Saturday, August 3, 2013

Saturday Entertainment: What's Krugman Doing with his AD/AS Model Now?

When someone is using the wrong tool, we at least hope that they're using the tool properly. For example, suppose Paul Bunyan wants to cut down a very large tree with a very small hatchet. It will certainly work better if he holds the hatchet by the handle and strikes the tree with the head, and not the other way around. Case in point: this "wonkish" post by another Paul.

Krugman obviously didn't read this post of mine, where I try to figure out what he is up to. You'll see in Krugman's post that his current preferred AD/AS configuration is Figure 3 in my post. AD is steeper than AS, and upward sloping. Here's what Krugman says about his second Figure "AS-AD with ZLB," which is my Figure 3:
Now, the reality is that prices and especially wages are sticky — which is why we don’t see runaway deflation. But that stickiness isn’t what’s keeping unemployment high, it’s just something we have to let into our models to make sense of what we see out there.
The AD/AS model Krugman has constructed is indeed a sticky wage model. If he's thinking about the quantity of output that gets determined as being less than "full employment," then the stickiness is that the nominal wage wants to rise but it can't. You would get full employment with an increase in the nominal wage, which shifts the AS curve left, and increases the price level and output. Output goes up because the increase in the price level deflates the value of private debt, and shifts the IS curve right. So what's going on in the labor market in the background of Krugman's second figure? There is an excess demand for labor. Firms really want to hire workers, but they can't find enough people to work at the market wage. That's a very funny kind of unemployment. Maybe Krugman can explain it to us.**

The other problem is that the Krugman narrative seems to be that we would be in a deflation, but for the wage stickiness that is holding up wages and prices. But what's going on in Krugman's second figure and my Figure 3 is that the increase in prices and wages that would give us full employment is being suppressed. I'm really confused.

I've been trying very hard to understand what Krugman thinks a liquidity trap is. As far as I can tell, price stickiness seems necessary to get it. But the liqudity traps I'm familiar with are summarized nicely in this paper by Cole and Kocherlakota and in a more recent paper by Ricardo Lagos. Those papers are about Friedman rules - monetary policies that will give you zero nominal interest rates forever. The basic idea is that the restrictions on the policies that will give you this are weak. That's a liqudity trap - at the zero lower bound we can alter policy in various ways and it doesn't matter. An interesting result I can get in this paper and this one is that you can get a liquidity trap away from the Friedman rule. That's due to an asset scarcity which makes the real interest rate low. If assets are extremely scarce you can have very high inflation rates at the zero lower bound. In all of those models prices are flexible. Conclusion: Liquidity traps need have nothing to do with sticky prices and wages.

**Addendum: Can't believe I woke up in the morning thinking about this. In the short run in Krugman's figure, the real wage is indeed too high, and there is an excess supply of labor, which is part of Krugman's narrative. The key problem in the narrative for this case is that the flexible price equilibrium has a higher nominal wage and higher price level, so the problem can't be a failure of wages to fall. Further, now I'm wondering how Krugman excludes Figure 4 in my previous post. That's the one where the slope of the AD curve is smaller than the slope of the AS curve and you have excessive aggregate demand. Does he know something about how strong that negative wealth effect is?


  1. Damn! I just wasted a couple of hours writing a blog post on the bit you crossed out! You owe me a beer!

    Assume prices adjust slowly and wages adjust very slowly.

    Draw a vertical LRAS curve assuming both output and labour markets clear.

    Draw an upward-sloping SRAS curve, holding W fixed, where the output market clears but the labour market does not clear.

    Draw a horizontal VSRAS (Very Short Run) holding both P and W fixed so neither labour or output market clears.

    Now draw an upward-sloping AD curve, which may be steeper or flatter than the SRAS curve.

    Assume that P and W fall in response to excess supply in the output and labour markets respectively. This may or may not move P and W towards their equilibrium values.

    In the SR this model may or may not have a stable equilibrium, depending on whether the AD is steeper or flatter than SRAS.

    In the LR this model has an unstable equilibrium.

    I'm pretty sure that's what PK had in mind. He just forgot to draw the LRAS and VSRAS curves.

    1. Hi Nick,

      I was going to put something in there like "where is Nick Rowe when we need him?" This is right up your alley. Nick, you have to understand that this model has nothing to say about dynamics. The best it can do, as in standard textbook stories, is deliver two things: a prediction about happens when the nominal wage is fixed, and a prediction about what happens when its flexible. When the nominal wage is fixed, we're assuming that quantity is determined in the labor market by reading it off the demand curve. When its flexible quantity demanded equals quantity supplied in the labor market. As to how you get from the short run to the long run, that's a story you're making up. There's little enough economic discipline on the static short-run and static long run story, let alone this dynamic nonsense you're throwing in. I can tell you another "dynamic" story where you get from short run to long run through increases in both the price level and the nominal wage and, along the adjustment path, the real wage falls. There's no economics that allows me to choose between your story and mine. If you're claiming your story is consistent with the data, then I can claim you're just cooking your results. This is not a model in which you can have a coherent discussion about what is going on in the world.

    2. Steve:

      Let's take a partial equilibrium example, for simplicity.

      Assume that potatoes are a Giffen good, so the demand curve slopes the wrong way near the equilibrium price. Start with an elastic supply curve (flatter than the demand curve), and start in equilibrium.

      Now suppose the Potato Marketing Board imposes a quota on potatoes, so the supply curve becomes perfectly inelastic when it hits the quota. And now suppose the PMB reduces the quota below the previous equilibrium quantity, so there is an excess demand at the previous equilibrium price.

      What happens?

      I say the price rises without limit (or until potatoes are no longer a Giffen good so there's a second equilibrium up there somewhere), even though I don't know what determines how quickly price will rise. The (original) equilibrium price has fallen, but it's an unstable equilibrium so we won't go there.

      Do you say the price of potatoes will fall?

    3. (Strictly, to make my above example work simply, I need to assume the quota is both a maximum and a minimum, so producers are required by law to produce exactly the quota amount even if the want to produce more or less.)

    4. Either way, what you've done is to give we a model with no equilibrium, so it doesn't tell me anything about the results of the experiment. You've got the wrong model for the problem at hand.

  2. I don't think Krugman cares about consistency or plausibility in whatever model he is using. He's just interested in promoting a political view to NYT readers and his friends at fashionable cocktail parties using a veneer of economic language. Of course, he must also avoid nuance, lest he leaves room for any possible uncertainty, or the use of any modern economics, so he can assume the role of the wise outsider.

    On the AS / AD model, in my experience a lot of people teach it almost qualitatively, rather than taking it particularly seriously. But even if we look at Krugman's story qualitatively it's really bizarre. It's fine to think that rigidities matter for high-frequency fluctuations, but, qualitatively, the AS / AD story was always a short-run story. The recession began 5 years ago - for how long is it plausible that the problem is nobody can set price and wages correctly? Then again, doesn't Krugman think this is still Japan's problem? Maybe there is some special model involving a liquidity trap that can cause labor markets to be caught in some self-reinforcing spiral of doom from which nobody can escape without policy forcing a "jump" to some new equilibrium. But that model isn't AS / AD or IS-LM or any recognizable New Keynesian model for that matter. Most models that deliver anything even close to that kind of story have a lot of short-cuts; implausible, or a least unjustifiable, assumptions built in; and typically make predictions that we should be strongly skeptical of until given further evidence (e.g. Oil shocks good! Tax increases good! Productivity increases bad! etc).

    1. Unlike New Keynesians who rely exclusively on price rigidities in their models Krugman has never claimed that price rigidity is the cause of the current liquidity trap and just like Keynes (Ch.19 of the General Theory) he frequently points out that more price and wage flexibility could even have negative effects upon output/employment.

      You guys might not like the simple old-school textbook models which Krugman uses and rant about how theoretical inadequate they are (they are indeed) but guess what, they perform pretty well empirically. Krugman has predicted inflation better than this blog author and he has also predicted that this recession will endure unless far more expansionary fiscal policy were used.

    2. Afraid you don't get it. The point is that this model doesn't predict anything.

    3. The notion that these "models" are successful empirically is outright false. And the idea that Krugman is testing them empirically is equally ridiculous. Anon 10:09, just because you want him to be right doesn't mean he is.

  3. "Market efficiency is simply an assumption of rationality. As such it has no implications. If it has no implications, it can't be wrong." -

    Tell me again about nor predicting anything.

    Whether you like it or not, a simple Hicksian analysis of the current crisis has predicted that inflation will remain low. Your model on the other hand has predicted inflation around the corner so old-school macro does a better empirical job.

    Nobody is arguing that IS-LM is the final wisdom. But neither is some DSGE model without finance or whatever model you use. It failed empirically but as you are obviously no scientist (few economists actually are, most of them live in their parallel world) you do not care one iota about real-world connectedness.

  4. nice work HMM SO WELL DESIGNED BY THE WEB OWNER I APPRECIATE U bye the web owner lots of information from owner

  5. Planning to enjoy with desi girls and boys then join Pakistani chat rooms and make desi girls and boys friends.