Wednesday, December 14, 2011

Nick Rowe, Joe Stiglitz, and Jean-Baptiste Say

I had plans to write a big long post about Say's Law, prompted by Nick Rowe's excellent criticism of a Joe Stiglitz Vanity Fair piece that a lot of people are scratching their heads over*. But I'm having a little trouble writing it because honestly I have trouble thinking about Say's Law, gluts, monetary disequilibrium, etc.

It's not that I don't understand the concepts and how they all fit together, I just never quite understood the preoccupation with Say's Law and these sorts of ideas. I certainly never understood why anyone would believe the Law itself (as it's come to us - I should clarify - Say said lots of things at lots of different times), but I've also never quite understood why people would think that supplying excess demand for money itself is necessarily a solution to the problems we really care about. So rather than post determinately, I'm just going to raise a bunch of questions that you all can solve decisively for me in the comment section while I'm taking my math exam.

First I never quite understood why we care whether markets clear or not. There are microeconomic reasons to care associated with efficiency, but from a macroeconomic perspective it seems largely irrelevant. I'll take a full employment labor market that doesn't clear over a clearing labor market significantly below full employment. I'll also take a developed Western democracy with lots of problems with clearing markets over an undeveloped banana republic where everything clears.

Most crucial to my interests is that the phenomenon we follow so closely and call "unemployment" is not inherently about a clearing labor market. The CPS surveyors don't go around asking what peoples' reservation wages are and then only count them as "unemployed" if their reservation wages are at or below the market wage. That's ridiculous. The social fact we claim to care so much about is the people who are to the right of the labor market equilibrium (whether that equilibrium clears the market or not) who show some appreciable interest in working. So again - what is the concern with markets clearing and "excess supply" and "excess demand"? In a market-clearing sense, unemployment doesn't have to be "excess supply", despite our regular conflation of those terms.

Say's Law also bothers me because it seems so static, when the actual activities of human beings depend on more than a contemporaneous balancing of the scales. The usual discussions that go on about monetary disequilibrium act as if investment demand is low because money demand is high - a whack-a-mole theory of recessions where when you push down demand in one area it pops up somewhere else, and money plays this special role of being (1.) fixed - at least base money, and (2.) a transmission mechanism to make a glut "general". It follows, then that if you get money right so there's no more excess demand for money, then there will be no more deficient demand for anything else. But this acts as if the economy is just a big self-regulating, self-organizing calculator - determining a bunch of relative prices that the weird qualities of money can throw out of whack sometimes. But this is a very limited view about what human economic behavior consists of. We're not just price-equalizers. We also anticipate and plan for the future, so any reasonable market equilibrium needs to be intertemporal. Deficient investment demand now is very likely to be a response to excess supply of a good in the future (or equivalently, deficient demand for a good in the future), and have nothing at all to do with any monetary disequilibrium. Currrent demand for investment goods is intimately tied to future supply of consumer goods. Indeed, that's practically the definition of "investment good". In that case, a depression could be general if there is a general expectation of lower future demand than previously anticipated without any excess demand for money at all. Excess demand for money could be the source of a general depression, but it doesn't have to be when you realize that Say's Law or monetary disequilibrium is really an intertemporal equilibrium.

None of this is to denigrate monetary disequilibrium or concerns about an excess demand for money. I am not one to downplay concerns about excess demand for money. But these are some of the reasons why I've never been able to fully jump on board with this idea that monetary policy is by itself a silver bullet. They're underdeveloped concerns, I know, but they're also not concerns that are really spoken to by the usual explanations of monetary disequilibrium that we all know.

As I noted yesterday, I see monetary disequilibrium in the way that I think Keynes did - as a "limiting factor", but not as an "operative factor". This means I am 100% on board with Sumner/Glasner/Rowe and am second to no one in validating the wisdom of going off the gold standard in the 1930s. But I still have a big concern that while that is going to improve things, it might not get us where we want to be.

*I want to say two things about Stiglitz. First - I hope Arnold Kling mounts a spirited defense of Stiglitz, because this is essentially Kling's theory of the Great Depression. I also hope that all the libertarians that cheer Kling also defend Stiglitz and throw some thoughts in. Too often I see people praise PSST and technological unemployment when it comes out of Kling's mouth but mock it when it comes out of someone else's. Second, I do want to draw attention to this comment on Nick's blog defending Stiglitz: "Stiglitz is telling a story that, in principle, cannot be modeled by the kind of highly abstract story. A structural change is not the same as a short period reallocation of incomes due to changes in productivity. This is what Marshall knew that economists today seem to not know. See his "Distribution and Exchange." And that is a lesson that Keynes very clearly learned from Marshall (but Pigou didn't) but that Keynesians seem to have not learned from Keynes."


  1. In other words, doesn't Say's Law in mathematical translation, effectively mean the following?

    MPC + MPI = 1

    Or at least that's according to Brady.

    Then again, according to Brady, there are multiple equilbria found in Book V of the GT, so there may be no need for a disequilibrium analysis.

  2. Well yes, this is essentially what has come to be called Say's Identity says (abstracting from a few things), and that's been recognized to be wrong for almost two hundred years now (indeed, although it's absolutely something Say wrote it's not even clear it's something he ever thought, strictly speaking).

    Say's Law has also been associated with the idea that recessions are relative, rather than general gluts - that there can't be such a thing as a general glut. This is more what I'm talking about here.

    Monetary disequilibrium theory talks about why this is not the case - how a general glut is really just an excess demand for money and money (as a good that is traded in every market in a monetary economy) can propagate general depressions when there is excess demand for it.

    This is all quite right. I'm concerned it's not the end of the discussion, though. I can think of situations where all markets clear and where there isn't necessarily excess demand for anything and it still looks like something we might call a "depression", empirically.

    I can also think of complications that arise from an intertemporal monetary equilibrium and adjusted expectations of future aggregate demand that have little to do with an excess demand for money currently.

    I just have this constant concern that monetary disequlibrium gets all the right answers as far as it goes, but then calls it a day before we've really worked through the entire problem.

  3. In the recent discussion about "overinvestment" I pointed out some of the troublesome definitions there.

    Another problem that closely related to the "overinvestment" controversy and Say's law is the definition of "good" in "general glut of goods". Does it mean "output good" or just "good/asset". That is, does it means something produced this period or does it mean anything that can be traded. I think a lot of people, present company excepted :), aren't sure what they mean here.

  4. I should note, Blue Aurora, that Current's point about assets is very important to combating the Identity version of Say's Law too.

    I only mention Say's Law because I think it's lead people into thinking about macroeconomic equilibrium and macroeconomic outcomes in a way that misses the larger point.

    Or to put it differently, what we now know of as "monetary disequilibrium theory" would look very different and maybe not even be around if the whole Say's Law thing had never come up. I think saying "it's the right answer to the wrong question" is too strong, but something along those lines.

  5. I see. That unfortunately, I have little to say on.

    But speaking of preparing for your mathematical economics exam, I have a question for you - do you think your mathematical economics professor or any of the American University faculty would be interested in reading the Boole/Keynes paper that I showed you?

  6. Say's law is a slightly off-kilter question because it's about what tendencies exist in the market economy. It doesn't concern itself with whether those tendencies can be *accelerated* by government action (or indeed something else).

    I went through a time when I thought it was a useless debate. I don't think so now though because I think the confusions that surround it are useful to solve because they are the same confusions that surround much else.

  7. Dan: "I had plans to write a big long post about Say's Law, prompted by Nick Rowe's excellent criticism of a Joe Stiglitz..."


    1. It's interesting you mention Say's Law right up front. That's exactly what I was thinking about, after writing my post. I was thinking: "Before anyone starts doing any macro, you first need to understand Say's Law, and second you need to understand why Say's Law is wrong. And then you can start doing macro."

    2. Yep. All my discussions of Say's Law are in the context of a simple competitive model where "excess supply" is well-defined. But I don't believe in that model. I believe in monopolistic competition; and that we need to take search costs seriously. And one of these days I ought to reformulate what I say about Say's Law to take that into account. Hmmm. That's hard. Maybe I'm too old, and not up to the job. Time for you kids to take over!

    3. "In that case, a depression could be general if there is a general expectation of lower future demand than previously anticipated without any excess demand for money at all. Excess demand for money could be the source of a general depression, but it doesn't have to be when you realize that Say's Law or monetary disequilibrium is really an intertemporal equilibrium."

    If there's excess supply of newly-produced goods today, what are people planning to do with the proceeds from the sale of those newly-produced goods?

  8. Daniel, FWIW, I still don't understand the full employment / clearing labor market distinction you keep making. I'd like to say I was brainwashed by Mankiw, but he claimed to know what you were talking about (unlike me).

    Let's say there is only one skill level in the whole economy. The labor market clears at $15/hour, meaning the quantity demanded equals quantity supplied.

    I would love to work, but only for $18/hour.

    Are you saying we have less than full employment? If so, if I tell you I want to start for the Chicago Bulls at point guard, did I just make full employment impossible?

  9. I have to agree with Bob.

    The statement that, "I'll take a full employment labor market that doesn't clear over a clearing labor market significantly below full employment" seems ridiculous to me. I would call it contradictory if not for the fact that it seems worthless on its face. It's like posing a hypothetical that doesn't exist.

  10. Look - while the modeling implications are harder, the basic point I'm trying to make isn't that hard. If you care about the 8.6% number the BLS puts out, and the sort of people that comprise that number, you are caring about a number that has nothing at all to do with a reservation wage.

    If you care about the people who have a reservation wage at or below the market wage and cannot get work, then you are caring about something other than the 8.6% unemployment number. We don't know what that number is, because we don't collect it and we probably couldn't reliably collect it, but it's going to be smaller, and for all we know it may not even be cyclical.

    These models weren't designed to bear the burden we're putting on them. They're designed to explain the determinants of the employment level and the wage. There's no obvious way to get "unemployment" out of them. "Unemployment" is as much a subjective personal judgement of your situation as it is anything else.

    So what causes this thing we call "unemployment" and that we care so much about? Mostly, fluctuations in employment (which we are much better at measuring and have much better models for talking about).

    I should say - search theory considerably improved our ability to talk about the phenomenon of "unemployment" more meaningfully.

  11. I should say - some people have said that concentration of unemployment in lower wage categories is hard for AD-AS thinking to understand.

    I don't see why.

    The fact is, there isn't just one class of labor in the world - labor comes with all different skill sets. In a recession, people will lower their reservation wage and a firm that previously employed lower skill workers gets the opportunity to employ somewhat higher skill workers at a lower price. That tendency works its way down the skill ladder and you see unemployment concentrated among lower skill workers. I don't see all that much of a problem.


All anonymous comments will be deleted. Consistent pseudonyms are fine.