolos/ Martin Shubik framing the CAPM as a GE with incomplete markets : http://cowles.econ.yale.edu/P/cp/p07b/p0759.pdf

Nick, Jon

"There are only two sticky behaviors in this model: the market rate of Interest which is pegged by the banking cartel and the price level expectations which remain anchored."

I.e., the real rate on bonds rb is sticky. Or, the LM curve, properly drawn in (Y, rb) space, is horizontal-ish. Or, Keynes, at least the Keynes that follows from Wicksell. I'm not sure why you would call this particularly Austrian - which author do you have in mind? My read is that over-investment through banking system disequilibrium does not co-exist with anchored price expectations in the Austrian model. For that, we must turn to Wicksell. From Leijonhufvud, page 32 here - http://www.libertarianismo.org/livros/sgpeotmpaase.pdf#page=19

But nevermind. Even if what you say is Austrian, if a sticky rb was to explain booms and busts, then it's easy to resuscitate standard Keynesian IS/LM results even in a model where the IS curve is not necessarily downward sloping.

[rsj]

OK, let me just back up a bit. Assume that everything is in equilibrium and prices are flexible.

1. (WDWW) What did Wicksell write about the result of a lowering of the financing rate below that of the natural rate? Infinite prices? Undefined prices? I'm thinking I don't know Wicksell like I thought I knew Wicksell.

2. Where did he go wrong?

[Ritwik]

Nick

This may be off-topic, but I believe that it is mistake to think that a CB that sets the short rate has made the LM curve horizontal. The LM curve is (or should be)a money/bonds curve, not a money/money curve.

Let me propose a re-interpretation of the IS/LM.

1) Money is the numeraire. As in, not just the unit of account, but also the only 'risk-free' asset. Near-moneys are money. So, T-bills are (also) money. Money bears real interest rm(which can be 0 or negative or anything).

2) The only other asset is 'bonds', loosely defined. Say, T-bonds/ G-secs of 10 years.

3) The curve space is (Y, rb) where Y is the expected/trend path of nominal spending and rb is the real rate on bonds. Note that Y is not the current NGDP, but the path.

4) The price level/ inflation is not fixed.

5) The IS curve represents the schedule of inter-temporal optimization (Wickseliian natural rate/ savings-investment equilbrium), and is horizontal(Yes, I believe Keynes is perfectly reconcilable with a horizontal IS curve). Keynesian shocks to MEC are up and down shifts of the IS curve.

Start with the Old Keynesian insight that the demand for money varies negatively with the nominal rate on bonds. Add the Fisherian insight that the demand for money varies negatively with the expected inflation. The two are perfectly reconcilable. How? Keynesians hold the price level/ inflation expectations constant (they also assume zero interest bearing money, but that doesn't change things). So they are changing rb, keeping rm constant. Fisherians are invoking superneutrality, non-interest bearing money and a varying price level/inflation. So they are changing rm, keeping rb constant. In either case, Md varies negatively with (rb-rm).

Notice that this is also an easy generalization from Tobin's 'liquidity preference as aversion to risk'. Again, he held the nominal interest rate on money to be 0, but that doesn't matter. The nominal rate and inflation can both be allowed to vary. What matters for the portfolio decision (as well as the Baumol-Tobin transactions demand for money) is the difference between rb and rm, and risk aversion.

rm is determined endogenously by the central bank, but is exogenous to the money/bonds equilibrium schedule (LM curve). Risk aversion is determined endogenously by animal spirits/ the financial system, but is exogenous to the LM curve.

LM curve, with rm and risk assessment as parameters, is drawn in (Y, rb) space. A cut in rm by the central bank is an outward shift in the LM curve. An increase in risk aversion is an inward shift of the LM curve. And vice versa for both. It is a mistake to think that a cut in rm is a downward shift of a horizontal LM curve. The LM curve is not drawn in (Y, rm) space, it is drawn in (Y,rb) space. A neo-Wicksellian central bank does not make the LM curve horizontal. An NGDP targeting central bank does not make the LM curve vertical. Central banks shift the LM curve around, to the best that they can. Its shape is what it is.

A classical economy, one that is within Leijonhufvud's corridor and has a tendency to auto-correct, is one where the LM curve is roughly 45 degrees. The relative prices remain the same and the economy transitions smoothly between paths of output in response to MEC changes. This is also my interpretation of Wicksellian price stability, which I believe is analogous to but cannot be reduced to price level targeting.

An NGDP targeting central bank tries to make the LM curve as-if 90 degrees and thus over-reacts to changes in MEC.

This LM curve captures both the monetarist intuition that there is no zero lower bound (as rm can be decreased even when the nominal rate on money is zero) and the Old Keynesian intuition that the liquidity trap has nothing to do with the ZLB (if the LM curve is horizontal-ish, shifting it around may not be of much help).

Those who worry about balance sheets, debt-deflation, animal spirits, ZLB etc. are basically saying that there are forces trying to shift the LM curve inwards. Those that say that the central bank can always create inflation are saying that the LM curve can always be shifted outwards. But both narratives, while important and true in their own right, are tangential to the Keynesian story, which I interpret as saying that the LM curve is horizontal-ish for a wide-ish range of Y.

[Ritwik]

rsj

I think Wicksell said infinite prices, if un-anchored ratex is assumed. But actual prices mean-revert due to metallic backing, which anchors expectations, making price-change expectations as-if adaptive and hence stable.

The absolute price level is indeterminate even when the natural rate is same as the market rate, and is purely historical.

http://www-ceel.economia.unitn.it/staff/leijonhufvud/files/wick.pdf

[Nick Rowe]

rsj: Wicksell said the price level would *eventually* go to infinity if the central bank held the actual rate below the natural rate permanently. I'm saying it would go there *instantly*, without sticky prices. (I should probably have said instead "without *some* sort of nominal rigidity like sticky prices").

My memory of Wicksell is too fuzzy, and it was too long ago, for me to say exactly where he went wrong. Plus, sometimes you can't really say where he went wrong, if there isn't a formal model. Maybe, as Jon and Ritwik say, it's got something to do with expectations. Certainly, if you assume the right form of adaptive expectations, where people make consistent mistakes and always underestimate what inflation will be, you could get a Wicksellian process in which the price level continuously rises slowly, but always more quickly than people expect.

[Nick Rowe]

Ritwik: you are close enough to be on-topic. And you may be on to something. But you lost me right at the beginning.

"The LM curve is (or should be)a money/bonds curve, not a money/money curve."

In a monetary exchange economy, with n goods (including the medium of exchange), there are n-1 markets. for each good except money, there exists a market in which that good exchanges against money. Money appears on one side of every market. (Clower). There is an "apple market" in which apples trade for money......There is a bond market in which bonds trade for money. The term "money market" is an oxymoron. What is a "money/money curve"?

(Sorry, just one of my pet peeves!)

[Ritwik]

Nick

Sorry, I was unclear. I was just referring to the usual interpretation of the LM curve as money demand vs money supply. I was re-formulating it to mean the locus of points of the economy where there is no excess demand for money relative to bonds. Perhaps that has always been the definition and I am just stating the obvious. But a lot of discourse tends to view the interest rate in the (Y, r) IS/LM space as the short rate, which strikes me as plain wrong.

Just scratch that sentence out if you will. My basic point simply is - money, properly defined, has always paid interest - the short rate. The short rate is not the interest rate where the LM curve is presently at. The short rate is a parameter not a coordinate axis. MMTers say 'wake up to reality, cenral banks set the interest rate, LM curve is horizontal'. You say 'equilibrium is unstable when the IS curve slopes up/ is horizontal and a central bank sets the short rate'.

Wrong and wrong. The LM curve, if horizontal, is horizontal because of investor preference. Setting the short rate has nothing to do with it. The equilibrium, if unstable, is unstable because the IS slope exceeds the LM slope. Setting the short rate has nothing to do with it.

[Nick Rowe]

Ritwik: "I was just referring to the usual interpretation of the LM curve as money demand vs money supply. I was re-formulating it to mean the locus of points of the economy where there is no excess demand for money relative to bonds. Perhaps that has always been the definition and I am just stating the obvious."

You are not stating the obvious. Few treatments of the LM even acknowledge the distinction. They treat them as synonymous.

This is something I have never been able to get my head around properly.

"My basic point simply is - money, properly defined, has always paid interest - the short rate."

Dunno. Currency doesn't.

[Ritwik]

Nick

Sure, currency doesn't. But currency demand is potentially purely inventory theoretic. Cost of storage, number of transactions made per day, number of days to procure more currency etc. It cannot explain business cycles, even from a monetarist *excess money demand* perspective.

And the special nature of currency does not reconcile with any expectations theory of monetary policy, which should be independent of the existence of currency. A general theory of monetary policy should work with any real rate of money, at any level of inflation.

For the longest time in the US, reserves didn't pay interest. So the fed funds rate was construed as the scarcity of money rate. I think it would be fairer to say that the system was operating at peak elasticity - there was no scarcity of money, the price of liquidity was zero - Friedman style - and the entire fed funds rate was effectively the interest on money.

[Nick Rowe]

Ritwik: "But currency demand is potentially purely inventory theoretic."

Isn't that true of the demand for all media of exchange? A "temporary abode of purchasing power", because it is hard to predict, and hard to match, desired spending and receipts of the medium of exchange? So we hold an inventory, or buffer stock, precisely because equilibration of planned spending and receipts is difficult. The buffer stock is there to handle shocks. But if the shock is too big, or seen as being too permanent, we decide to take action to adjust our planned spending and/or receipts to rebuild or sell off our inventory. (Leijonhufvud corridoor). And when a lot of people try to do this at the same time, this is what causes booms and busts. And it's not even obvious precisely what "equilibrium" and "disequilibrium" mean when we are talking about buffer stock/inventories.

[marris]

> A possible answer. A rate of interest has the units 1/time. All those other prices have $ in the units. The nominal wage is $/time. The price of oil is $/barrel. Etc.

Very nice! So we have two ways to look at interest: as a "real thing" (maybe "real-ish") since the units don't have $, and as a "money thing" (since it's a "price of money" and everyone knows that the CB can both set price floors and create money (the relevant qty in this market) at will).

Funny that it's one of the things that the CB *could* set most easily and probably the one "real thing" in the economy that it should stay furthest away from.

[Nick Rowe]

marris: Yep!

[Jon]

"But nevermind. Even if what you say is Austrian, if a sticky rb was to explain booms and busts, then it's easy to resuscitate standard Keynesian IS/LM results even in a model where the IS curve is not necessarily downward sloping."

Ritnik, anything involving capital structure theory is Austrian. Marginal analysis is Austrian.

The one point which everyone falls on is the claim by some Austrian authors that some capital goods cannot be repurposed and this leads to unemployment. This claim is a mere assertion--if you struck that claim from the texts I challenge you to derive it from Böhm-Bawer's, Menger's, or Mises's other writings. It cannot be done. It's an ad hoc supposition.

[Ritwik]

Nick

I'm not denying the issues with cash-flow matching. Indeed, Perry Mehrling argues that cash flow matching is the survival constraint that is often lost in treating money as one of many assets or one of many goods.(Though he also argues that medium of exchange is a category error, for it blurs the line between means of payment and promises to pay. The two differ for different entities that operate at various levels in the hierarchy of money.)

My contention with the inventory of currency is far simpler. Suppose my salary account pays 1%. Suppose I suspect issues with my employer paying me. Do you expect that over the next 30 days I will buffer up my wallet? Or will I just let some funds accumulate in my salary account?

There is nothing special about 0% money(except for the fact that the existence of a 0% alternative makes it impossible for the nominal rate on money to be, say, -5%)

Demand deposits/ current accounts pay interest in many countries. Reserves have paid interest in many countries. But demand deposits are money, and reserves are money. Like I said, how can you hold an expectations theory of monetary policy while at the same time according a special status to currency? If you want to build a general theory of money, whether Wicksellian or Chuck Norris-ian/ hot-potatoist, it should not depend critically on the existence of a 0% alternative.

[Ritwik]

Nick

Also, I think my definition of the LM curve is similar in spirit, if not same, as that of the later Tobin with his three-asset flexible price attempts to converge monetary general equilibrium, asset prices and IS/LM.

He argued, contra Basil Moore, that the endogeneity of the money stock does not make the LM curve horizontal.

[Patrick]

"it's not even obvious precisely what "equilibrium" and "disequilibrium" mean when we are talking about buffer stock/inventories"

FWIW, over the last couple of years of reading your posts on the topic, the analogy that comes to my mind is a network of pipes with reservoirs at each node. Money is the water in the pipes. Stocks of money are the levels in the reservoirs. Flow can be diverted in and out of the reservoir. People have desired flows in and out of their nodes, but sometimes (most of the time?) people can't quite get the actual flows they want, no matter how they fiddle with the valves, level in the reservoir, or even the arrangement of the pipes.

[Nick Rowe]

Ritwik: "If you want to build a general theory of money, whether Wicksellian or Chuck Norris-ian/ hot-potatoist, it should not depend critically on the existence of a 0% alternative."

Agreed, And Perry Mehrling is one of the people I should read more of. (Oh God, there are so many, but he is high on the list).

Patrick: be careful or you will make me sound like a "Hydraulic Keynesian"! ;-)

[Patrick]

Nick: Time to resurrect the MONIAC! BTW, it's an interesting bit of cross-over between computing and economics.

[Ritwik]

Nick

There are many good papers on his page, but you may find these ones particularly interesting.

1) 'Monetary transmission' - [link here pdf NR]

2) Paper introducing the hierarchy of money - [link here pdf NR]

3) On David Laidler (and by extension, you , I presume) - [link here pdf Yep! I just read it, very good paper NR]

4) On Woodford 2003 - [link here pdf NR]

[W. Peden]

On Woodford 2003 - here's the charmingly titled article "The Quantity Theory of Money is Valid - The New Keynesians are Wrong!". That's what's called "wearing your heart on your sleeve".

[link here NR]

I'm sure there's a lot to be said for Woodford's work, but based on the extremely partial account in this article I almost find it plausible that Woodford's research programme is the result of a drunken bet ("Hey guys, if I can't drink all this vodka without throwing up, then I'll re-write monetary theory without money having any importance!"). I like this article, even if I hate the use of 'valid' to mean 'sound'. (Deductive validity is very cheap.)

[Nick Rowe]

W Peden:

(They are German? so maybe just got the English words muddled.)

I skimmed it, but I can't figure out what the empirical test was. I don't think it's that the Cambridge k is constant across countries. It must be that the Cambridge k is (roughly) constant across time?? But it isn't *very* constant across time. They lost me at equation 2.5. What's so special about k at time T (the end time in the data series)?

[W. Peden]

Nick Rowe,

I'm not in a position to defend the details of the article. As for the wording issue, I would say so, were it not for the facts that (a) educated Germans are often more precise in their word selection in English than native speakers and (b) it's a very common mistake among native English speakers. The most useful thing that most philosophy students learn in a formal logic course is the soundness/validity distinction.

[Alex1]

Nick,

In regards to Wicksell exploding with flexible prices and wages, may I suggest that you are overthinking the issue a bit.

Lets put Wicksell in his context. Lets forget IS-Lm/AD/AS and all that stuff. We are with flexible prices, so why would we bother with that stuff. Lets go old classical. We have an aggregate production function with the usual classical dimnishing returns. We have a labour demand and a labour supply, both dependent on the real wage, and the L=N equilibrium condition. We have a real capital market (in which both savings and investment are independant of income - that's crucial) and for all things monetary we have the quantity theory. Now we introduce Wicksellian loanable funds as S+DH+dM=I. Next we assume a positive investment shock and the banking cartel continuing to lend at the old (now too low) rate, thus expanding the money supply. What happens? Well, the usual dynamic explanation is that investors use the new money to divert production from consumer to capital goods. But we have a real wage based and flexible labour market, so the influx of new money will not only raise prices (as demanded by the quantity theory), but also wages, thus keeping the real wage constant and output completely untoched by the strange wicksellian workings. By extension the pull into capital goods will be offset by an opposite pull into consumption goods when the workers get their new wages. If this is simaltenous we get a non-exploding economy, which has inflation proportionate to money supply growth, consistent with the basic quantity theory and the classical dichotomy. The Austrian addendum is that pull and counterpull are of course not simoultaneous, thus we get the same basic conclusions, but we also get relative price distortions.

Once you invoke LM or anything comparable, the model of course becomes either indeterminate or overdeterminate. Why would you take poor old Mr Say to the money markets? He really has no business there and all these veils just confuse him

[Nick Rowe]

Jon: "There are only two sticky behaviors in this model: the market rate of Interest which is pegged by the banking cartel and the price level expectations which remain anchored."

I said I would think about this slowly, and then return to it. OK.

Assume all prices are perfectly flexible. Assume that the current expectation of the price level 10 years from now remains fixed. Let's not ask how.

Start in equilibrium. Then assume the central bank and banks cut the nominal interest rate by 1%. There is now excess demand for output at the existing price level. So the price level immediately jumps by 10%. So people expect 1% deflation per year. So the real interest rate is back to where it was before. Immediately.

[Nick Rowe]

Alex1: am I now saying the same thing you are?

[Ritwik]

Nick

Your conclusion is being driven not by flexible prices/ rational expectations, but by homogeneous expectations/ representative agent/ full information. You cannot get Wicksellian/Phelps-ian results in a Lucasian world.