Monday, January 9, 2012

A few links questioning neoclassical economics

- Unlearningecon argues that static neoclassical models are useless. I could agree with him if he could tell me a good way of learning dynamic models without static models. To agree with him I would also require a dynamic model that one could teach to freshman non-econ majors in one semester so that they can go into the world with a solid intuition. Without impugning Unlearningecon at all, I'm not sure he's up to the task. He also criticizes comparative advantage, and I think that's more straightforwardly wrong. Is comparative advantage the only thing that matters in trade? Of course not - we know from Paul Krugman's Nobel winning work (and really since Adam Smith and before) that it's not the only thing that matters. But it does matter a lot. And what's even more important - comparative advantage explains transactions between individuals very well, even if a lot of other things are going on in international trade.

- Evan sent me this recently. It's a Guardian article suggesting that economics is not value-free, as it claims, and that: "This theory [mainstream economic theory] of how the economy would work if there were free competition has thus been put to the test. The result is what I believe will prove to be the worst economic disruption in the history of the developed world." First - that last sentence is ridiculous. As far as I'm concerned that was the Great Depression, not this episode, and mainstream economics helped this episode avoid a Great Depression. I also obviously disagree on the value-free point. True, many economists do overlay their own values, but that does not mean the science itself is a normative one or that identification of rational behavior and self-interest valorizes those things (as the article claims). It's simply an explanation of how humans act in the real world.

36 comments:

  1. It's simply an explanation of how humans act in the real world.

    That a value statement!

    ReplyDelete
  2. Ah, Victoria Chick, the eminent British Post Keynesian economist...I was not surprised. I've read similar articles by her in The Guardian before. (Not that I disagree with her in this particular instance, BTW.)

    ReplyDelete
  3. Yesterday I was defending quasi-static models in the context of penalties in Roller Derby. The problem is interestingly similar to what UnlearningEcon says about supply and demand.

    The simple explanations about supply-and-demand are of-course inadequate for more advanced economics. They give way to more complex ideas. That is quite common in education, the first things engineering students learn are unrealistically simple models which are later fleshed out and made practical.

    One thing that has to be understood about those simple models is that they are based on the ceteris paribus assumption. We have a particular demand curve in a particular period of time. At a later period that demand curve changes and all else remains the same. That "all else" includes assumptions about expectations of the future and the state of supply. Of course in practice when demand changes expectations of the future change too. In these type of models that is dealt with by considering too changes rather than just one, demand changes and so do expectations of the future. The acting individuals who make up the market may not think of things this way, but that's how the economist is looking at the problem. The economist can look at things differently, one of the ways of doing that is to discuss "expected" and "unexpected" changes. Another is to draw up an inter-temporal choice model of some sort by having expectations shift the demand and supply curves at various points. Many "mainstream" economists have constructed ways of doing that, including Fisher and Hayek. The way's I've read about so far aren't entirely satisfactory, but I believe there are more sophisticated ones.

    The "feedback" issue is a bit simpler. The question we must ask is: in practice is it likely to be an issue? If I raise my demand for tea now what change will it make to the price of soup ladles in azerbijan? Probably not much. On the other hand, if the central bank raise their demand for bonds significantly during this year what feedback effects will that have? Probably quite a lot. The marginalists were not initially interested in macro variables, what they laid out was principally a microeconomic theory with adjunt ideas about growth in the long-run. Things like the supply and demand model are not meant to be used on aggregate supply and aggregate demand. The more intelligent defenders of AS-AD models don't fall into that trap. These large-scale problems or macroeconomic problems must be treated separately.

    Some of the early neoclassicals were quite aware of the feedback issue. Marshall discusses Giffin goods for example.

    ReplyDelete
  4. It seems strange to me to pose the question, are static models useless, simply because they are not dynamical systems.

    For example, to compare static equilibria, you might need comparative statics (well, duh). Consider a parametrised optimization problem. Does the value function inherit continuity from the function? How does the value function change when the parameter changes?

    All fundamental questions with which economics is concerned.

    How then can static modelling be without use?

    ReplyDelete
  5. Also, on the issue of feedback in static models: perhaps unlearningecon needs to un-unlearn the chain rule...? ;-)

    ReplyDelete
  6. 'I could agree with him if he could tell me a good way of learning dynamic models without static models. To agree with him I would also require a dynamic model that one could teach to freshman non-econ majors in one semester so that they can go into the world with a solid intuition.'

    My problem is and has always been the diagrams rather than the qualitative aspects of neoclassicism. Sure - at its heart, demand supply capture a few truths: if something is less scarce, the price will go down; if it is wanted more, the price will go up, generally speaking.

    Dynamic models: I don't know. I don't have the capacity to come up with my own just yet, but Steve Keen has come up with a good macro one that's fairly easy to understand, and he actually remarked that it was easier without assumptions.

    I don't think teaching flawed models on the grounds that they are easy to teach (actually, a lot of people struggle when they first learn economics) is a valid defence.

    On CA: The entire point of my argument is to provide a theoretical justification for the 'infant industry' argument, seeing as basically every developed country used protectionism when they were starting up. I'm trying to show that CA is irrelevant for the purposes of that argument, at least in the short run.

    I admit I'm not that happy with that post as it takes on too much in one go. However, I have critiqued ceteris paribus and CA more clearly in a couple of older posts, so I just felt like that one was a round up of the general principles.

    Thanks for the link.

    ReplyDelete
  7. Re: economics being value free. Naturally, I don't think it is, but that article is a bit off base. However, I'd strongly recommend a book called 'The Skeptical Economist'. I know you're reluctant to read heterodox books because you have conflicts with your work, but it's more about ethics than economics and I expect you'll be largely in agreement with it. It's also a really good read.

    ReplyDelete
  8. "I don't think teaching flawed models on the grounds that they are easy to teach (actually, a lot of people struggle when they first learn economics) is a valid defence."

    I think it is because they're the starting points for realistic models. It may well be that economics students can only understand the supply-and-demand model in the first semester of their course. It forms a basic starting point from which other models can be understood. Then later things like expectations, the theory of monopolistic competition and economising over time come in.

    Your post belongs to a commonly encountered set of criticisms that group "Neoclassical Economics" with "Economics 101". To really criticise neoclassical economics you'll have to understand it in detail, not just superficially.

    ReplyDelete
  9. I'll leave it as an exercise to the reader to find the quasi-static assumptions in the arguments put forward against free trade.

    A nice starting point is to compare arguments against free trade that rely on capital export to those that rely on the infant industry argument.

    ReplyDelete
  10. 'Your post belongs to a commonly encountered set of criticisms that group "Neoclassical Economics" with "Economics 101". To really criticise neoclassical economics you'll have to understand it in detail, not just superficially.'

    I'm well aware of this rebuttal and have put it into my FAQ. The point is that if the foundations of the theory are flawed, it doesn't matter how many 'frictions' you've incorporated later on.

    Steve Keen sees a lot of the later models as attempts to shield the neoclassical core.

    ReplyDelete
  11. Also, a lot of the 'later' models such as Arrow-Debreu and Walrasian equilibrium are as hopelessly flawed as the foundations.

    ReplyDelete
  12. Here's a choice quote from that FAQ:

    How do neoclassical models contradict themselves internally?

    The movement of or along one line creates ripple effects throughout the economy that violate Ceteris Paribus. This leads to a constant feedback loop, and so the static nature of the models renders them useless.


    Unlearningecon appears to believe that, since he has never encountered dynamic neoclassical models, they therefore do not exist.

    In other words, "Unlearning Economics" is unlearning economics before first learning economics--as Current suggests.

    ReplyDelete
  13. "The point is that if the foundations of the theory are flawed, it doesn't matter how many 'frictions' you've incorporated later on."

    In your FAQ you refer to Steve Keen and Noah Smith. But you don't actually give specific criticisms.

    Let's say we have a monopolistic competition model that takes place over time with expectations that can be of varying degrees of accuracy. It's not enough to say that more complex neoclassical ideas only add frictions. Now of course monopolistic competition is a 'friction' but expectations are not.

    Exactly what is wrong with this model? Specifically?

    You could say "feedback" then I'd say your misapplying a microeconomic model to a macroeconomic situation.

    ReplyDelete
  14. The point of the blog is to provide a criticism of neoclassical theory as a whole rather than focus on critiquing specific areas. Why? In my view the entire discipline is flawed from the ground up and attacking individual models will only result in new models with similar flaws.

    Having said that, I *will* do some more in depth criticisms in the future, it's just that I'm still establishing the foundations of my position.

    Re: monopolistic competition, the start of the problem would be using MC=MR as profit maximising, since firms simply don't do that in the real world.

    I don't think 'feedback' only applies to macro. The point is that as the firm alters price, production inputs and so forth it changes the composition of the economy, which creates ripple effects that come back and alter demand/costs.

    A large part of these fluctuations may be internalised by the firm, but that doesn't mean they don't exist.

    ReplyDelete
  15. But that's why we have general equilibrium models and mathematical tools like the chain rule and envelope theorems.

    Do you know what the chain rule is? What mathematical economics textbook do you use for your undergrad studies?

    ReplyDelete
  16. I've been casually reading these comments, and I agree that a lot of the criticism of mainstream economics is entirely unfounded, but (with no particular commenter in mind), let's make sure we keep this polite w.r.t. unlearningecon.

    One of the advantages of heterodoxy is that it challenges the mainstream (this goes for Austrians too - who often are treated the same way). The mainstream challenges itself all the time, but heterodox challenges come from a different angle. That's a good thing.

    I'm fine with all of these modeling techniques - I don't consider them fundamentally flawed. I do consider them fundamentally imperfect or incomplete. Those are somewhat different accusations. For those of you critical of unlearningecon, what do you think of Israel Kirzner's critiques? I'm not a huge fan of Kirzner myself (for much the same reasons that I disagree with unlearningecon here), but surely some of you see merit in them.

    Anyway - same goes for the heterodox side - civility matters.

    Carry on!

    ReplyDelete
  17. Daniel,

    Sure thing--sorry, didn't mean to appear rude; my question about the chain rule and textbooks was meant sincerely.

    A static model will (obviously) not capture effects that occur over time (by definition of static). For that you would (again, obviously) need a dynamic model (by definition of dynamic), of which there are plenty.

    But a static model can certainly capture feedback effects. If Unlearningecon can point me to his/her math econ textbook I will try to find a an excerpt that covers some of the relevant material.

    ReplyDelete
  18. No - I was serious when I said I didn't have anyone specific in mind and that I was tracing the whole comment thread.

    I just don't want it to turn into a "you're just dumb", Cafe Hayekesque thread.

    The questions you raise about simple issues of chain rules and envelope theorems form the very core of a good critique of unlearningecon's position. Of course this stuff is known and handled by the mainstream.

    Don't let my friendly reminder intimidate you! Some comment threads just clearly have a higher probability going off the rails than others :)

    ReplyDelete
  19. Understood, and thanks.

    ReplyDelete
  20. Vimothy: yes I am aware of the chain rule, you learn that in 6th form..

    I don't really use textbooks (as a side note, why are Americans so obsessed with textbooks? Every time I say I study economic they ask which textbook I use. Weird.)

    General equilibrium models make a lot of assumptions about heterogeneity and break down if you try to incorporate too many 'real' factors.

    They are also ergodic and unable to model the recent crisis, as opposed to something like Minsky, which destablises itself endogenously, and thus appears to reflect the real world more accurately.

    ReplyDelete
  21. Unlearningecon,

    I'm not American.

    If you are already familiar within the chain rule then you should understand how a static model can capture feedback effects.

    Of course, it's not like this is unique to economics per se. A static model in any context (pick your favourite discipline) can capture feedback effects via a chain of nested functions (for an econ e.g. take the value function as a function of the objective function, as with the envelope theorem).

    ReplyDelete
  22. My mistake, I just find the textbook thing strange as I have never encountered it in the UK.

    The feedback effects can be 'captured', but since it's a static snapshot, when we observe at any one time it doesn't necessarily tell us anything useful, much like a static snapshot of weather patterns wouldn't. We need complex/agency models to get a clearer picture of what's going on.

    Daniel, your commenters are generally very good and so if you think I'm being rude try and invoke the principle of charity, i.e. read it in a jokey tone. I get fired up when engaging some people but there's always good discussion here, so I do my best to remain civil.

    ReplyDelete
  23. Unlearningecon,

    No probs re me not being American.

    But are you really complaining that although a static model captures feedback effects, it does not capture effects over time?

    That's like me complaining that my calculator is useless because it doesn't allow me to surf the web. The calculator is not designed to surf the web; that's what my laptop is for. For general arithmetic my calculator works just fine.

    In order to capture effects that occur over time, you need a dynamic model *by definition*.

    ReplyDelete
  24. The whole point is that the static models might be 'correct', but they are not useful as we don't know which direction things are moving in.

    However, I wonder if we're getting off base here.

    My initial complaint was actually that D-S and IS/LM *don't* capture the feedback effects created when one curve moves, due to altered expectations, the impact of changes in prices on the rest of the economy and therefor eon the firm in question, etc.

    Maybe I'll clarify this objection in another post.

    ReplyDelete
  25. But these are just learning aids for introductory econ lectures. They don't represent actual current research.

    If you ever go on to postgraduate study, you will have so many dynamic models rammed down your wazzoo that you'll be solving them in your sleep. Trus' me bruv.

    ReplyDelete
  26. They may be dynamic but they are still ergodic, though that is a separate issue. They are also unable to model the recent crisis, which makes alarm bells go off in my head.

    ReplyDelete
  27. What do you mean by "ergodic"? Are you talking about measure theory? (Really?)

    We can certainly write stochastic dynamic models.

    Also, when you say "they are...", it makes it sound like you have a specific dynamic model in mind. Do you? Earlier in the thread, you seemed to suggest that no neoclassical dynamic model exists, which contradicts this.

    ReplyDelete
  28. I using 'dynamic' wrong - what I actually meant was 'chaotic'.

    The problem is that the vast majority of these models, perhaps all of them, tend to oscillate around or converge to some sort of long run equilibrium, e.g.

    DSGE
    Solow-Swan
    Arrow-Debreu

    The real economy doesn't seem to do this.

    ReplyDelete
  29. No, they're models. The stability of a system depends on its parameters. A dynamical system can converge onto a steady state or diverge away from it. (And there are various ways that we can classify that trajectory e.g. on linear planar systems and using the linearisation theorem to qualitatively classify nonlinear systems).

    It all depends.

    BTW, if you think the real economy is never in dynamic equilibrium, how would you describe, e.g., the long-run behaviour of GDP? Chaotic?

    ReplyDelete
  30. I'm not sure what point you're trying to make in your first paragraph. None of the neoclassical models can model the crisis as endogenous and the economy as inherently self-destabilising, so for me they aren't sufficient.

    'e.g., the long-run behaviour of GDP? Chaotic?'

    There certainly seems to be a long run trend in particular countries, but on a global scale there are massive differences between areas, and even in particular countries there is a lot of volatility. I'm also not sure it is sustainable in the long term, so to describe it as 'ergodic' might be a bit of a stretch.

    ReplyDelete
  31. Any dynamic model can be non-stable (or stable, of course) depending on its parameters.

    Given those parameters, either it converges to the steady state or it does not. We can easily classify that trajectory of convergence or divergence qualitatively in the case of linear planar systems, and even in non-linear systems by taking advantage of hyperbolic steady states with the linearisation theorem.

    You can get all sorts of results with these models.

    Here's an example of that, using a simple dynamic Keynesian multiplier-accelerator model, where s in (0, 1) is the marginal propensity to save, Y(t) and I(t) are output and investment at time t and G-bar is a (constant) govt expenditure:

    (1-s) Y-dot = -sY + I + G-bar
    I-dot = aY-dot – I

    The system is simple, with a unique steady-state at I = 0, Y = G-bar/s. It’s easy to show that this cannot be a saddle-path—instead it is a source if a > 1, a sink if a < 1. Under some parameter values, that sink is an attracting spiral. Under others, that sink is a repelling spiral, and the economy goes through a series of ever more extreme booms and busts until corrective action is taken or output goes to zero.

    ReplyDelete
  32. The reason I prefer Minsky (and by extension, Keen) is that their models are INHERENTLY destabilising, rather than depending on whether certain parameters are 'off' or 'just right'. I can't believe that the only reason capitalism continues to destabilise itself is because certain values are off - I think it's a more fundamental tendency towards self destruction.

    Maybe I'm biased, I don't know. I'll give you the last word!

    ReplyDelete
  33. I'm betting that since it's January the three people who aren't Daniel in this discussion are Brits.

    Daniel brings up Austrian economics. Although I learned mainstream economics first I've since read much more Austrian economics. My thinking is Austrian-leaning if not entirely Austrian school.

    He's right to point out the Austrian's criticisms of neoclassical economics. I agree with many of those, though I disagree with some and I think for other the mainstream have cleaned up their act.

    But, that Austrian criticisms are often different to the post-Keynesian ones. In the Austrian criticisms a theme repeated again and again is that underneath the assumptions used to build a particular model there are more general idea. While the assumptions may be fragile and dubious the general idea beneath is often much more robust.

    Take the monopolistic competition model we were discussing earlier. Is it really necessary to assume anything very complicated about business profit maximizing? Firstly, lets take a situation where an unexpected increase in demand occurs, cet paribus. Let's assume that this increase is not thought to be permanent. In that case businesses will increase prices if they can within the period of time when demand stays high. Why would they do otherwise? If there is one particular brand of good that increases in demand then the business that supplies it will increase the price, again if they can within the period of time when demand stays high. There are perhaps exceptions there, if they believe customer loyalty will be higher if they don't for example. Now it's true that by doing this those businesses may not increase their profits in line with any particular model about that. But, who cares? it isn't directly relevant.

    (Hayek's criticism of monopolistic competition in "The Meaning of Competition" contains similar elements. But I see that more as an attempt to advance beyond simple ideas than as an attempt to demolish them).

    Of course if expectations of the future are changing then things are different. But even in that case the *direction* that a change in demand pushes the price in is the same. The change in expectations may be the overwhelmingly larger factor and decide what happens to the price, but that's a different matter.

    (I'll leave relating this back to the chain rule and perturbation models as an exercise for the reader.)

    ReplyDelete
  34. As for feedback. What I would like to see is a situation where realistically a microeconomic model isn't right because of feedback. I can think of a couple of those myself, but they're not very realistic. We all know everything is connected back to everything else, as some ancient greek said we can never step in the same river. But that doesn't mean that practically changes in one market will have a significant feedback effect on the opposite side of that market. If one draws markets to be extremely broad then that's a different matter. But, in marginalist economics the markets picked out by the economist shouldn't be broad, getting away from unrealistic ideas about broad markets was one of the things marginalism was created to do.

    The last few comment here have gone into macroeconomics. On this blog in the past there has been a lot of criticism of "microeconomic hegemony". Pete Boettke says something like "macroeconomics must be founded in robust microeconomics" and Daniel replies by saying that "microeconomics must be founded in robust macroeconomics". What I've learned from these debates is that the two sides aren't really talking about the same thing. Critics of the Austrians, for example, think that their ideas are microeconomics scaled up to an level where it no longer works. The Austrian's themselves though know that though their theories may have microeconomics at the bottom the process of reasoning that leads to macro introduces much more. For their part the Austrian's criticise macroeconomic ideas that don't have much basis in micro. The problem here is that many who say they practice "positive economics" really don't. They really are interested in the micro underpinnings, they just assume different things about them. Keynesians assume various things about excess capacity and existing stocks of intermediate goods. Monetarists assume the same sort of things, or something about substitution. (Austrians assume things about substitution too.)

    What everyone often seems to forget is the complexity of building a complete argument. I often see it taken for granted that if school X argues A, B & C then those things must be consistent with each other, and that the only question can be if their assumptions are consistent with the facts. That's all wrong, the process of working out a theory is fraught with peril, so an economist (or school) can easily believe logically contradictory things through error.

    Unlearningecon seems to believe that if Minsky's view of economic crises is right then all of neoclassical economics must be discarded. Because it somehow can't lead to predictions that are like the recent behaviour of real economies. That assumes that neoclassical economics is some sort of incredibly intelligent and diligent Dr.Pangloss. It assumes that every alleyway that a micro theory leads to has been chased down and it's been shown that it must lead to some optimistic view of Capitalist economies. Of course that hasn't happened. Mainstream ideas can be taken in all sorts of directions, as indeed can the ideas of most schools.

    ReplyDelete
  35. 'Is it really necessary to assume anything very complicated about business profit maximizing?'

    Well we need to know what firms do to maximise profits, surely.

    'Firstly, lets take a situation where an unexpected increase in demand occurs, cet paribus.'

    Ignoring the CP objection, your example doesn't address my problem. I was suggesting that, dependent on expectations, an increase in price might lead to an increase in demand, rather than a decrease, if there were scares of a shortage.

    'Unlearningecon seems to believe that if Minsky's view of economic crises is right then all of neoclassical economics must be discarded. Because it somehow can't lead to predictions that are like the recent behaviour of real economies.'

    Not all of neoclassical economics, but a good deal of their macro models, yes. That's how theory usually progresses - the one that corroborates with reality best is advanced whilst tohers are discarded.

    People have this weird cognitive dissonance with economics, a sort of 'let's not throw the baby out of bathwater' mentality despite empirical contradictions. The funny thing is, the same people don't apply it to other sciences/social sciences.

    'As for feedback. What I would like to see is a situation where realistically a microeconomic model isn't right because of feedback.'

    Supply/demand in a stock market isn't right because of feedback.

    IS/LM cannot be right because of feedback through expectations, which both curves are partially derived from.

    I think you need to ask yourself if we assumed ceteris paribus in another discipline, such as engineering, and violated it 'only a little bit', would it be safe to build a building ignoring that?

    I mean, in planetary motion a slight change in starting conditions completely changes the end result. In an even more complex system like the economy I refuse to believe we can ignore minor contradictions on the grounds that they are minor.

    ReplyDelete
  36. You are free to take the view that much neoclassical economics must be discarded. But, what are we to replace it with? Can you eliminate these problems? If you can't then many people will stick to mainstream economics. For this reason specific issues are interesting.

    You write "Well we need to know what firms do to maximise profits, surely."

    For the purpose of discussing competition it isn't essential to know exactly how firms profit maximise. All that's needed is realistic ideas about the direction prices will move in response to short-run changes in conditions. Profit maximization is interesting and an important topic by itself, but that's another issue.

    "I was suggesting that, dependent on expectations, an increase in price might lead to an increase in demand, rather than a decrease, if there were scares of a shortage."

    It may, but "scares of a shortage" are a change in expectations aren't they? In a sense I agree with you here. A change in price may lead to a change in expectations of the future. There are many other interesting nuances here that Kirzner pointed out. For example, a temporary rise in price may cause agents to seek out substitutes once the price falls they may stay with those substitutes. I don't see how this is a stake in heart of neoclassical economics, especially since more complex models allow for things like this.

    I don't understand what IS/LM has to do with anything, since it's not even a neoclassical model. I agree with you that the IS & LM curves aren't independent.

    In engineering, including structural engineering, principles that are violated "only a little bit" are routinely used. Structural engineers often only model first and second order effect. They design things so that the problems in their models don't come up in practice. Economics is more difficult because it's a science, the market is not something that's been engineered in any precise way. Economists have no control over the issues that it brings up.

    If you can come up with some ideas without these problems then people will be all ears.

    ReplyDelete

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