Tuesday, January 25, 2011

The Keynesian Cross Crusade

After complaining about Shiller in my last post, I might as well continue while I'm in the mood. I just read this Krugman blog post. This reminds me of the annual Fox News stories about the War on Christmas. The Keynesian Cross is as shrouded in myth as Christmas, and about as likely to go away, unfortunately.

According to Krugman, here's the problem:
Basically, in the face of what I would have said is obviously a massive shortfall of aggregate demand, we’re seeing on all-out attack on the very notion that the demand side matters.
As I've said before, on more than one occasion, "shortfall in aggregate demand" has no meaning, either in fully-articulated New Keynesian models, or any other modern models of macroeconomic activity. If the statement has no meaning in terms of the theory, we can't observe the shortfall empirically either. Of course, Krugman is thinking about a Keynesian Cross model, which is not what modern macroeconomists (of any stripe) work with. If you buy into the Keynesian Cross, what recession would not be a shortfall in aggregate demand?

Then, we get:
This isn’t entirely new, of course. Real business cycle theory has been a powerful force within academic economics for three decades. But my sense is that the RBC guys had very little impact on public or policy discussion, simply because what they said seemed (and was) so disconnected from actual experience.
You could actually say more than this. Real business cycle theory is based on work on economic growth by Solow, Cass and Koopmans, which stretches the "powerful force" back more than five decades. Real business cycle theory, in its original form was hardly "disconnected from actual experience," indeed part of the influence and strength of the approach came from its quantitative nature and the attempt to confront models with the data. Now, you won't find many practitioners today who actually use the RBC framework in its original form. However, what it became was New Keynesian theory (and applications), with the addition of sticky wages and prices with monopolistic competition; and any number of branches of the macroeconomic literature (asset pricing, credit markets and bankruptcy, for example) that use competitive heterogeneous agent constructs. To say that the "RBC guys" had no influence on policy is false. People who I might think of as RBC guys are now running Federal Reserve Banks, for example, and the ideas permeate policy discussion.

Krugman seems to think that people are out to attack the Keynesian Cross, but that certainly is not anyone I know. Most of them don't think much about it. Apparently the attacks are coming from the commenters on Krugman's blog posts. Well, good for them! Krugman's defense is that the Keynesian Cross has been in the Econ 101 textbooks for 62 years. I have also heard that a majority of Americans believe that angels exist.

Krugman goes on by resurrecting his babysitting co-op story. To me, the lesson from this story is that the designers of this co-op did not do a good job. About 25 years ago, my wife and I participated in a babysitting co-op in Minneapolis that seemed to work fine, without anything that looked like Keynesian intervention. I don't get it.

Anyway, we know how this will finish off. Krugman will be alarmed - but sad.
Seriously, though, this is truly sad — and dangerous. Demand-side understanding, in my view, played a big role in helping us avoid a full replay of the Great Depression; if enough people had shared that understanding, we might have avoided even the minor-league Depression we’re going through. But willful ignorance is on the march — and the odds are that we’ll handle the next crisis very badly.
I think Krugman and I have different ideas about where the ignorance resides.

Addendum:
1. People who think that Old Keynesian (or more accurately, the Hicksian interpretation of Keynes) economics has something to say typically argue that the "demand deficiency" is obvious by looking out the window. But the fundamental source of inefficiency in their framework comes from sticky wages and prices. To make a good case that Old Keynesian forces matter for recent history, the Old Keynesians should be showing us direct evidence of the sticky wages and prices, as they relate to the current state of the economy.
2. What I think Krugman is worried about is that, if the non-Keynesians are correct then somehow we no longer have a role for the government. Nothing could be further from the truth. Not accepting Keynes as your savior does not mean that you think the government's role should be very narrowly defined, that you don't care about the poor, or that you are a Republican.

58 comments:

  1. "Of course, Krugman is thinking about a Keynesian Cross model, which is not what modern macroeconomists (of any stripe) work with. If you buy into the Keynesian Cross, what recession would not be a shortfall in aggregate demand?"

    This is silly. Why can't you have a supply shock in a Keynesian cross model? It's just income, expenditures, quantity consumed, and quantity invested. There's no aggregate demand in it to speak of. Let's say there was an oil supply shock, reducing oil consumption. How would that show up in the Keynesian cross? Wouldn't it look EXACTLY like an oil demand shock?

    The Keynesian cross isn't an illustration of demand driven recessions - it's an illustration of the concept of the multiplier.

    And why do you even think Krugman has that in mind here?

    ReplyDelete
  2. Daniel,

    What Keynesian cross have you been looking at? In the one they taught me, there was no production. This was it:

    Y=C+I+G
    C=A+cY

    I, G, and A are exogenous, and c is a parameter, with 0<c<1. Where's the oil?

    I think this is what Krugman has in mind because he refers to what has been in the Econ 101 books for 62 years - that's the above.

    You can throw in the LM curve if you want, and I know Krugman is also a staunch defender of IS-LM - not much difference.

    ReplyDelete
  3. Where is the demand in those two equations, Stephen? All I see is quantity consumed and quantity invested. Quantity consumed is the assumed to be the equilibrium point of quantity demanded and quantity supplied - it's as much about supply as it is about demand.

    I and C are composed of something. One thing that certainly enters both is oil (or whatever else - that was just one product I picked).

    If you have a supply shock it's either going to raise oil prices or lower oil ouput or both, depending on the elasticity of demand. Let's assume there's some of both since no specific elasaticity of demand is implicit in the Keynesian cross. What does this decline in oil output - resulting from the oil market equilibrium after experiencing a supply shock - look like in a Keynesian cross?

    b probably wouldn't change I'd guess - A would shift down, and I would shift down. It looks the same as a demand shock in the Keynesian cross because the Keynesian cross is about income and expenditure levels - not the distinction between supply and demand. The whole little illustration is just an identity, E=I, designed to demonstrate multiplier effect when 0 < c < 1. That's all it is. You have consumption and investment. Is it "consumption demand" or "investment demand"? No, it's not. It's an actual consumption level and investment level. "Consumption demand" is a relationship between price and quantity, not a number. C and I are the intersection of Cs and Cd, and Is and Id, respectively.

    This is how the textbooks use the Keynesian cross - to illustrate the multiplier, not to say anything about exclusively about demand. The Keynesian cross can only say something about demand if you bring a demand story to it (such as IS-LM, as you suggest). But you can just as easily bring a supply story to the Keynesian cross.

    ReplyDelete
  4. How do you even talk about demand in a Keynesian cross?

    You talk about it by saying "oh, I goes down, now look what happens to everything else" or "oh, A goes down, now look what happens to everything else".

    I and A shift exogenously (exogenous to the Keynesian cross, that is). Why can't the exogenous shift come from a supply shock as easily as from a demand shock?

    You're trying to get far more mileage out of the Keynesian cross than it offers, and then you're turning around and attributing that attempt to Krugman, and then you're coming back and blaming it all on the poor little Keynesian cross that never claimed to be anything but an illustration of the multiplier concept in the first place!

    ReplyDelete
  5. Daniel,

    Again, I don't know who you have been listening to, but what I hear from Krugman, or Brad DeLong, is entirely consistent with what I was taught in Principles of Economics in 1975. The way this was described to me, in terms of the above, was that A, I, and G, were the exogenous components of aggregate demand. Some of the current intro books will go straight to some AD/AS construct, but the AD is coming from the same place, with the LM thrown in. By the way, this always ends up happening to me. I have to explain the Keynesian models to the Keynesians.

    ReplyDelete
  6. Steven you are quite right about your interpretation.
    But you can think of production taking place in this economy.
    The production function is just linear in labor and the number of firms of measure 1.
    Y=L.
    The nominal price of goods is fixed, P=1, and the firms commit to supply whatever is demanded at the fixed price.
    Modern New Keynesian model look pretty much like this too. But there the consumption function is derived from household maximization, and instead of being fixed for ever, firms prices are adjustend once in a while, depending on the assumptions.
    Then the details of why prices are fixed differ accross models,some just assume it, other derive it from the fact that when there is a lot of strategic complimentaritie in price setting, each firm don't want to deviate from the rest of the pack too much and you can get somewhat sluggish pricing behavior with relatively modest extra tricks.

    ReplyDelete
  7. Yes, exactly. There has to be some notion of output being demand-determined, and firms are willing to produce whatever is demanded. But it's a little funny. You want the supply of goods to be perfectly elastic, but with respect to what? Keynesian cross is a real model. You don't want money in there, I think. I think it works if output is linear in the labor input, and labor supply is infinitely elastic, at a real wage lower than the marginal product of labor. Then, the demand for goods as determined in the Keynesian Cross gives the demand side of the market. Demand is perfectly inelastic with respect to the real wage, and supply is perfectly elastic, so output is demand-determined. Strange little model to be teaching to first-year undergraduates. Woodford is in some sense more straightforward.

    ReplyDelete
  8. I don't really think of the Keynesian cross as real. I sort of think that the old-timers thought of real and nominal as being the same in the short run and kind of made it (sometimes) explicit in their writing. So its really P=1. Meanwhile, people Hicks were very well aware of the difference between real and nominal, but were just content with making the dirty little assumption that in the short-run that distinction was not crucial due to price/wage rigidity.
    The big change that happened with the New Keynesian economics was to incorporate expectations, and then inflation expectations start playing a key role and the distinction between real and nominal becomes fundamental.
    I think you are a little bit off base about how to think about the current crisis in terms of the New Keynesian models by emphasizing the price rigidities in this particular way in your "PS". The way I think people think of this is that there are some big real shocks that hit the economcy (which are nowadays sometimes explicitly modeled by incorporating financial frictions) and then relative prices (say price of things today relative to the future, i.e. the real interest rate) needs to adjust a lot. It models where firms have a lot of strategic complimentarities in their pricing it is very hard to bring about that sort of change in relative prices (even if you cut the nominal rate to zero) due to some sort of coordination failure that is sometimes modeled by exogenously imposing that firms cannot change their price exept at stochastic dates, but also sometimes modeled explicitly with menucosts. The point of the menucost models is that you need really just tiny menu costs in models where there is lots of strategic complimentarities to bring about big time coordination failures that then have big real effects.

    ReplyDelete
  9. "Strange little model to be teaching to first-year undergraduates. Woodford is in some sense more straightforward."

    This pretty much summarizes why I didn't like my freshman macro course back in undergrad. A lot of it requires some really tortured logic to make any sense at all. Frankly, I think we do our students a disservice by teaching this kind of stuff to them.

    ReplyDelete
  10. anonymous:

    Yes, I was pretty much sticking to old-fashioned Keynes here. New Keynesian economics is very different. Typically, in that paradigm, people are not thinking about strategic complementarities and multiple equilibria - that was the province of Bryant, Peter Diamond, Cooper and John, and later dynamic sunspot equilibrium business cycle models (Benhabib, Farmer, Woodford). I think I've seen New Keynesian multiple equilibria (equilibria with different self-fulfilling expectations about future prices), but can't remember the paper. However, that's not what Woodford and company tend to focus on these days.

    Joe,

    Yes, exactly. At some level it's very simple, but sometimes a good student can be totally confused, and for good reason.

    ReplyDelete
  11. Stephen, I think they (Woodford et al) are not focusing on the coordination failure thing because they feel they are standing on the sholder of a giant (or midget, depending on your perspective). So what the literature from the 80's and 90's showed is that you don't really need a lot of sand thrown into the wheels of the economy (that is very little cost of changing prices) to make it a rational decision by firms to change their prices very little in response to shocks (assuming others are not changing at the same time). That was sort of Mankiws original point in "small menu costs big welfare implications" QJE paper. So they assume e.g. Calvo prices because they feel confident that they don't need a lot of "menucosts" to justify staggered prices if there is enough strategic complimentarities. In other words, the sticky price thing won't have too big effect on the firms profits (its not that profitable to deviate from the fixed price thing). It's not really about multiple equilibria, its is a about slow adjustment of prices of firms that are committed to satisfy any demand. And this, Dr. Watson, has huge impact and can very well explain situations such as the one we are in today.

    ReplyDelete
  12. Hey Steve, are you talking about the dynamics indeterminacy of Taylor rule, a la Benhabib-Farmer RBC model? Here is the criticism by John Cochrane:

    http://faculty.chicagobooth.edu/john.cochrane/research/Papers/taylor_rule_jpe_revision.pdf

    Essentially determinacy in NK model requires gov't commitment to "blow up" the economy once deviated from the targeted path. Cochrane shares similar criticism with Atkeson-Chari-Kehoe's argument for NK lack of sophisticated monetary policy, once upon a time you have told this one before :)

    rus

    ReplyDelete
  13. Hi previous anon, I am afraid I want to ask: why should we believe menu cost is a way to start? Even Mankiw himself is aware menu-cost models cannot captures some aspects of reality well, like inflation persistence. On the other hand, Golosov-Lucas JPE 2007 show that, if we believe in menu-cost, then we should not believe in any monetary policy that is helpful to push up the economy.

    rus

    ReplyDelete
  14. Golosov and Lucas forget to include strategic complimentarities. Gertler and Leahy introduce them in a follow up JPE article and overturn their result.

    ReplyDelete
  15. Hi previous anon, thanks for pointing out Gertler-Leahy JPE 2008, it makes the point sharper- menu-cost model itself (as state dependent pricing) is not enough and much replies on strategic complementary to replicate the nonneutrality of money.

    I am not try to say we should use time-dependent pricing instead; my worry is the menu-cost argument to justify sticky price in previous previous anon. Mankiw, who advocates menu cost model in 80s, has switched to sticky info argument for sticky price.

    rus

    ReplyDelete
  16. Dr. Watson anonymous,

    The Mankiw argument applies to all kinds of decisions. There are other results like this in the literature where, for example, the loss from following some suboptimal decision rule gives you a second-order welfare loss. If there are costs to changing prices, there are costs to the firm of making all the other decisions it has to make. Why is it so hard to change a price but so easy to change employment and output? Further, it would be nice to have a better idea of what the costs of changing a price are, so that we can actually measure this directly.

    ReplyDelete
  17. Another thought on this:

    "So they assume e.g. Calvo prices because they feel confident that they don't need a lot of "menucosts" to justify staggered prices if there is enough strategic complimentarities."

    No, they use Calvo pricing because it is easy. Once you get out of Mankiw's static framework, the prices are going to have to change over time - they can't be fixed forever. Thus you have to have some kind of staggered pricing. How do you do it? You could have the menu costs in there, but that is hard. Though, actually you can think of Calvo pricing as a special kind of menu cost technology. Either the cost of changing a price is zero or infinite, but I get access to the zero-cost technology at random.

    ReplyDelete
  18. ^that's right. The point is that you don't need this random cost to be too big if there is enough strategic complimentarity. Hence, they are happy with just assuming the most basic type of staggered pricing, i.e. Calvo.

    ReplyDelete
  19. Anonymous and rus: actually Golosov and Lucas and Gertler and Leahy cite each other, so they acknowledge the results. As a quant guy, I agree with rus that it is real rigidity, which comes from strategic complementary, rather than menu cost that matters for the result.

    Stephen: it seems you don't like menu cost models too, so are you fine with Calvo pricing?

    ReplyDelete
  20. ^^^Stephen: Yes it's true that the strategic complimentarities etc can be found in other contexts.

    The point is this: It came up in this context because of data: To explain facts.

    Ever since Hume, economists have realized that output incerqases if you increase the nominal stock of money. Friedman and Schwarts are all about this too.

    So yes, if there is a stylized fact that has several hundred of years of empirical backing, by all means lets write models to explain them.

    What I've never understood is why people like you seem for some random idiological or historical reason adverse to take into account the very sensible and pedestrian hypothesis of strategic complimentarties and nominal rigidities which seem quite helpful to explain real events. Menu costs are only only one example, but people have used informational costs as well as somebody was pointing out -- at the end all that matters is that that people use "money" as a numeraire and there are costs in changing the prices for whatever reason.

    ReplyDelete
  21. On menu costs and Calvo pricing:

    As I explained, these are basically different versions of the same thing. I don't have a problem with thinking about these things. We want to be able to explain pricing behavior, and that could be important for macroeconomic activity. The difficulty I have is with the unanswered questions. When a firm gets to make a pricing decision, which must it commit to a price until the next pricing decision? Why can't it commit to a growth rate, or commit to something more sophisticated - a contingent pricing rule? Once a firm is committed to a price, it is also committed to supplying demand at that price, which is sometimes suboptimal, ex post. The latter is critical for how the model behaves, and gives you all the output effects, but it is just made up.

    Strategic complementarities:

    I'm all for that. You could argue that this is the key "Keynesian" idea - not the sticky wages and prices. Sometimes this slips into New Keynesian economics (as in Gertler and Leahy), but usually the New Keynesians are focused on the relative price distortions. Sometimes there seems no other good explanation for particular phenomena other than strategic complementarity and multiple equilibria (see for example http://www.jstor.org/stable/2117343). There are some credit market phenomena that look like this - I'm currently working on a problem like that, where the credit market is gummed up with a low level of lending, as non-creditworthy borrowers is a self-fulfilling phenomenon.

    "all that matters is that that people use "money" as a numeraire and there are costs in changing the prices for whatever reason."

    I don't agree. What matters a lot is what money is, why it is used in transactions, how it competes with alternative means of payment, why we write contracts in nominal terms, and why the pricing decisions are made the way they are. You have to build from that base or you can't understand where the macro implications are coming from, and whether we should take them seriously.

    ReplyDelete
  22. "Once a firm is committed to a price, it is also committed to supplying demand at that price, which is sometimes suboptimal, ex post."

    Sometimes suboptimal, but don't most New Keynesian models specify conditions in which it will always be optimal (e.g., firm is a monopolistic competitor, so that price exceeds MC in the neighborhood of equilibrium, and the shock is sufficiently small that we remain in that neighborhood)? In that case it's not just made up.

    ReplyDelete
  23. ^^ One question: What have the "deep" foundation of the monetary frictions -- the "base" --taught us during this crisis? That the Friedman rule is optimal? Just wondering ....
    Meanwhile, seems we have learned a lot from them ad hoc models with the nominal frictions. Just saying....

    ReplyDelete
  24. People who think that Old Keynesian ... economics has something to say typically argue that the "demand deficiency" is obvious by looking out the window.

    Yes.

    But the fundamental source of inefficiency in their framework comes from sticky wages and prices.

    This is one explanation for what they observe out their window.

    To make a good case that Old Keynesian forces matter for recent history, the Old Keynesians should be showing us direct evidence of the sticky wages and prices, as they relate to the current state of the economy.

    If I advance a theory about why it's snowing outside, and you find that theory unpersuasive, is it snowing?

    If the statement has no meaning in terms of the theory, we can't observe the shortfall empirically either.

    Heh. So I guess I have my answer. It is not snowing.

    ReplyDelete
  25. Suppose the Federal Government increases spending in this calendar year by 1 trillion dollars. In your view, will this
    a) raise output but not prices and interest rates compared to what would have happened without the trillion dollars, or
    b) raise prices and interest rates but not output
    c) raise both output as well as prices and interest rates?

    ReplyDelete
  26. As I've said before, on more than one occasion, "shortfall in aggregate demand" has no meaning, either in fully-articulated New Keynesian models, or any other modern models of macroeconomic activity.

    That's just not true, unless you want to quibble about semantics. In New Keynesian models you have an "output gap" which means a shortfall in aggregate demand (if the gap is negative, or excess demand when the gap is positive). The same idea applies to Monetarist or Lucas misperception theories.

    Krugman and old Keynesians put a lot of emphasis on fiscal policy as a determinant of aggregate demand, whereas others tend to focus on interest rates or the stock of money, but it's the same basic idea.

    ReplyDelete
  27. Andy,

    This depends on the specific model, but you can get something like the following. I am a monopolistic competitor, and I can't change my price. Also suppose that the labor market is perfectly competitive (no sticky wages, for simplicity). Now, let D* be the quantity of output the firm would like to supply at the given price. D* is determined in the same way it would be for a price-taking firm (which is what this firm is under the circumstances). But consumers want to purchase D at the given price. So long as D <= D*, it works as in a New Keynesian model. But there is not reason we can't have D > D*. The firm wants to produce less than what is demanded, but the modeler says the firm has to produce D anyway. When would this happen? If you surprise these firms with an increase in the money stock (sorry, no money in here, so Woodford would say it's a surprise reduction in the nominal interest rate target), the firms in the lower tail of the price distribution will look like this. Essentially you need to constrain the firm to meeting demand in order to get the New Keynesian Phillips curve.

    "What have the "deep" foundation of the monetary frictions -- the "base" --taught us during this crisis? That the Friedman rule is optimal?"

    No, we're well past that. You could start with this:

    http://www.artsci.wustl.edu/~swilliam/papers/intermed09-10.pdf

    "Meanwhile, seems we have learned a lot from them ad hoc models with the nominal frictions."

    Such as?

    Snow Anonmyous: I rest my case. Obviously you have accepted Keynes as your savior.

    "Suppose the Federal Government increases spending in this calendar year by 1 trillion dollars."

    Tell me what they are spending on.

    Last anonymous:

    The output gap is the difference between output in a model where all the wages and prices are flexible and output with sticky price and/or sticky wage frictions. Not sure why you want to use that word "demand" in a description of the outcome.

    ReplyDelete
  28. Prof W: Krugman has a post about persistently large output gaps and deflation. Surely this evidence is naturally interpreted in terms of a Keynesian model - evidence of deficient AD. Its hard to explain it in RBC or indeed a New Monetarist model.


    not snowing in texas anonymous

    ReplyDelete
  29. Apparently it never snowed anywhere until someone articulated a convincing theory of the origin of snow. And until that time, anyone who nevertheless insisted on the existence of snow was obviously a mindless religious fanatic.

    ReplyDelete
  30. Snowpeople:

    I think you mean the following: Sometimes we look out the window and see a lot of people searching for work. Therefore, it is obvious that, if the government buys more stuff whenever we see a lot of people searching for work, we will somehow on average (over time and across the population) be better off. That's a brave leap, and I am perennially surprised at the number of people who are willing to take it. I wish it were true. I wish there were a Santa Claus and a tooth fairy too.

    ReplyDelete
  31. No, I said nothing about the government buying more stuff. My question is narrow: does the presence of snow, or demand deficiency, depend on whether you find persuasive any particular theory as to its origin? Because remarkable statements like

    If the statement has no meaning in terms of the theory, we can't observe the shortfall empirically either.

    lead me to believe you think it does.

    ReplyDelete
  32. Here is a coherent statement: I have a theory, which I can show you is consistent with the data. My theory tells me that, when I observe x, then there exists a market failure. The government can correct this market failure, and make us better off in some well-defined sense by doing y.

    ReplyDelete
  33. The output gap is the difference between output in a model where all the wages and prices are flexible and output with sticky price and/or sticky wage frictions. Not sure why you want to use that word "demand" in a description of the outcome.

    Because in the long run supply is fixed at the flexible price level, and changes in demand only affect the price level. Thus, if there is an output gap, where current output is below the flexible price level (characterized by long-run aggregate supply), say, then it must be due to a shortfall in aggregate demand. In other words, in a purely classical (e.g., RBC) model the concept of an output gap is meaningless: aggregate demand only affects inflation, and output is always determined by supply. If there's an output gap, it implies aggregate demand is important, so Krugman's views are broadly consistent with New Keynesian analysis.

    As for why we classify things this way, I guess it's by analogy to micro. An increase in demand (supply) leads to an increase (decrease) in prices.

    ReplyDelete
  34. You're giving me an Econ 101 textbook story. As you say "we classify things this way..." to make it seem easy for a freshman, after you have given him/her only one tool: two curves, one called supply (slopes up), one called demand (slopes down). But we're grownups. We can actually think in general equilibrium terms.

    ReplyDelete
  35. But we're grownups. We can actually think in general equilibrium terms.

    Well, yes, we can do. But if it's simpler to express ideas using basic, Econ 101, terminology, than why not do so? Especially if one is addressing a lay audience, as Krugman is. If output and employment are below their natural levels because prices are too high, then it seems reasonable to me to describe this as a shortfall in aggregate demand, rather than obfuscating matters by digressing into the various theories as to why this may or may not be the case.

    ReplyDelete
  36. It is impossible to observe demand deficiency, which is an object only defined internally to a model. If your theory is bad, your definition is bad.

    ReplyDelete
  37. "But if it's simpler to express ideas using basic, Econ 101, terminology, than why not do so?"

    But it's not simpler. It's actually misleading, and fundamentally confusing.

    ReplyDelete
  38. But it's not simpler. It's actually misleading, and fundamentally confusing.

    Well, you may find it confusing but it's standard economic terminology. Personally, I don't like the output gap terminology, but I woudn't claim it was devoid of meaning.

    ReplyDelete
  39. I've always some problems to think in GE terms. Never understood the concept. Different markets use different time frames which make it impossible to reach any kind of General Equilibrium. Markets are connected, but in a very imperfect way.

    On the other hand - just look at the 2008 international trade statistics, the inventory statistics, the National Accounts of that year - that's what a shortfall in demand looks like.

    And and Austrian version of 'Okun's law'is a very good way to explain why the labor market is not reaching equilibrium: increasing productivity is, again and again, surprising companies who aks prices wich are to high and therefore making unexpected profits while at the same time not using all capacity and labor available - a surprise not ofset by for instance suddenly increasing exports of increasing government demand.

    So, forget about the GE fairytale

    ReplyDelete
  40. "it's standard economic terminology."

    That doesn't make it right of course. What's interesting is how hard it is to stomp out a bad idea in economics. You think Keynes is dead and gone, then someone digs him up, dresses him up in a tuxedo, and brings him to dinner.

    ReplyDelete
  41. ''To make a good case that Old Keynesian forces matter for recent history, the Old Keynesians should be showing us direct evidence of the sticky wages and prices, as they relate to the current state of the economy.''

    Why is the burden of truth on showing that prices and wages do not conform to market clearing? Market clearing prices and wages are a single vector out of a large continuous space of possibilities. Perhaps the burden of the proof should be on showing that prices and wages are indeed such that markets clear?

    And if markets don't clear, then you have rationing or gluts. Isn't that a good, simple way to explain unemployment?

    - RV

    ReplyDelete
  42. That doesn't make it right of course. What's interesting is how hard it is to stomp out a bad idea in economics.

    It's not a question of right or wrong, it's a question of using phrases that have well-defined, informative meaning to people who know a little economics. True enough, one can split hairs about the distinction between the supply side and the demand side, and it's arguably preferable to avoid such terminology in a technical context (although I think describing New Keynesian models in terms of aggregate demand actually clarifies things greatly) but if Krugman's greatest failing was speaking in terms of aggregate demand his blog would be a great deal more informative. In this case, Krugman was guilty of a classic straw-man argument, since the majority of academics and policy-makers do believe the demand-side matters. What Krugman seems to object to is that a few people question this orthodoxy, and many others don't agree with his policy conclusions.

    You think Keynes is dead and gone, then someone digs him up, dresses him up in a tuxedo, and brings him to dinner.

    I don't know why you would think Keynes was dead and gone when so many prominent economists describe themselves as New Keynesian or post Keynesian. In any case, the idea of a shortfall in aggregate demand isn't limited to Keynesians, as I said, the Lucas misperceptions story is really a story of why changes in aggregate demand may affect output and employment, and Monetarists blame a shortfall in aggregate demand (due to fall in the money supply) as the cause of recessions and, indeed, the Great Depression itself. What I don't know, is whether aggregate demand is important in the New Monetarist paradigm. I would assume that it is, but perhaps you can enlighten me.

    PS I forgot to mention before, but in response to your comment "we can think in general equilibrium terms": all macroeconomics is general equilibrium, including the Keynesian cross. Perhaps you meant to say dynamic general equilibrium?

    ReplyDelete
  43. "all macroeconomics is general equilibrium, including the Keynesian cross."

    What do you mean by "general equilibrium"? Walrasian equilibrium with all markets clearing? If so, then the Keynesian cross is not necessarily in general equilibrium.

    -RV

    ReplyDelete
  44. What do you mean by "general equilibrium"? Walrasian equilibrium with all markets clearing?

    I mean an equilibrium (i.e., a state of rest, no tendency to change) for the whole economy, rather than just one particular market, as in partial equilibrium. The whole point of the Keynesian cross model is that the economy can get stuck in a bad, high-unemployment equilibrium, such that it will not return to full employment without some exogenous change (such as an increase in government purchases).

    If market clearing is necessary for equilibrium, then New Keynesian models don't qualify either, because prices are assumed not to change to clear each market. In fact, if you want to be a purist then any model with monopolistic firms is not in equilibrium because the monopolist ensures that Supply < Demand.

    ReplyDelete
  45. When we write down a general equilibrium model, we first specify the environment: the economic agents who live in this fictitious world, the goods and services they consume, endowments, and the available technology. We also need to specify who knows what, and possibly the technology available for observing things that would otherwise not be observed. Then we need to specify an equilibrium concept. Every serious macroeconomic model used today has those features - New Keynesian, old RBC, heterogeneous-agent incomplete-markets, Lagos-Wright monetary exchange, whatever. If you want the model to generate business cycles, there are two ways to go about. Some models (e.g. New Keynesian and old RBC) have a unique equilibrium, and we can get aggregate fluctuations with exogenous stochastic shocks. Some other models (coordination failure models, for example) have multiple equilibria. Sometimes these models have equilibria where there are fluctuations that are driven by extraneous uncertainty (sunspot equilibria). There are various roles for government policy that come out of these models. In standard flexible price models we can think about credit frictions and government debt policies, and optimal taxation. In New Keynesian models we can think about some nonneutralities of money, price distortions, and how those distortions can be mitigated by monetary policy. In coordination failure models we can think about how policies can kill off bad equilibria. Using the words "aggregate demand" is not useful in understanding what those models have to do with reality and what they have to say about how we should conduct policy. You and other people use the words "aggregate demand" and "aggregate supply" because someone long ago thought that would be a useful shortcut in getting some ideas across to undergraduates. Unfortunately, this sometimes requires that those students unlearn some things. Apparently this unlearning is hard for you. Aggregate demand is a crutch, and the sooner you throw it away, the better off you will be.

    ReplyDelete
  46. You could make a very similar argument regarding market supply and demand in a micro context. I wonder if you have the same aversion to that terminology, e.g. "a rise in demand for oil" doesn't tell us if this is due to a change in preferences, incomes etc.

    Krugman favours a certain set of policies including low interest rates, QE, deficit spending, increased government purchases, increased unemployment benefits and higher exports, which are traditionally classified as part of aggregate demand. Indeed, he wants them because he believes they will stimulate demand. He is against cutting taxes to increase incentives to find work or expand business, deregulation etc. He does not believe the recession is due to changes in productivity etc. In other words, he does not believe that issues that are normally classified as affecting aggregate supply are important. It seems to me that using the AD/AS terminiolgy is extremely useful if you view the economy the way he does.

    At the other extreme, I would agree that if you view things from a RBC perspective the terminology is pretty useless, except perhaps when communicating with a lay audience. For New Keynesians, I would say the terminology is potentially useful, potentiall misleading. But being dismissive and patronizing about people using standard economic concepts is pedantic a best and arrogant at worst. Now, Krugman is consistently patronizing and arrogant, but two wrongs don't make a right.

    ReplyDelete
  47. 1. No, for micro partial equilibrium, demand and supply is fine. That works.

    2. "It seems to me that using the AD/AS terminiolgy is extremely useful if you view the economy the way he does."

    It's not "if." He views the economy the way he does because his macro training was essentially AD/AS.

    3. "...if you view things from a RBC perspective..."

    No, you're missing the point. This has nothing to do with RBC or government is good vs. government is bad.

    4. If trying to promote better ideas is dismissive, patronizing, or arrogant, then I guess I would be all three.

    ReplyDelete
  48. So, suppose you have a view of the world where markets don't clear, i.e., you get a lot of people who want to sell stuff at the going price but cannot find buyers (or vice versa) Then, for some reason, everyone wakes up in the morning and decides they are going to DEMAND goods rather than just keep their money in the bank. They will go to stores to buy goods, firms will commission new investment projects etc. Then, so long as the markets where demand increases are in a state where they would be willing to sell more stuff at the going price if only they could find buyers, they will sell more stuff and, if they are short of inventories, may want to increase output employment etc.

    Note that in this view of the world it is perfectly appropriate to speak of aggregate demand. It is something that has to do with how much people collectively *want* goods rather than something to do with their willingness or ability to produce goods.

    -RV

    ReplyDelete
  49. A model economy is a system of equations derived from optimizing behavior by economic agents subject to various constraints.

    The equations in this system are interdependent, so exogenous shocks, say an exogenous government spending shock, eventually will affect all equations when the new general equilibrium is reached.

    The exact general equilibrium can be defined in numerous ways, such as: prices adjust to make sure that all markets clear (... and all constraints are satisfied).

    Everything changes, not just consumption and investment demand.

    KP

    ReplyDelete
  50. Professor Williamson:

    I looked at the paper http://www.artsci.wustl.edu/~swilliam/papers/intermed09-10.pdf that you cited in one of your responses.

    Very interesting and elegant. But as far as I can tell you don't provide any empirical support for the mechanisms that are proposed in the model.

    I am struck by the fact that you and other types that you tend to cite approvingly always argue on a priori methodological grounds.
    The guys you always criticize, particularly the new Keynesians, actually roll up their sleeves and show how their models do when they confront the data.

    Doesn't it take a quantitative modelto beat a quantitative model?

    Everything else is debating how many angels can dance on a pin.

    ReplyDelete
  51. 1. The quantitative work is around the corner.
    2. Debating the theory is certainly not a waste of time.
    3. Don't tell me Keynesians are the only ones doing serious empirical work. That's not true.

    ReplyDelete
  52. 1. Everyone is eagerly anticipating your empirical work.
    2. I agree that debating theory is not a waste of time - it's a useful warm up for empirical work.
    3. What specific empirical work in the new Monetarist lit are you referring to? I do not of a paper of Randy Wright which land up showing inflation can reduce output. When all is said the mechanism is the same as Cooley and Hansen CIA model of 20 years ago? Is that the big empirical victory for `deep' models?

    ReplyDelete
  53. I was actually planning to be an econometrician at one point in my life, but got distracted. You can take a model like the one in my paper and go in a fairly straightforward way to empirical work. Add some aggregate shocks and solve a few problems involving matching what is in the model with the data, and it's not a big deal. Any decent economist could do it, given the inclination. I can understand, though, that playing with New Keynesian models has its attractions. In spite of what you say, the work isn't hard. The technology for solving the models is well-developed, there is ample software available to help you out, and a there are many like-minded people who will referee your papers and get them into journals. That's the easy route.

    ReplyDelete
  54. RV,

    I think the problem is that talk about "demand" takes attention away from the key mechanism which is at the core of the Keynesian story, i.e. sticky wages and prices. Why aren't people saying: "Damn, those sticky wages and prices sure have messed things up since 2008," rather than, "Damn, we have deficient demand." I think the answer is that they think that the idea is easier to sell the second way. No one seems to be pointing out wage or price stickiness in particular sectors of the economy or in particular firms. But if the story is so important, the friction should be obvious. What I'm looking for here is not the usual narrative about infrequent price and wage changes, but specific evidence relating the price and wage stickiness to output in specific sectors or firms.

    ReplyDelete
  55. You make empirical (and publishing) sound awfully easy. A task left for the less creative proletariat. Sounds awfully cynical and, to be honest, self - serving, to me.

    ReplyDelete
  56. Not sure if you are the same anonymous as the one with his/her sleeves rolled up above, but I'll assume so.

    This is how it goes, and you can see it in any number of strands of economic research. (i) Sims writes "Macreoeconomics and Reality," then some guys put together RATS, and everyone can run VARS and produce impulse responses. It's easy. Result: A proliferation of papers. (ii) Kydland and Prescott write "Time to Build and Aggregate Fluctuations," other people learn the computational algorithms and refine them. It's easy, and there is another proliferation. (iii) There may not be a single seminal New Keynesian paper, but this comes out of work by Woodford and others in the mid to late 1990s, and the estimation methods used by, for example, Smets/Wouters and Christiano/Eichenbaum/Evans. Same thing.

    I'm not trying to make out that I'm superior. You were denigrating a literature that I work in, and I wanted to point out that you are on shaky ground. What I see among some New Keynesian researchers (just as in any other strand of economic research) is a reluctance to ask fundamental questions about their paradigm.

    ReplyDelete
  57. I don't mean to denigrate the literature that you are working on. It is full of interesting ideas. At the same time we are social scientists. So issues can't be settled purely on a priori methodological basis. After all one person's notion of deep is another person's notion of ad hoc. It seems to me that the literature which you are associated will only have have a broad impact if you and others show that it does a better job at organizing and explaining the data than than alternatives (RBC, new Keynesian,...). I'm not being aggressive here. It's entirely possible that you can succeed in this task. But do it you have to try.

    ReplyDelete
  58. I agree completely. I'm an admirer of Neil Wallace, and I have learned a lot from him. However, Neil never wanted to get into giving policy advice. He would say something like: "We're not ready for that yet." In the meantime, Mike Woodford was getting a lot of traction with his ideas at central banks. "Deep" monetary economics attracted a lot of people who were not much interested in quantitative work, but there are some people, like Boragan Aruoba (Maryland) who know the monetary theory and can do the empirical work too. See this paper:

    http://econweb.umd.edu/~aruoba/research/paper16/paper16.html

    ReplyDelete