Critique of Classical Economics

A blog called “The Wild Peak” has an interesting critique of classical economics, making the case among other things that its failures are rooted in physics envy.

Given the absolute centrality of space and time for all things economic you might be forgiven for thinking that they would be important components in modern economic ‘science’. They are not. In neoclassical economics, which retains a vice-like grip on both academia and policy making and provides the intellectual support for the ideology of neoliberalism, space and time are almost totally absent. Why is this? Why does it matter for people and the planet? And are there more useful alternatives?

It also tend to ignore most important aspects of scale, energy use, resource limits, how aggregate markets are not scaled up individual demand and supply curves, and how economic actors actually interact, adapt, behave and choose. There is also no concept of time’s arrow, i.e. the irreversibility of processes and how such “path dependence” is crucial for economic development. The point is that even though all these factors have been studied by some excellent economists (usually of the non neoclassical variety), they are still marginalised within academia, business and government policy making. Stripped down classical mechanics still rules the roost.

I start to part ways with the article at the end; the author is a bit credulous regarding Georgescu-Roman’s overdrawn attachment to entropy. (As long as the sun shines, the world is far from its entropy bounds in a physical sense of entropy. Entropy in the mathematical sense should not be confused with thermodynamic entropy because it has no real second law constraints in that context.) I am not sure a satisfactory economic theory exists. But I think the argument for the failure of the dominant view (still called “liberal” in the UK, better understood in American contexts as “market-libertarian” of “Friedman school”) and its variants (Keynes-Krugman school) is solid.

h/t @gl33p

Comments:

  1. Dear Michael,

    Thanks for your comments on my short article and also the English to American translation. Winston Churchill once said Britain and America were two countries divided by a common language.

    When I have a little more time (at the weekend) I'll say a little more about Nicholas Georgescu-Roegen. Paul Samuelson once commented that when any potential PhD student mentioned the word entropy he ran a mile. But remember he also said he wasn't much bothered if his economics was right so long as he wrote the text books - which he did.

    Very pleased to discover your interesting site.

    Stephen Lewis

  2. I think its close to "not even wrong" territory. Or "Intellectual Impostures" (http://philosophynow.org/issues/25/Intellectual_Impostures_by_Sokal_and_Bricmont). Or most closely http://scienceblogs.com/stoat/2012/05/11/on-getting-out-more/.

    The entire thing seems so misconceived that I don't know where to start.

    Economics, on the large-scale, is trying to abstract out behaviour, so yes its quite natural to remove space and time. Its comparable to, say, using conservation of momentum to say "we don't care in detail about the results of this collision but we know the sum momentum of the products".

    And yet on the smaller scale, of course both space and time come in, perfectly naturally, as costs.

    As to whether this matters for having economics address environmental issue: well, lets consider the problem of internalising externalities: for example, the emissions of CO2. Economics can help you study the different effects of a carbon tax versus regulation versus cap'n trade. In none of which do space, or time, appear prominently or usefully. but that's OK, because they aren't really needed.

  3. William, I'm a bit surprised by this critique given that you once stipulated that a lack of dynamics might well be a key failing of economics. This is not much different than my claim to that effect.

    It is certainly the case that science proceeds by abstracting away things that don't count. But it does not proceed by abstracting away things that do count.

    In fluid dynamics, when we neglect a term, e.g., time derivatives, we are obligated to demonstrate that those terms are small compared to the dominant balance.

    It certainly appears to be the case that mainstream economics is built around a "general equilibrium" model, one which is conveniently mathematically intractable at scale, leaving the discipline stuck at the level of toy problems.

    I also agree with Stephen that there is no reason to believe that a general equilibrium actually exists; the system is sufficiently self-referential that it quite likely never settles down.

    Where I am not convinced is in the claim that anyone has done much better than the mainstream. I'd like to see it if it exists.

    I tentatively believe that agent modeling is the only approach that shows promise, but that it will be very difficult to get it even qualitatively right; i.e., a climate-sense prediction may be possible.

    I do hope some predictive power is possible, because we have some very complex decisions impending. These are decisions that, for the reasons Stephen identifies, conventional economics (with its resource-substitution presumption and its lack of a dynamical formulation) will be of little assistance.

  4. > I’m a bit surprised by this critique given that you once stipulated that a lack of dynamics might well be a key failing of economics

    I didn't read the original post to be saying that. Had it done so, I wouldn't complain. Yes, I agree, that there is potentially / probably a big problem in the way that economics deals with "dynamics"; but phrasing that as "not dealing with space and time" is very odd. Perhaps this is just a language problem.

    > I am not convinced is in the claim that anyone has done much better than the mainstream. I’d like to see it if it exists.

    Yes, seconded.

  5. Re both the above comments.

    First I was taken aback by William's rather dismissive tone - so I wasn't going to reply. But as his reply Michael was more reasonable I'll add a couple of very brief comments. First re General Equilibrium models. I see you both agree. I was taught this by one of it's leading exponents, Avinash Dixit, now at Princeton, so I hope I know what I'm talking about.

    Also I think agent modelling of the economy (as a complex adaptive system)is a very useful approach indeed - see for example Alan Kirman's work. I also teach this subject at the Bank of England.

    Re the Space and Time bit, I wanted to explore this tentatively because I do think the point is valid, though needs more work.
    Economies do happen in time and space and many of the features of the way they work can't can't be explained by neoclassical economics even if you put in "transport" costs as a proxy.

    I also do think there are alternatives to neoclassical economics (but that's for another time)yet neoclassical economics still has it's uses for looking at some of the issues William mentioned.

    In general I agree with Michael's comment:

    "It is certainly the case that science proceeds by abstracting away things that don’t count. But it does not proceed by abstracting away things that do count."

    By the way I'm half way through writing a piece on prediction in economics, which I hope won't receive another blast from William.

    Best wishes

    Stephen Lewis

  6. MT, I think you probably know more about economics than me precisely because you have a solid grounding in physical mathematical modelling. I crammed a small quantity VERY late into my brain and in a half-arsed manner, coming from a social science background. And it was fine to do a course on mathematical modelling precisely because, yes, C19th economics built on physics since those were the available tools. Agent modellers use available tools now, but I think much of it is all fur coat and no knickers (and I say that as someone still trying to get some value from the method).

    So while I never really feel confident being critical, and do think many things are wrong about economics, I'm still wondering which things are wrongest. As always, being a krugman blog fanboy, here's a recent one on 'gadgets vs scratchpads'. He's completely right about supply and demand to start with. (Krugman's also the most prominent recent economist to have done the most to deal with the dimensionlessness of economics.)

    On the article: AFAIK, there isn't actually any 'auctioneer' in GE models, any more than there's a maximisie-pixie in a Lagrangian. It's something that's often raised up as an example of how obviously farcical the models are - agent modellers often start with this point. It was just a thought experiment that Walras added in a later edition, I think. You won't find it anywhere in the model spec. That's one of the, er, 'fun' things about economic models: they're this rather unmanageable tangle of maths and verbal supporting argument. Having spent waaay too long trying to untangle Krugman's core-periphery work, it's not obvious what is what. It is probably obvious to people apprenticed in it. The reason goes back to what Krugman's talking about in that blog entry, and something I've blathered about before: the models are best used for mulling over problems and helping develop researcher insight. Hence Krugman saying, "use gadgets wisely and in moderation, and never let them become too important to your thinking".

    I wonder if there's a comparison to GCMs. Had a chat with someone about this recently which chimed with some things I'd been wondering. Policy makers have demanded higher-res models with, it appears, the full expectation of higher-resolution prediction - will this tiny cell be in drought in 20 years' time? But is that the point of the models? Or is it more to do with the much wider range of reasons for model-building that policymakers don't really understand or appreciate, ie. developing a deepening understanding of the system you're addressing? If better prediction were the goal - as the person I was talking to said - surely a better goal is more manageable GCMs able to run much larger ensemble forecasts?

    On the question of what approaches are marginalised in places of political power or academia - there's plenty going on in academia. It's certainly right that a specific school of economics has its foot in the door - again, a point that Krugman has been (pdf) quite specific about when he's explaining his approach to introducing geography into economics: "mainstream economics isn’t going away: like it or not, the White House has a Council of Economic Advisers, not a Council of Geographical Advisers, the World Bank hires lots of economists and not many geographers."

    But on the kinds of models that are actually used for making decisions, and how they're used - there's a book to be written there. I think there's a whole ecosystem of different approaches depending on the policy area, and neoclassical economics doesn't have quite the grip that it appears to - certainly not globally. (Though the flipside of that is often a kind of colonial megalomania where newly developed model approaches get tested in developing countries.)

    I think I'm particularly resistant to being so dismissive partly because of some of the alternatives that seem to be on offer. But the point, really, is how the hell you build a polity capable of dealing with such a plethora of ways of modelling the world. Democratically, presumably, but with what criteria? It's sort of easy with, say, heart surgeons: it's pretty quickly apparent if their mental model of the human body isn't up to much. Economic policy is considerably more challenging. Stafford Beer quotes Macmillan: "running the economy is like trying to catch a train using last year's timetable."

  7. Related thoughts: it might make more sense to think about this problem backwards and start with the obvious political point. A rich king advertises for a new economic advisor. Two apply. One says it's important to tax the rich and distribute the money widely. The other says Kings should tax salt, crops, imports, exports and walking. Both have models to prove it. Who gets hired?

    A somewhat daft way of saying that the 'mysterious pre-established harmony' between neoclassical economics and a particular version of liberal democracy (paraphrase of Strauss) isn't very mysterious. I'm just not sure what you do about that, if one thinks modelling has any value at all to social questions (which presumably is the case, given that physical modelling of climate has now become so tangled with social questions.)

  8. Stephen, fascinating to hear you were taught by Dixit. Having spent such a long time wrapping my noggin round the Dixit-Stiglitz model, I'm jealous of having actual access to him!

    We've got into discussions on P3 before, here and here, though not really coming to any conclusions. But you'll see we've picked up on a fair few things you've mentioned - one of my bugbears being the misuse of Friedman's 'F-twist' argument, claiming he says "unrealistic assumptions don't matter" (something I also repeated until, looking for ways to think about model-building, I actually read what Friedman said, which is subtly but vitally different to the caricature so often used to paint economists as naive Vulcan-like imbeciles.) A lot of that's come from my (still not quite finished!) thesis work; I've stuck the current version of the navel-gazing model chapter here. There's a lot of overlap, even many of the same quotes, with things your blog's discussing.

    The most recent article of yours shares some of the same view of modelling in general. I'm not sure whether this eco-economics attack on modelling stems from quite the same source as MT's skepticism, would be interested to dig more into that. I think this starts getting close to the heart of this, which again comes down to what we're claiming models are *for* (which I waffle on about interterminably in that thesis chapter.) You quote the Daly/Solow argument (Daly: "If we want a bigger cake, the cook simply stirs faster in a bigger bowl and cooks the empty bowl in a bigger oven that somehow heats itself.") But you don't provide Solow's response; quoting meself quoting Solow:

    "Solow's reply to this highlights a recurring argument used to defend the abstract nature of many economic models - critics are taking them too literally, and not considering how the models are used: 'We were trying to think about an interesting and important question: how much of a drag on future growth, or even on the sustainability of current production, might be exercised by the limited availability of natural resources and the inputs they provide? ... The role of theory is to explore what logic and simple assumptions can tell us about what data to look for and how to interpret them in connection with the question asked' (Solow 1997 p.267/8). Solow goes on to point out that the argument should be about how substitutable renewable and non-renewable resources are, given that the former are likely to be highly capital-intensive. (Ibid.) He appears to be saying that his critics have mistaken economists' models for their actual understanding of the world, rather than tools that aid that understanding."

    Which is what I was arguing Krugman also says in the comment above, and what Friedman was saying way back in the 50s (though whether Friedman follows his own advice, I'm not qualified to say - I only really know his stuff from his 'essays on positive economics'). E.g.: when assuming s = 1/2gt^2, "under a wide range of circumstances, bodies that fall in the atmosphere behave *as if* they were falling in a vacuum. In the language so common in economics this would be rapidly translated into: the formula assumes a vacuum. Yet it clearly does no such thing."

    And of course there are plenty of circumstances where it would be a completely inappropriate way to think about falling objects. But that doesn't make using it 'naive' or invalid, any more than using temperature and pressure measurements does (when we know that at atomic reality is more complex).

    I haven't got to the bottom of this stuff to my own satisfaction at all. It would be great if we could carry on exploring them here at P3, but I wonder if that might require some kind of agenda!? I wouldn't mind starting with actually working up a common understanding of some economic models as they are. Stephen, it sounds like you have a solid economic background. I don't really; I've gone off and done this by myself while my supervisors looked on horrified, and found it immensely tricky to find economists to check my understanding with (perhaps my own networking failings, I shan't blame economists' closed-shop attitude!)

    Some suggested things to discuss: understanding a basic general equilibrium model, how they've been used, how they're related (or not) to the actual policy workhorse DSGE models, what other models inform policy; whether economic training in these models is (as Krugman's own view seems to suggest) more a set of shared heuristics almost incidental to how economists are apprenticed in a policy episteme - so it's not 'about the truth of the models' so much about training and embedding policymakers. Which would mean we could attack the models til the cows come home and we'd be missing the point. You can actually see how that might function in a completely different context: Lansing's work on Balinese rice management, where a shared collective model allows autonomous ritual actions to both re-create the landscape and manage water to maximise crops/minimise pests across the Subaks. There's no 'correct' top-down model, there's a shared social technology tied into a specific landscape, kept alive in people's minds through ritual. Lansing calls it 'sociogenesis': "when Balinese society sees itself reflected in a humanised nature, a natural world transformed by the efforts of previous generations, it sees a pattern of interlocking cycles that mimic these cycles of nature" (priests and programmers p.133). Mainstream economics may partly work the same (esp. ritual!) but on quite a different scale (and of course with the open question of whether it's capable of ending in a cyclical self-maintenance or is instead a global virus; cf. Agent Smith).

    It'd be good to explore a recent-history example of an economic model's impact: Krugman's core model again. Since 1991 it's gone from 'thought experiment to lure economists into thinking about geographical questions' (while also radically changing the profession's view of the central dynamics of international trade) to a World Bank report on geography coming straight from it, despite Krugman's own apparent caution on using it - and a continued suspicious lack of empirical support that shouldn't be a surprise given it was only ever meant as a toy model to make a point. (I've never quite understood what Krugman's own view of this is; initially caution and a notable silence as the Nobel prize arrived and some of the key ideas got absorbed by the body politic. But I'm sure he'd have some clear Views on it.)

    And to ask the same question again: given all this, what role do we actually think models can / should play in both analysing and organising society? I have a book in front of me by Stan Openshaw, an ex-professor from my department, called "Using models in planning" (1978). A quote:

    "Without any formal guidance many planners who use models have developed a view of modelling which is the most convenient to their purpose. When judged against academic standards, the results are often misleading, sometimes fraudulent, and occasionally criminal. However, many academic models and perspectives of modelling when assessed against planning realities are often irrelevant. Many of these problems result from widespread, fundamental misunderstandings as to how models are used and should be used in planning."

    While he's writing about town planners, the same applies. We don't really understand how this is meant to work. It happens anyway, but without asking about this, we're stuck in this strange place where one side carry on using their models while others keep on going "ha! look at those stupid assumptions!" - as if by some miracle of alchemy, the models being criticsed will crumble like vampires at dawn. Kuhn's stuff makes clear it doesn't even quite work like that in the physical sciences: the relationship between discovering problems, errors, better models and the structure of a discipline is way more complex. If that discipline is then tangled into political power, there's a whole other bunch of stuff going on...

  9. On dematerialization, physicist Tom Murphy:

    http://physics.ucsd.edu/do-the-math/2012/04/economist-meets-physicist/

    and my own thoughts:

    http://init.planet3.org/2009/05/cruel-hoax-growth-equity-cannot-be-sustained.html

    http://init.planet3.org/2009/04/tautology-its-weaknesses.html

  10. Dear Dan,

    I just read all this. Some very interesting points I'll try to respond to in the near future.I just wrote two popular pieces on Economic Prediction and Dematerialization which are obliquely relevant:
    http://thewildpeak.wordpress.com/2012/08/05/thoughts-on-the-limits-of-economic-prediction/, http://thewildpeak.wordpress.com/2012/08/02/humpty-dumpty-economics-the-myth-of-dematerialization/

    Stephen


Leave a Reply

Your email address will not be published.