Winner of the New Statesman SPERI Prize in Political Economy 2016


Friday 25 August 2017

Medicine and the microfoundations hegemony in macroeconomics

Mainly for economists

I’m beginning to think I should have made much more of analogies between economics and medicine in discussing what I call the microfoundations hegemony: the idea that the only ‘proper’ macroeconomic models are those that have all their equations consistently derived from microeconomic theory. The analogy I have in mind is that biology represents microfoundations, and statistical analysis linking, say, smoking to lung cancer are the non-microfounded models. (I've used the analogy in other contexts.)

I was thinking about this in the context of a paper I have just finished which uses a diagram that Adrian Pagan used to describe different types of macromodels. The diagram, which you can find in an earlier post, has ‘degree of empirical coherence’ and ‘degree of theoretical coherence’ on the two axes. Particular macromodels can be placed within this space. At one extreme involving the highest theoretical coherence but weaker empirical coherence are microfounded DSGE models. At the other are VARs: statistical correlations between a group of macro variables with no theory-based theoretical restrictions imposed. In the middle are what I call Structural Econometric Models and Blanchard calls Policy Models, which use an eclectic mix of theory and econometric evidence.

If you have a simple view of the hard sciences, this diagram looks very odd. Theories either fit the facts or they do not. But I think a medic could make sense of this diagram by thinking about medical practice based on biology (for example how cells work and interact with various chemicals) and practice based on epidemiological studies. Ideally the two should work together, but at any particular moment in time some medical ideas may borrow more from one side or the other. In particular, statistical studies could throw up links which do not have a clear and well established biological explanation.

Now imagine the microfoundations hegemony in macroeconomics applied to medicine. Statistical longitudinal studies in the 1950s showed a link between smoking and lung cancer, but the biological mechanisms were unclear. The microfoundations hegemony applied to this example would mean that medics would argue that until those biological mechanisms are clearly established they should ignore these statistical results. The investigation of such mechanisms should remain a top research priority, but for the moment advice to patients should be to carry on smoking.

OK, that is perhaps a little harsh, but only a little. That some macroeconomists (I call them microfoundations purists) can argue that you should model and give policy advice based not on what you see but on what you can microfound represents something that I cannot imagine any philosopher of science taking seriously (after they had stopped laughing).        

19 comments:

  1. I look forward to the day when complexity theory is fully developed and adopted by macroeconomics such that the dilemma here posed, a perceived trade-off between microfoundations empirically verified versus coherent theory, will be understood instead as distinctive levels of complexity, in the way that biologists for instance understand biochemistry on the one hand (as their microfoundation) and emergent properties of complex biological systems on the other. I gather we are not there yet, but I believe we one day will, hopefully in the not too distant future (after my lifetime perhaps, but hopefully before the self-destruction and disintegration of the world economy).

    ReplyDelete
  2. The idea of quarantine existed centuries before Pasteur explained the microbial origins of disease.

    ReplyDelete
  3. I think another analogy you might find helpful is that between Statistical Mechanics (micro-founded) and the Gas Laws, so important in chemical thermodynamics.

    Most people with a certain level of education are probably familiar with the formula PV=nRT, which gives the kinetic energy contained by a given number of moles of an ideal gas.

    https://en.wikipedia.org/wiki/Ideal_gas_law

    Historically, this law is a combination of other laws, Boyle's Law, Charles' Law etc, which were discovered empirically.

    However, by making certain assumptions about the behaviour of gas molecules, one can derive an equivalent to the ideal gas law from first principles:

    PV = ⅓Nm(v²)
    P = pressure. V=volume, N = no of molecules, v is the speed of an individual molecule; (the v² should have a bar above it to indicate that it's the mean of the square speed)

    Amongst other things, this formula enables a definition of temperature in terms of the mean kinetic behaviour of molecules.

    https://en.wikipedia.org/wiki/Kinetic_theory_of_gases#Temperature_and_kinetic_energy

    BUT, as soon as a physical system becomes interesting, the ideal gas laws break down, or need extensive modification, and so do the kinetic/statistical mechanical equivalents, as the assumptions behind the kinetic theory of gases aren't realistic, although under certain circumstances they may be an acceptable approximation.

    In practice, the laws of physical chemistry are better determined empirically, although then working backwards to the "micro" foundations of statistical and quantum mechanics may then provide useful insights.

    My undergraduate physical chemistry textbook, by Walter J Moore states: "In principle*, all of chemistry can be derived from quantum mechanics, plus a few basic empirical properties of electrons, protons and neutrons ....."

    But it's the footnote connected by the asterisk which is the killer:

    *From "en principe oui", a French expression meaning "non".

    ReplyDelete
  4. The real issue is something called "emergent properties." Let's look at chemistry (even better than medicine as an analog). No competent chemist would be ignorant of quantum mechanics, which is the equivalent of microfoundations. QM makes everything go. Yet chemists don't spend their days solving QM equations before they mix stuff in text tubes. Why? Because the realm of chemistry is the next level of "physical organization," those chemical bonds between atoms. Now, QM explains why those bonds exist, why they do what they do, etc. But chemists rely on properties they know about the bonds themselves: stereochemistry (which in the social sciences would be "structure"), type of bond (single, double), and so on.

    Now, you could do good chemistry without knowing QM and just knowing about bonds, stereochemistry, etc. But knowing QM not only helps explain WHY things happen, but can also let you see & try out reactions you didn't know you could do, or explain reactions that otherwise would not make sense.

    Long story short, you do want microfounded models 1) for the sake of knowledge (you know the how & why), and 2) to use that to reveal possibilities you otherwise would not have known if you stay at your level of analysis. If, however, like a good "social chemist," you focused on the emergent properties--how upper-level configurations have their own independent properties and outcomes--then you could do your work and probably do it well. Krugman sometimes hints at this with his stuff on IS-LM.

    And even if preferences, tastes, expectations, etc. shifted at the micro level because of previous events & policies (which is sort of what the Lucas critique suggests), this might not ruin your model. Take "power." There is a structural side, one form of which is that, all else equal, more people on one side will win over a smaller number of people. Now, if we add norms & expectations, institutions, etc., we can reduce the effect of numbers--but we can't eliminate it.

    Georg Simmel really took a good step in this direction, with structural thinking--that structure itself has properties independent of those of the actors. Macro could do the same.

    Last point: this is where the social sciences are screwed up. Discipline int he natural sciences are organized hierarchically: from the smallest to the largest. Each discipline really is dealing with a level of physical organization & its own emergent properties: QM (physics) --> chemistry (bonds) --> biology (systems). The social sciences are organized vertically: all levels in econ, in soc, in polisci, etc. But our boundaries are entirely artificial: the economic is never divorced from the social & political. So we end up with these screwed-up assumptions and approaches the confound what they should unravel, and unravel what they should be discovering.

    Moral of the story: medicine is applied science. Think in stead of biology & chemistry. Oh yeah, and look at what some political scientists and sociologists are doing with structure (e.g. Ronald Burt).

    ReplyDelete
  5. Very interesting. On smoking, though I don't know of a great deal of argument from medics in the 1950s, exactly your 'microfoundations hegemony' line was certainly taken by the great geneticist and statistician Sir Ronald Fisher. See his contributions gathered together at https://www.york.ac.uk/depts/maths/histstat/smoking.htm , and there's more at https://priceonomics.com/why-the-father-of-modern-statistics-didnt-believe/

    ReplyDelete
  6. Maybe macroeconomics should look to engineering instead of science as a way to form models. We have our theories but the equations can get complicated making it easier to just do the experiments, and then create reference tables for later use.

    ReplyDelete
  7. Cause and effect relationships in medicine are, as in economics, regarded as controversial if differing interpretations are offered to coherently explain the available data. In medicine the controversies are generally resolved over time as knowledge expands and techniques for observation and measurement reveal which interpretation is supported by the further development of knowledge. During the controversial period, we often see one research unit produce research supporting the chief's known opinion regarding the controversy. Once the controversy is ended, everyone moves on to new questions. The resolution of controversy is regarded as progress. In economics, the same people hold the same positions on controversial matters for their entire careers. When this occurs, the basis for holding to a belief appears to be political. Improvements in techniques of observation and data collection over time do not appear to resolve controversies in economics as they do in the practice of medicine or engineering. Rather than a search for objective truth, the practice of economics appears to be a contest to win political support for one's own opinion.

    ReplyDelete
  8. See this from Robert Sapolsky (Stanford Biologist and Neuroscientist): https://www.google.com/url?sa=t&source=web&rct=j&url=%23&ved=0ahUKEwiFudTQ2fTVAhUG5GMKHatiCOEQwqsBCCowAA&usg=AFQjCNF1sHlk2z9QIx28jtvfnGP_UlkifA

    ReplyDelete
  9. Interestingly A clinical Scientist I am acquainted with drew an analogy that economic modelling can in no way be compared to biological structures, as in Biology you can track each micro organic trace back to its origins, whereas in economics it's like the weather and a gust of wind can unexpectedly blow your theory out of the water.

    https://www.youtube.com/watch?v=_rUk-r0G1eQ


    ReplyDelete
  10. I'm curious, do the microfoundations purists even consider the accuracy of their microfoundations? I seem to recall that Milton Friedman argued that accuracy of the microfoundations didn't matter. And I also seem to recall all too much outright rejection of the argument that micro theories should be reconsidered if they predict incorrect phenomena.

    About medicine and smoking (medicine and sleep deprivation, too): I think the history is perhaps more like the microfoundations purists and what they think. The statistical connection between smoking and lung cancer (or at least health danger) was dismissed -- as pseudoscience, or otherwise. That was true even with the obvious fact that smoking involves inhaling smoke into one's lungs.

    19th Century had physicists and chemists deriving the (phenomenally-already-established) macroscopic laws of thermodynamics. The results were both accurate and inaccurate in a most nerve-wracking way. They knew that they were on the right track, but that there was a huge problem. They had no idea how to fix things; it ultimately took quantum theory to resolve the problems.

    I think that this illustrates the huge difference between certain schools of economics and actual science.

    ReplyDelete
  11. I'm still smoking. My sister, who didn't, is dead, from lung cancer.

    "they should ignore these statistical results" - quite. And most of what we read / hear, apart, obviously, from your good self.

    ReplyDelete
  12. I'm glad that you are making these sort of searching evaluations of how macroeconomics gets done. I'm still very troubled and confused though by the case you have made on previous posts that macroeconomics would have nothing to gain by adopting a culture of systematically evaluating forecasts made using various macroeconomic approaches. As an example of a missed predictive opportunity : in the last few decades, South Korea and Botswana transitioned to using government debt denominated in their own local currencies. Some people (eg MMT ) would predict that that would have had a transformative positive effect. Other economists (such as the standard IMF guidance) would predict that there would be problems from "crowding out" or their other mysterious (to me) justifications for instead promoting hard-currency (eg USD) denominated government debt. If prospective predictions had been made, then we would be on the way towards debunking.

    ReplyDelete
  13. Speaking as both a biologist and a doctor - by which I mean, I have a degree in molecular biology and am currently doing some research which is very much about basic science and am a practicing surgeon...

    Speaking as both a biologist and a doctor, I completely agree with your analogy.

    The epidemiological data often throws up associations not previously suspected or explained by knowledge of the biological processes. This is not surprising because biological systems are immensely complicated and our knowledge is certainly far from complete. Incremental studies gradually add to the picture and the interplay of genetic and environmental factors is difficult to understand. The great quote on this is: "If everybody smoked, then lung cancer would be a genetic disease." The point is that most people who smoke don't get lung cancer (if fact most smokers die from heart disease) yet smoking causes lung cancer. Some people are more genetically susceptible than others and hence when exposed to smoking develop the disease. And that is one of the simplest interactions around. It is also the case that things that work in the laboratory do not always translate to clinical practice. The only solution to this problem is to have big data - big studies with lots of patients to see the effects.

    Conversely, such studies suffer from the inherent weakness that controlling for all factors is difficult. Which is partly why you see reports in the media about doctors changing their minds and offering one piece of advice one week and a different piece the next. Often this reflects how someone did a small study which threw up something unexpected which warranted a bigger study to support or refute this finding. Or it could be that the linkage was found to just be more complicated. It is about a progressive process of accumulating knowledge and evidence. As with any science.

    Conversely there are plenty of examples of things that work in vitro but do not translate well to clinical practice. The fact that we cannot explain why they do not translate is yet more evidence of how complex a system we are working with but does not change the fact that basic science is not enough when deciding on which treatments are effective and which are not.

    Hence the best of medical science is where the epidemiology and the basic science agree. The data of smoking-related cancer is overwhelming. This is combined with a good understanding of how the countless carcinogens in cigarette smoke damage the lung cells and lead to cancer. But the epidemiology predated the basic science by decades.


    Interestingly, I think that both fields are fighting the same battle in the public sphere around what we know and don't know. The main difference is that, fairly or unfairly, we start from very different places. Medicine, on the whole is well respected and for the most part the critics are seen as quacks. With many notable exceptions such as the vaccine debate. Conversely the preconception that economists are basically just making it up to suit their own ends is quite prevalent.

    It seems to me that their are a lot of economists doing some very good work out there. I also get the impression that there is a weakness in the field of some who are only interested in good theoretical models - regardless of how much the model may be refuted by empirical data. As I understand it, Ricardian equivalence is an archetypal example of this.

    It is deeply frustrating when faced with this propaganda war - by which I mean, I have lost count of the conversations I have had with worried patients triggered by the latest misleading story in the media.

    AFZ

    ReplyDelete
  14. A bitter pill for political economists
    Comment on Simon Wren-Lewis on ‘Medicine and the microfoundations hegemony in macroeconomics’

    When economists are at the end of their wits they turn to analogies and metaphors. Simon Wren-Lewis is a case in point: “I’m beginning to think I should have made much more of analogies between economics and medicine in discussing what I call the microfoundations hegemony: the idea that the only ‘proper’ macroeconomic models are those that have all their equations consistently derived from microeconomic theory.” And he concludes: “That some macroeconomists (I call them microfoundations purists) can argue that you should model and give policy advice based not on what you see but on what you can microfound represents something that I cannot imagine any philosopher of science taking seriously (after they had stopped laughing).”

    Nobody can take this methodological absurdity seriously. The fact to start with is (i) that microfounded macro is one of the worst blunders in the history of modern science, and (ii), that the representative economist ― Simon Wren-Lewis included ― cannot get his head around it until this day.

    The political economist has an instrumental relationship to science and uses theory as a means to an end. Given his political priorities, he simply suspends theory if it comes in the way of agenda pushing.

    The difference between scientist and political agenda pusher is this: “A genuine inquirer aims to find out the truth of some question, whatever the color of that truth. ... A pseudo-inquirer seeks to make a case for the truth of some proposition(s) determined in advance. There are two kinds of pseudo-inquirer, the sham and the fake. A sham reasoner is concerned, not to find out how things really are, but to make a case for some immovably-held preconceived conviction. A fake reasoner is concerned, not to find out how things really are, but to advance himself by making a case for some proposition to the truth-value of which he is indifferent.” (Haack)

    Economics claims to be a science and, by consequence, the economist has to satisfy scientific standards and NOTHING else. Scientific standards are well-defined since antiquity: “Research is in fact a continuous discussion of the consistency of theories: formal consistency insofar as the discussion relates to the logical cohesion of what is asserted in joint theories; material consistency insofar as the agreement of observations with theories is concerned.” (Klant)

    If the facts contradict microfounded macro the latter is refuted has to go out of the window. It is as simple as that. It is NOT possible to say, if a theory contradicts the facts I suspend the theory for a while and march on with analogies, metaphors, common sense, and populism. A scientist under NO circumstances does what is second nature to a political agenda pusher: “In order to tell the politicians and practitioners something about causes and best means, the economist needs the true theory or else he has not much more to offer than educated common sense or his personal opinion.” (Stigum)

    Microfounded macro is NOT the true theory. In fact, Walrasian microfoundations are false since 140+ years.* Unfortunately, Keynesian macrofoundations are false since 80+ years. Both approaches have already been dead in the cradle without economists ever realizing it. What any philosopher of science has to say to the representative mico- and macrofounded economist (after they had stopped laughing) is: You are fired!

    Egmont Kakarot-Handtke

    * First Lecture in New Economic Thinking
    http://axecorg.blogspot.de/2017/05/first-lecture-in-new-economic-thinking.html

    ReplyDelete
  15. This seems to me sensible but perhaps a little late.

    Maybe I am wrong but I am almost sure that we have read a series of posts (going back to the early years of this blog) saying, with various degrees of emphasis, that while microfoundations have their limits there is really no alternative to using the consistent approach they purport to enforce.

    For those of us inherently sceptical of such an approach (if only for the reason that my economics education preceded the invention of such new-fangled approaches) this latest position is a long-awaited breath of sanity. It is not simply a question of favouring intellectual rigour over real world experience. I have long argued (well, ever since I learnt of their existence) that the establishment of micro-founded analysis in a sense conditions the answer that models can give: if every actor is constructed as a profit-maximising atom then the model pretty much must prove that profit-maximisation must be optimal.

    ReplyDelete
  16. I'm happy to read the common sense in this post - having engaged with the profession at a time of particular microfoundational fundamentalism. One other point I would make is that there is one type of crucial evidence the profession still really does like: unquantifiable evidence. But for example, if you really want to understand things like a liquidity trap, you need to trawl through the archives and see what borrowers and lenders say about their own lending and borrowing behaviour.

    NK.

    ReplyDelete
  17. The problem with microfoundations is that you can't derive the map to translate a micro founded inference onto a real economy, and you can assume any such map is nonmonotonic anyway, so all MF inferences are necessarily inapplicable to the real world. Yes, a philosopher of science laughs at MF, as do most mathematicians.

    ReplyDelete
  18. Philosophers of Science call this use of what you call micro foundations to establish macroscopic laws "Reductionism".

    But the biggest laughter from a Philosopher of Science for the economic fad for reductionism in macroeconomics is the lack of any rigor to it. We're all well aware that microeconomics is not attempting to reject any microeconomic models that cannot be shown to be founded on rigorously established psychological laws.

    This is not principled reductionism. This is an attempt to privilege theory over empiricism.

    ReplyDelete
  19. Don't predict the weather until you can do it via motions of individual molecules and atoms

    ReplyDelete

Unfortunately because of spam with embedded links (which then flag up warnings about the whole site on some browsers), I have to personally moderate all comments. As a result, your comment may not appear for some time. In addition, I cannot publish comments with links to websites because it takes too much time to check whether these sites are legitimate.