On the Absence of a Model

Does the absence of a workable model refute a hypothesis? I think this makes some sense – after all what is a hypothesis other than a model?

Here’s an extremely interesting article by a skeptic of economics, Jonathan Schlefer, claiming that there is no “invisible hand”, or  equivalently, that the general equilibrium model at the core of economic thinking is simply invalid. He also shows that the widespread belief to the contrary is consequential.

This is doubly interesting to us because it reflects in the opposite way on climate science skepticism. Nobody has ever proposed a model of climate which is not sensitive to greenhouse gases that actually works as well as the models in use in mainstream science. Schlefer asks:

The failure to model the invisible hand is ironically powerful. Any given economic model might well be implausible. But if the brightest economic minds failed for a century to show how some invisible hand could move markets toward equilibrium, can any such mechanism exist?

Similarly, if climate models are really off base in suggesting that the system is sensitive to greenhouse gases, you would think that the fossil fuel interests, with their armies of PhDs, would have by now constructed a model that works as well which is not sensitive in that way. It is the fact that nobody has done so that is as indicative of the success of climate science as the status of economic models is indicative of the failure of economic science.


  1. "Does the absence of a workable model refute a hypothesis?" As usual, Krugman's my favourite on this. Particularly from section 'the evolution of ignorance' where he compares theory-making to early map-making. Early maps were more report-based but, as they slowly became filled in, that heuristic knowledge was lost.

    "There was an extended period in which improved technique actually led to some loss in knowledge. Between the 1940s and the 1970s something similar happened to economics. A rise in the standards of rigor and logic led to a much improved level of understanding of some things, but also led for a time to an unwillingness to confront those areas the new technical rigor could not yet reach. Areas of inquiry that had been filled in, however imperfectly, became blanks. Only gradually, over an extended period, did these dark regions get re-explored."

    Krugman got his Nobel prize for his work on geographical economics (the 'new economic geography'). Here he is (pdf) reflecting on that. His express purpose in creating his 'simple' core-periphery model was to show economists that geography mattered (as well as that you didn't need comparative advantage to make it work, so some places could become the core through endogenous forces alone). But the bait he used to lure neoclassical economists out of their dimensionless 'wonderland' was a model able to preserve general equilibrium.

    I'm getting to a point: actually, what Krugman showed with that model had been known empirically for a long time: it's all about cumulative causation, and the first theorist to apply it to trade, I think, was Myrdal. Plenty of others (including Jane Jacobs) noticed the basic core-periphery dynamic. So - had people like Mydral or Jacobs actually created workable models? Did Myrdal's lack of general equilibrium make his weaker? Krugman probably thinks so - he's very big on building models as part of the thought process.

    I'm not so sure, given everything that had to be sacrificed to build his core-periphery model. Krugman cautioned policy-makers that it was only a toy model, but it still went on to become the basis for a World Bank report, and to get its tendrils into all sorts of other policy areas. There's a recent case study of the birth of a theory and it's being let loose in the wild that hasn't been written yet.

    But anyway, to your question. Hypothesis: some cities or regions can become stronger simply by being slightly ahead in some tiny way, and cumulative causation can do the rest. Or: the history of the world's spatial economy would not turn out the same if you ran it again; it's not entirely physically determined. Did we need Krugman's model to tell us that? Was the idea refuted by a lack of model prior to Krugman making one? Actually, neoclassical economists would have said yes - hence Krugman wanting to show them the error of their ways in the only way they understood - a general equilibrium model. The rest of us, hopefully, don't suffer so much from an institutionalised streetlight effect. But if a model is descriptive but deeply empirical, is it still a model?

    Which relates to a second point: you make the comparison to climate skeptics, and the piece you link to does something that has a parallel to the climate skeptic approach. Economics-wise at least, I've spent a few years learning not to do it: emperor-is-stark-naked arguments. Five or six years ago I was going round being just as gob-smacked by the obvious stupidity of Milton Friedman claiming that people can be treated 'as if' they are utility-maximising agents. It actually turned out that 99.9% of people making that point hadn't read what Friedman had said, taking it completely out of context. Sound familiar? Here's the Friedman piece to read (pdf again): plenty to be critical about, but it's good stuff.

    As a comparison, physicists can treat collections of atoms 'as if' they can be described using heat, pressure and volume. They're even comfortable knowing that's not the whole story; it's just a very good working description at the level it works at and doesn't mean kinetics isn't important. Here's Feynman nailing that perfectly: 'When you explain a 'why', you have to be in some framework that you allow something to be true. Otherwise you are perpetually asking why.'

    Friedman was also very big on Marshall's approach to using theory: he believed Marshall 'took the world as it is; he sought to construct an 'engine' to analyse it, not a photographic reproduction of it' (Friedman 1953a p.35). Walras - who Schlefer pins the whole of economics on - in comparison, built what Blaug calls a 'a peculiar vision of a sort of 'realistic utopia' ' (Blaug 1997 p.569). Here's Brad DeLong talking about that point.

    So I don't like Schlefer's piece because it seems sensationalist and one-sided, it doesn't cover any recent economic developments modelling self-organisation (e.g. the entire of agent-based economics) and it's doing what climate skeptics do to climate scientists. On his basic point - 'there is no invisible hand'. I'd reframe it by asking, does he mean one of the following: 'if general equilibrium is economists' gas law, describing the macro stability of the economy, then it lacks the equivalent of kinetics'. Or is he saying, no self-organising feedback mechanisms - i.e. that can work through human actions without purposive central oversight - exist in the economy? Or, feeback mechanisms exist but no current model can help us understand them? (A nice overview of the direction that modelling self-organisation in the economy, minus 'assume an equilibrium has been reached' assumptions: the origin of wealth by Beinhocker.)

    None of that massive rant is to say there aren't serious issues in economics, but actually I'd be more worried about the stuff Krugman's been cataloguing recently: macro economics has about as solid a basis in empirical fact as you could hope for, but our global political structures, of course, don't have to listen to that. They can pick n mix economic theory to suit. Of course, if large swathes of our political structures are institutionally capable of ignoring actual physics, that shouldn't come as a surprise.

    Sorry, didn't know that was going to turn into an essay.

  2. danolner:

    To me, the key questions are simple: (1) Do there exist economic models that could predict the onset of the 2008 global financial crisis -- give or take a few years? (2) If such models exist, then what are they? (3) If such models do not yet exist, then do economists think of this as a major problem in economics that needs to be solved? (4) If they don't, then why not?

    -- frank

  3. Hmm, this reply button is above your comment, frank, hope it appears in the appropriate place.

    Two things: do there exist meteorological models capable of predicting exactly if, when and where hurricanes are going to make landfall, and how much damage they will cause? No. Does our level of understanding of hurricane seasons in general help us prepare for the worst? Yes. Has economics got to that stage? No. Does that mean it's entirely useless? No.

    I don't think it's fair to require economic models to predict catastrophic events, any more than you'd require meteorological models to do so, except in the very short term. I also think economic models are never going to be physical models since, as Gaz put it a while ago: "How would you physicists like it if you had to survey a bunch of molecules to find out what they planned to do, only to have most of them change their minds anyway, and the government restructure the laws of physics because of some opinion poll?"

    There's a lot been written on the 'physics of society' (Philip Ball's book is great), but at root you can't treat the problems as directly comparable.

    Some other good questions there, though. Dammit if Krugman hasn't written about that too, and very well.

    In his talk of 'mistaking beauty for truth', he's covering the same stuff I was discussing: a lot of economics went off in pursuit of "citadels of crystalline mathematical perfection that would shatter if touched by the harsh rays of reality" (Ball 2007 p.647). That's all the Walras stuff.

    The odd thing is - and this is only a pet theory of my own - the difference between that pursuit of virtual worlds and 'engines of analysis' can't necessarily be seen in the DNA of the models themselves. It comes down to how they are used, how they get their tendrils into policy - and especially policy legitimation. There's a vital difference between (1) treating an economic model as Reality and the meatworld as a poor reflection of them and (2) remembering that all models are wrong and, in economics especially, we should not always be asking them to provide us with predictive power. We certainly shouldn't ask them to do anything that even physical models can't do.

    Yes, plenty of people think the current state of economics is a major problem. Krugman makes that clear enough, but... well, is the financial crisis like the Earth being suddenly plunged into an ice age, thus completely stuffing all of climate science? No, but neither is it as completely PR'd and twisted as the CRU nonsense was by some, to make it appear there was a crisis where none existed. Nevertheless - and this is where I get cross with that piece - plenty of people with an axe to grind are doing exactly the same sort of triumphant baying.

    So why not e.g. get back to Marshall and away from Walras? There's also - the same kind of problem as with climate science - the issue of how you insert good analysis into policy, and build structures to stop it being politically hijacked or shouted down by power.

    Also, of course, how a functioning democracy is meant to manage the competing claims of different model builders, given that (J.C. Scott's point), states don't just use models to analyse, the state's gaze actually goes some way to making the world in its image (another problem economics faces that physics really doesn't!)

    Here's a recent Soros-funded example of people looking for alternative answers.

    I just think it's very, very important not to conclude that the entire history of economics has nothing to teach us. I know plenty of people, particularly on the left, want to boil the whole thing down to a simple "look, neoliberalism's been proved a sham and, oh, by the way, clearly we can't save the planet in a neoliberal framework" (Naomi Klein's saying that) - but it's just not a very helpful label for actually thinking through what *does* need to happen. I mean, what does 'get rid of neoliberalism' *mean*?

    Anyway, I'm wittering on again. Somewhere in there, some food for thought I hope.

  4. Very interesting Dan - I wish I had time to follow up all your links and educate myself a bit more about economic models.

    I wonder if the relevant question is something more like 'Do there exist economic models that show events like the 2008 global financial crisis?'

    But are economic models and physical, climate, models in reality so far apart conceptually that there isn't any meaningful equivalence to be drawn between them? That economic models model and feed in to human behaviour seems an important difference - does the efficacy of a climate model depend similarly on how it's used?

  5. My understanding is that "equilibrium" in economics has quite a tight meaning - the tendency of market systems to establish and maintain an optimally efficient level of output. The evidence for this is weak, but the claim has strong normative implications -for instance that political action will in most cases "disturb" a beneficial equilibrium, and is therefore to be avoided. A lot of human systems tend to equilibrium (for instance, political systems do not usually oscillate between autocracy and anarchy), but these generally result from active maintenance and enforcement, not from some innate tendency. I think it vfair to say that economics has been slow to explore these mechanisms.

    I would not draw too many parallels with models of the physical world. However complex that is, it lacks intentionality and the issues that come with it.

  6. Dan:

    True, climate models can't predict disasters with perfect precision. But they can predict a potential for disasters, and part of climatology is devoted to the study of attribution (which is where the anthropogenic warming theory is coming from), and moreover the problem of attribution is being integrated right into models of climate.

    In contrast, even though there have been no less than two economic bubbles in recent history -- the dot com bubble, and the housing bubble that led to the 2008 crisis -- there's still close to no work on trying to model the formation of such crises.

    Instead, bubbles seem to be treated as 'anomalies' that lie outside economic models. Efforts at attribution are pretty much ad hoc -- a possible exception being the work by Bouchaud and Guedj.

    And most importantly, economists don't seem to see this as a problem. And that I think is very telling.

    If the prevailing thinking, even among well-intentioned economists, is not 'yeah, we should really figure out more systematic ways to model the potential for crises' but rather 'we don't know how to model crises, this stuff is too hard, but trust us, economics is useful', then there is a huge problem.

    -- frank

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.