Saturday, 28 July 2012

What is New Keynesian Economics really about?

For economists. This post arose out of remark I made in response to comments. I’m not sure I want to defend the original remark too much, but I think there is a discussion here that some may find interesting.

In my discussion with heterodox economists, I was asked by someone whether I had actually read any major heterodox thinkers, or just looked at blogs. In response I listed some of those that I had read, including Axel Leijonhufvud. In passing I said I thought you could make a case for Leijonhufvud being the first New Keynesian economist. I guess this was a red rag to a bull, and Lars Syll wrote another post saying how wrong I was. (He is fast overtaking Scott Sumner in the perceived Wren-Lewis error count.)

So what did I mean? Of course I knew that Leijonhufvud is highly critical of much of modern macro. But what I had in mind was what I remembered of ‘Keynesian Economics and the Economics of Keynes’, which I read when young and which made a big impression. (For a more recent assessment, see this review by Howitt.) At the time there were attempts being made to partially microfound Keynesian economics by looking at rationing regimes where either goods or labour markets did not clear. Leijonhufvud was critical of this, and the earlier, emphasis on wage and price rigidity as being at the heart of Keynesian economics, both in terms of an interpretation of Keynes but also as useful macroeconomics. What I also remember, but did not fully appreciate at the time, was a good deal of discussion of the importance of intertemporal coordination failures, in which people’s expectations about long-term interest rates differ from the marginal efficiency of capital.

For many, New Keynesian economics is just a microfounded version of Old Keynesian economics. But what that means in practice is that New Keynesian theory is explicitly intertemporal, which IS-LM is not. Another way of describing New Keynesian theory, which I have used myself, is that it is a RBC analysis with price stickiness added on. In a technical sense that may be true, but I think it can be misleading in that it perpetuates the idea that these models are ‘all about’ wage and price rigidity.

Now it may or may not be the case that wage and price rigidities are preconditions for Keynesian analysis. I have argued elsewhere that a better way of saying it would be that price flexibility coupled with an appropriate monetary policy may rule out Keynesian problems, but that at the Zero Lower Bound (ZLB) inflation targeting does not. However it is also plausible to argue that the existence of money is a precondition for effective demand to matter in determining output, but not many people say that money is what New Keynesian analysis is all about. (Particularly as money is often not even part of the model.) You could also argue that imperfect competition was an essential ingredient in New Keynesian analysis, but again not many would characterise New Keynesian analysis as the macroeconomics of imperfect competition.

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. The natural rate is the real interest rate that would obtain if output was determined by an RBC model – less precisely, if output was determined from the supply side. We often say the natural rate of interest is the real interest rate that would occur under flexible prices, but that is shorthand for the above, and perhaps misleading shorthand on occasion.

Once we have the ‘wrong’ real interest rate, then (using imperfect competition as a justification) New Keynesian analysis determines output and perhaps employment only from the demand side, and the determination of effective demand becomes critical to the model. Perhaps a better way of saying this is that if real interest rates are at their natural level, we do not need to think about demand when calculating output. In most cases, it is the job of monetary policy to try and get the economy back to this natural real interest rate. This gives you the key insight into why, ZLB problems apart, it is monetary rather than fiscal policy that is the primary stabilising policy.

We can make the same point another way. In the New Keynesian model, business cycles are generally an intertemporal mismatch between demand and supply (unless, for some reason, we get stuck with demand deficiency). What is the relevant price that influences this intertemporal allocation of demand? – the real interest rate, not the price or wage level. So if we have to think about New Keynesian economics as being about some price being wrong, that price is the real interest rate. And thinking in an explicitly intertemporal framework also makes you think twice about adding M/p into an IS curve, which is partly where the idea that sticky prices are crucial comes from.

Of course at some level the idea of characterising a class of models by one key idea is a bit pointless. But at another, more intuitive, level I think it can be helpful. In this case, for example, it suggests why monetary rather than fiscal policy is the policy of choice for stabilisation. This is also a good example of where IS-LM is less revealing: in this framework both policies look as good as each other within the context of the model itself.

This is why I claimed a connection between some of the ideas in Leijonhufvud’s Keynesian Economics and the Economics of Keynes and my interpretation of New Keynesian economics. I could now make sense of Leijonhufvud’s emphasis on talking about real interest rates, rather than wage and price rigidity. But I would be the first to admit that it is easy to see in what other people write what you want to see, so I have no problem with other people disagreeing with me on this interpretation. And even if there is something in what I say, I know of course that Leijonhufvud is highly critical of other aspects of New Keynesian analysis, and that his views may well have changed since that book. But what I do hope is true is that my interpretation of New Keynesian analysis is of some help to those who struggle to see what the nature of the market failure is in some modern Keynesian analysis.    


  1. Simon, I am wondering if you are so deeply steeped in the neoclassical paradigm that you just cannot comprehend how your world view is preventing you seeing just why mainstream macro theory cannot explain why the global economy is in mini depression. For instance, you talk about categories of "market failure" - prices and wages or intertemporal prices that deviate from some Walrasian benchmark. I do urge you to have a look at Robert Clower's 1990s speech to the Southern Economic Association where he fingers what I am getting at - if you are brought up in that Walrasian paradigm, it precludes you from ever figuring out how the macro economy works. His tentative solution was to turn to agent based modeling - I think he did some work with Peter Howitt.
    Do you know the work of Brian Loasby at U of Stirling ? Much of his fantastic but dense oeuvre also helps one to at least see why Walras and modern macro is a dead end.

  2. There are many brilliant ideas in economics that are unfortunately labelled as 'heterodox' because of the status of the person who came up with them. The terms 'orthodox' and 'heterodox' are never used in any hard Science - just in Religions and Economics. I think that is telling.

    A list of important 'heterodox' ideas:
    Loans create deposits
    The money multiplier model has the causation backwards
    Credit = debt
    The change in the stock of credit/debt affects the flow of credit in circulation and hence aggregate demand
    The ratio of debt:income matters for individuals and for the economy as a whole
    Balance sheet recessions / Private sector deleveraging
    Equilibrium models have little predictive value
    The central bank can control interest rates on bonds of any maturity.
    A sovereign currency issuer need never go bankrupt.
    Inflation is the only restraint on fiscal policy.
    Government borrowing is not inflationary when the private sector is paying down more debt than it borrows.

    1. Honestly the only people I see use the term 'heterodox' frequently are the 'heterodox' economists; it seems like that label is willingly self imposed. I had never heard the term before outside of the blogosphere, of course I had come accross papers and ideas in my classes/lecture notes that might now be labelled as 'heterodox', but back then they were just another approach to conventional economic problems, they were never prevented as in anyway sacrilegious.

    2. presented*

  3. Off-topic:

    BT London: "A list of important 'heterodox' ideas:
    Loans create deposits..."

    Oh God! Can some of you soi-disant "heterodox" guys actually read a first year textbook?? The idea that loans create deposits is in every (OK, there must be some exceptions, but I can't think of any) first year textbook. It's as old and orthodox as .... a Greek archbishop! Hilariously, it's usually introduced at the same time as the money multiplier model, which is a model that many (maybe most?) orthodox (e.g. New Keynesian) macroeconomists think is hopelessly wrong (because it has the causation backwards).

    "The terms 'orthodox' and 'heterodox' are never used in any hard Science - just in Religions and Economics."

    And I almost never hear them in economics either, except on the internet. And the people who use those words all the time seem almost always to be people who describe themselves as "heterodox".

    Sorry Simon. Back on topic. Interesting post. Yep. It's one of those subjects on which I can say "I disagree", without wanting to say "you're wrong". More than one perspective can be OK on this, like blind men and the elephant. I would emphasise more the "price-setting" idea, as one of the two strands coming out of the '69 Phelps volume (the Lucasian supply function being the other). And imperfect competition was integral, because it is the only way to make sense of a model in which quantity sold is demand determined (as opposed to Q=min{Qd,Qs} in the short run with sticky prices. And I see the Wicksellian stuff you emphasise as coming much later, as an add-on.

    For example: take a canonical NK model. Throw away the Euler-IS curve, and replace it with (say) a 2-period OLG model and MV=PY. But keep Calvo pricing and monopolistic competition. I say it's still New Keynesian. You still need "activist" monetary policy.

    Must collect my thoughts on this.

    1. "Oh God! Can some of you soi-disant "heterodox" guys actually read a first year textbook??"

      Nick, you seriously won't get anywhere with an attitude like this. The idea that heterodox economists have not learned economics or don't understand is simple arrogance.

      My first year textbook contains absolutely nothing of the sort:

      "In any discussion of the creation of bank deposits
      it is customary to begin by recognizing that not all of
      the funds deposited with a bank will be withdrawn at
      any one time. Indeed, under normal circumstances inflows and outflows of funds will be such that on any one day banks will require only a fraction of the total funds deposited with them to meet withdrawals by customers. This implies that the remainder can be lent to borrowers. But this is not the end of the story
      because funds lent by one bank will flow back into the banking system, again a fraction will be retained and the remainder will be available for lending to other borrowers. This process is known as the money supply multiplier."

      I won't further the derail by explaining why this is clearly wrong, but suffice to say heterodox economists are not straw manning you.

      If economists really incorporated endogenous money, they would find it easy to see that credit HAS TO expand more aggregate demand to increase, and debt therefore ADDS to income rather than redistributing it.

      But from the Keen/Krugman debate it is clear that you do not believe this.

    2. Unlearning: let's continue this on my blog, so we don't disrupt Simon's. You will immediately see where.

  4. I like your characterization of the market failure as i not being equal to its 'natural' rate, but I'm puzzled by your claim that this shows that "monetary rather than fiscal policy is the policy of choice for stabilisation". Doesn't the IS-LM insight that "both policies look as good as each other within the context of the model itself" apply equally to NK models? If the goal is to get i equal to its natural rate, one way to do it is to change that 'natural' rate itself by using fiscal policy. So if i is below its 'natural' rate then you can cut government spending to lower the 'natural' rate itself. Within the model, either you correct the market failure or you don't. Within the model there's no such thing as a "policy of choice". Or am I missing something?

  5. I am fascinated by the macroeconomic discussions here and on related blogs. Is there some rule against using mathematical formulas? How can one argue precisely without writing down the models using mathematical notation?

  6. "The terms 'orthodox' and 'heterodox' are never used in any hard Science - just in Religions and Economics. I think that is telling."

    Economics is a mono-paradigmatic discipline - either you're with the mainstream paradigm, or not. Other disciplines have multiple paradigms, e.g. one can be a realist or a feminist or a post-Marxist or a structuralist or whatever. Where there are multiple paradigms, people don't end up defining themselves as orthodox/heterodox in the same way.

  7. Prof. Simon, I think you need to re-think the assumption that the "natural" risk-free rate necessarily has anything to do with the return on capital.

    The risk-free rate essentially amounts to paying people a subsidy for not-investing (Keynes had a pithy and apropos quip about piling up claims to enjoyment that one does not intend to exercise at any particular time).

    I don't think you'll find any economist who argues that paying a European farmer for not-growing crops brings the European grain market closer to the "natural," or barter, clearing price. In point of fact, the entire public purpose of the Common Agricultural Policy is to make the grain market depart from the barter economy and secure its place in the industrial economy.

    So why does paying capitalists a subsidy for not-investing his money bring the capital market closer to the barter state?

    - Jake

  8. Jake,

    "the assumption that the "natural" risk-free rate necessarily has anything to do with the return on capital."

    I don't really see "subsidy for not investing" thing, but in any case in that formulation it obviously has something to do with the marginal return on capital - if the marginal return to capital is 20% or 2% it's obviously going to have something to do with the size of the "subsidy" people are going to need "not to invest". If European farmers made huge profits growing crops they'd need paying more not to growth them wouldn't they.

    1. Isn't the problem here that "the return on capital" (even more so the "marginal return on capital") is a quantity that doesn't actually make any sense and can't be given a non-circular definition?

      This would actually be my candidate for a definition of orthodox vs heterodox macroeconomics at present - whether one takes the Cambridge Capital Result (it isn't a "Controversy" and hasn't been one for thirty years) seriously, or whether one continues to use models in which there is a "capital stock" and a "marginal return on capital".

    2. I think models in which there is a "capital stock" and a "marginal return on capital" are useful fictions, or highly stylized formalizations of something real, but I'm open to arguments that they are seriously misleading. And I think you have to treat these models carefully and not think they tell us more than they do.

      We know production does involve capital goods, and that there is a stock of capital goods, and that something will happen to output when more capital goods are created and put to work, but trying to model all that precisely is a real can of worms and I think a case can be made for taking a some very broad strokes, such as using amount of money spent on capital goods to construct a measure of capital, even knowing that really the price of capital goods cannot be separated from the return on them.

      To my mind it makes perfect sense to say that in some economies at some times the economy-wide marginal return to capital is high - thinking of something like an average, meaning just "there are lots of good investment opportunities out there" - an example might be an under-developed country that has just experiences a positive political regime change, whilst another economy may be already very capital intensive where, looked at as an aggregate, the marginal returns on adding more capital to the economy are lower. I think that's a good description of reality and I think simple models can get at such things (I hesitate to say I have any expertise in anything, but if I aspire to anything, it's to be a long-run development macro-economist - think Rodrik - so that's why I think of aggregate models with capital stocks like that).

      Business cycle frequency stuff I know a lot less about, but I imagine there you have to think about spare capacity and expected demand dynamics and the whole business of what the marginal return on capital is gets even more fiddly. But I don't think you can escape having to think about it, and I still think that making some pretty brutal simplifying assumptions might be justified.

      Incidentally, I have never looked into the CCC (or CCR) in enough detail (life is too short) but there are various approaches to capital that are within the mainstream - say, vintage capital models, or capital as embodied technology - I don't know whether they would be more satisfactory to heterodox critics.

    3. The risk-free interest rate is a subsidy to not-investing because it allows people to place their equity in credit instruments rather than productive investment. A bond is a mixed credit/equity instrument (with the equity proportion given by the degree of overcollateralization demanded for rediscount operations in that bond). Paying a non-zero interest on the credit part entices owners of equity (which is not free) to waste their equity doing something that the discount window (which is free) can do just as well.

      Or, put another way, if the marginal risk-free return on equity is non-zero, you are not exploiting all the gainful investment opportunities available to your society at that particular point in time. Paying a risk-free interest rate ensures that this underinvestment is chronic.

      - Jake

  9. This comment has been removed by the author.

  10. "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 agree with you that the key feature of NKE is its explicitly intertemporal nature, but if I had to characterize it by one thing, it would be gradual adjustment of prices and wages. Maybe that is just the American version of NKE

  11. Normally Fischer and John B. Taylor's staggered wage models are seens as the first in the New Keynesian tradition (rational expectations, and natural rate cum imperfections). I think Rudi Dornbush's overshooting model precedes those, and should be seen as the first NK model. My problem with all these models (New Keynesians) is the acceptance of Friedman's natural rate hypothesis, for which there is no reliable empirical evidence in my view.

  12. Simon: good post. I offer a different perspective. But one where Leijonhufvud also plays a role!

  13. I highly recommend this post by Robert Vienneau (, which is illuminating about how some mainstream authors deal with heterodox perspectives. In that sense, I should say, that while I disagree with Simon, I think he tries to honestly incorporate heterodox insights into his New Keynesian framework. I would appreciatte a comment on your views about the natural rate.

    1. I think my main point is that few theories in macro are either absolutely true or absolutely false. It would be a shame if mainstream economists were so constrained by rational expectations that they could not examine the consequences of persistent mistakes. But it would be equally foolish if heterodox economists thought that rational expectations was never useful. There should be no great divide.

      I think the same about the natural rate. Do I think there is a concept out there called the natural rate that is important in helping understand inflation - absolutely I do. Do I think it is always and everywhere independent of demand side influences - absolutely not. Indeed, one aspect of this last point may be crucial in understanding what is happening in the UK (and maybe elsewhere) at the moment.

    2. Thanks for your reply. Mind you some Post Keynesians (PK) have dealt with Rational Expectations. Tom Palley for example uses it in his book on PK economics (see for example this paper

      The question of the supply constraint, which I agree might be important to understand inflation, is that once you accept that it can be influenced by demand, then in the policy relevant range of possibilities, at least in developed countries with no balance of payments problems, it's been really a long time since we have been even close to that limit. Certainly in the US and the UK there is an incredible amount of space for fiscal expansion without any serious risk of inflation.

  14. 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. ... Perhaps a better way of saying this is that if real interest rates are at their natural level, we do not need to think about demand when calculating output.

    I have no problem with this as a description of New Keynesian economics but this view was explicitly rejected by Leijonhufvud. His argument is that there are many possible equilibria with self-fulfilling expectations, each with a corresponding natural rate. So there is no way to define the natural rate independent of demand.

    "In the depressed state which is typically the Keynesian concern, the illusion is that [people in the aggregate] are insufficiently wealthy to purchase the full employment rate of output. ... The condition that the excess demand for'"bonds' be zero does not insure a 'correct' value for the interest rate. ... It is the very essence of Keynes' argument that the interest rate cannot be relied upon to 'equate' saving and investment (at full employment) and that bond markets may clear at a wide range of interest rate-employment combinations." (KEEK, p. 276)

    Or more pithily, "If expectations are not approximately right, there is nothing 'natural' about the natural rate."