Winner of the New Statesman SPERI Prize in Political Economy 2016

Thursday 6 April 2017

Economists as medics

I got some stick on twitter the other day for my (longstanding) view that economics is in many respects like medicine. It is of course not exactly like medicine: as the man said, economics is an inexact and separate science. But think about what most doctors spend their time doing. They are in the business of problem solving in a highly uncertain environment in which they only have a limited number of clues to go on. They have solutions to a subset of problems that work with varying degrees of reliability.

If you read Dani Rodrik’s book Economics Rules (which if you have not you should, and can the person who borrowed my copy return it please!), you will see that economists have a large number of distinct models, and the problem that many economists spend their time solving is which model is most applicable to the problem they have been asked to solve. Where doctors have biology as the underlying science behind what they do, they also rely on historical correlations to see if the science is appropriate. Think about solving the problem of why there had been an increase in lung cancer in the middle of the last century.

The science for economists is microeconomic theory, now enriched by behavioural economics. Most of the models economists use are derived from this theory. But as Rodrik emphasises, the trick is to know which model is applicable to the problem you have been asked to solve. To help solve that problem, economists, like doctors, want data. Many have observed how journal articles are now more likely to be about investigating data than establishing theoretical results. Economists have recently started adopting the terminology of medicine in economic studies, talking about treatment effects for example. We both do controlled trials (for economists, mainly in development economics).

Sometimes the paths of the two disciplines cross (as they do all the time, of course, in health economics). One of the big empirical discoveries of recent years has been by Case and Deaton, looking at mortality rates of the US white population. Here is a key figure from their 2015 study.

Mortality has been falling steadily almost everywhere, except since just before 2000 among US whites. Focusing just on the US, the problem seems to be mainly for non-college educated whites (this graphic comes from here).

As with anything to do with race and class in the US, this work has been controversial, but some excellent analysis from Noah Smith shows that the problem suggested by the data is real enough.

Case and Deaton have a new study which tries to understand why this is happening. They describe it as evidence of ‘deaths of despair’. In each age cohort among this group, deaths from suicide, drug overdose or alcohol have been steadily rising. Some useful data is shown here. The interpretation the authors give for the despair is the decline in economic circumstances and status of the white working class in the US.

One of the factors that they describe as an ‘accelerant’ in this development has been the overprescription of opioids drugs that provide short term pain relief, but which have negative consequences in the longer term. US policy over the last 20 years has led to what some describe as the
“worst drug epidemic in U.S. history. Enough opioids are prescribed in the United States each year to keep every man, woman and child on them around the clock for one month.”

They go on
“It is hard to believe that medicine, which prides itself on empiricism, could have taken such a wrong turn.”

Of course individual doctors make mistakes all the time, but the profession as a whole can make major mistakes. It is of course subject to pressures from individuals and large organisations (drug companies). In this, again, it is like economics.

Consider this chart, taken from Alan M. Taylor, ‘The Great Leveraging’, NBER WP 18290.

The blue line shows the percentage of high income countries experiencing a financial crisis each year. Crises were endemic until after WWII, when it appeared for two decades or more that they were a thing of the past. In the 1980s they returned, but without any major impact on high income countries. Then there was Japan’s lost decade, and plenty of papers were written about how that was a particularly Japanese problem. The 2000s seemed quiet, and some called it the Great Moderation, until the global financial crisis arrived.

Looking at this chart, it is hard to believe that economics, that prides itself on its empiricism, could have made the mistake of believing that now things were different. But economics, like medicine, can make big as well as small mistakes. The point I want to make here is the different nature of the response to these mistakes from outside these disciplines. No one says that medicine has failed us, and we need to find fresh voices. No one will say that ‘mainstream medicine’ is in crisis, and we need to look at alternatives.

They do not say that because it would be stupid to do so. With the opioid epidemic something has gone very wrong and it needs to be corrected, and the same is true for economics and the financial crisis. So why the overreaction when it comes to academic economics? One reason is that doctors are not generally asked how long people will live, and even when they do their forecasts are not published almost every day in the press. Most economists are as honest as doctors would be about that kind of unconditional forecasting, but it suits the media to appear shocked and surprised when things go wrong. Another reason is that ordinary people can see doctors doing good things all the time to themselves, their friends and families, but the work of economists is felt less directly. It also seems intuitive that medics are in some sense better than economists, although how you could measure that I do not know. Both factors may explain why medicine is internally policed to a large degree (doctors can be stopped from practicing), whereas economics is not.

Another big difference involves politics. Economists bring unwelcome news to both left and right, so it suits both sides to occasionally bash the discipline that brings the message. We have seen a great deal of that from the right over Brexit. For the left more than the right there are also non-mainstream economists who have an interest in arguing that the mainstream has been corrupted by ideology. Quite why so many on the left choose to attack mainstream economics rather than use the mainstream to attack the right I do not know. All I do know is that they have been doing it for 40+ years, as I remember being told by many economists that the mainstream was fatally flawed back in Cambridge in the early 1970s, which was before Thatcher and Reagan.

But these differences should not obscure the similarities between economics and medicine. We both deal with people, and their mind and body can be pretty complicated whether as individuals, or as a society. In some areas we have developed quite detailed degrees of quantitative understanding that allow us to make successful interventions (more so than in other social sciences I suspect). In other areas we do things that work most of the time but sometimes fail, but there are many important areas where if we are honest we do not have any real idea of what is going on. So we make mistakes, which can sometimes be extremely costly for huge numbers of people, but we also learn from these mistakes.


  1. I also took a "economists as doctors" message from Rodrik's book. But the implication of this analogy is that economists require another step in their training before they should be making policy recommendations. There should be a period of internship of 2-3 years after graduate school where junior economists are supervised by senior economists as to how to select and apply models given the data. Without this structure, economists are more likely to use the models they developed in their graduate school papers as their go-to solution.

    1. Junior economists often get a job after an internship or are assigned to basic tasks (database maintenance, coding, fixing mistakes in repports...). You rarely get to have a say before you acquire experience.

    2. Interesting point. An alternative approach, which is in part closer to the one I folllowed (but that would emphatically not be a recommendation to any 'real' economist) is to spend the first few years after college doing micro-economics in a commercial/industrial environment (i.e, internal forecasting and market analysis in a company), then step right away doing non-academic, and preferably non-economic, work and finally return to a more academic environment (though not necessarily in academe) by when you may be able to apply both what you learned in college (and hopefully kept up with to some extent later) and what you learned in the big wide world.

      Having both been one of, and later examined for employment purposes, young academic economists I am quite certain that no one should be allowed to take up a responsible position as an economist until he or she has had some genuine, non-economic experience.

  2. Anonymous makes a point that occurred to me (although I don't know how practical it is) - anyway, however you do it, the key point is that the training of new economists around model selection might be worth looking at.

    For myself, two other thoughts:

    1) Economists don't get the same level of public trust over mistakes as doctors in part because "ethics" is visibly a part of the setup & ethos of the medical profession and not at all visible in the setup of economics. Indeed, economists often delight in (c.f. the Freakonomics tendency) demonstrating their superiority by taking on "conventional ethics." (To be clear, they are often right that "conventional ethics" are flawed, but attitude matters, sneering declarations are not a good way of persuading people of your good intentions.)

    2) I've been mulling Noah's recent Bloomberg piece on DGSE vs "simple models" and I think we need to start asking "how similar are these economies in fact?" Should we be expecting the same (working) model to work for the USA in 1957, 1977, 2017? Let alone for Japan, UK, South Africa, France, Australia as well? The short analogy for this is maybe economics isn't even medicine, maybe it is veterinary practice (with a side order of accelerated evolution thrown in) ?

    1. «we need to start asking "how similar are these economies in fact?" Should we be expecting the same (working) model to work for the USA in 1957, 1977, 2017? Let alone for Japan, UK, South Africa, France, Australia as well?»

      It is extremely sad that in 2017 “we need to start asking” about that. JM Keynes was pointing that out nearly a century ago:

      The result is that political economists see their work as a social discipline, that has a set of tools, including quantitative tools to investigate, with "discipline", social theories and "stylized" models in specific, historic, contextual political economies, and that different ages and different places require different figuring out and investigating different theories and "stylized" models, not merely universally valid microfoundations as in “science for economists is microeconomic theory, now enriched by behavioural economics”.

      But the latter stance is beloved by "end of history" and "TINA" advocates, including I guess those who make analogies with the study of unchanging (but over millennia), externally defined (by nature) body biochemistry and systems.

      «The short analogy for this is maybe economics isn't even medicine, maybe it is veterinary practice (with a side order of accelerated evolution thrown in)»

      But veterinary practice still is based on the assumption that there it is investigating an external, objective, unchanging (or slowly changing) system, that is a system that is not political, not social, not historical, not contextual, and not reflexive (in the sense of Soros).

      I think that our blogger intends the “separate” and “inexact” to argue that Economics is not like physics, but like biology, still a natural science, not a political one. Consider these points:

      * In physics there is an assumption that the object of study is an order of nature, with fixed, unchanging, externally determined laws. The assumption is that not only the shape of the laws is fixed, unchanging, externally given, but also the parameters are "physical constants"; so the mass of the proton or the speed of light can be determined once and for all, regardless of when and where it is measured, to say 1 part in a million, and further accuracy is possible given better measuring apparatus.

      * In biology there is an assumption that biochemistry and body dynamics are also subject to fixed, unchanging, externally determined laws, but given their complexity and the range of variation they can only be modeled and their parameters estimated statistically, that is it is still a “separate” but “inexact” natural science.

      For me the better analogy is with a cross between sociology and cliometrics :-). Studies of the political economy, especially at the macro level, are rooted in "sociology" and corroborated by "extreme cliometrics" :-). Note quite a an art depending on pure subjective judgement like literary critique, nor an inexact natural science like biology, but a "discipline", that is something that requires both systematic insight and quantitative analysis.

  3. Brendan Kennelly7 April 2017 at 08:30

    Interesting post, thanks. Perhaps a better analogy for economics might be with public health medicine and a better analogy for medicine might be with financial advice/planning

  4. "we also learn from these mistakes"

    Not the paid for hacks producing ideological justification for the reactionaries. The "we" are too few, the mercenaries too many.

  5. Mainstream "evidence based" medicine uses clinical trials as its testing ground. The results of those trials tell the medical establishment what works and what turned out not to work. As such, mainstream medicine exhibits the sort of rigour that I wish macroeconomics would try to expose itself to. Since macroeconomics isn't easy to experiment with perhaps meteorology or geology might serve as more applicable comparators eg the rigorous way that meteorologists make full use of the weather data stream to continuously assess whether they are on the right track or not

  6. Economists ― medics or barber surgeons?
    Comment on Simon Wren-Lewis on ‘Economists as medics’

    When economists are told that economics does not satisfy the scientific standards of material and formal consistency they invariably fall back on J. S. Mill’s slogan of economics as ‘inexact and separate science’.#1 This, of course, is merely one of economists’ numerous unacceptable excuses.#2 There is NO such thing as an inexact and separate science. There is only science and non-science respectively cargo cult science.

    Feynman defined cargo cult science as follows: “They’re doing everything right. The form is perfect. ... But it doesn’t work. ... So I call these things cargo cult science, because they follow all the apparent precepts and forms of scientific investigation, but they’re missing something essential.”

    What is missing among economists is a proper understanding of what science is all about. Wren-Lewis’s comparison of economics with medicine is the paradigmatic defense of a cargo cult scientist. To be sure, the representative economist can by NO stretch of the imagination be compared to a present-day medic. The present-day economist compares to the medieval barber surgeon who, more often than not, killed his patients with bloodletting and toxic medicine.

    Because economists lack the true theory their economic policy guidance has NO sound scientific foundation since Adam Smith/Karl Marx. Ultimately, economists bear the responsibility for mass unemployment and the social devastation that comes with it.#3

    Wren-Lewis argues: “ The science for economists is microeconomic theory, now enriched by behavioural economics.” The fact of the matter is that microeconomic theory is based on a behavioral axiom set that contains three NONENTITIES. Nothing of scientific value will ever come from provable false axiomatic foundations.#4

    Economics needs a paradigm shift from false Walrasian microfoundations and false Keynesian macrofoundations to true macrofoundations.

    Like most economists, Wren-Lewis suffers from the social science delusion. The subject matter of economics is NOT individual or social behavior but the behavior of the economic system. Economics is NOT a social science but a system science.

    Accordingly, the Case/Deaton study about the mortality rates of the US white population is the proper business of sociology and cannot be taken as an example for good economics.

    Until this day, economists have NOT gotten the foundational concepts of their subject matter, i.e. profit and income, right. This is like medieval medicine before the blood circuit was properly understood.

    Wren-Lewis’s conclusion is beside the point: “No one says that medicine has failed us, and we need to find fresh voices. No one will say that ‘mainstream medicine’ is in crisis, and we need to look at alternatives. They do not say that because it would be stupid to do so.”

    Indeed, this IS stupid in relation to modern medicine which is applied science but NOT in relation to modern economics which is a cargo cult science. The four main approaches ― Walrasianism, Keynesianism, Marxianism, Austrianism ― are mutually contradictory, axiomatically false, materially/formally inconsistent and all got the pivotal economic concept profit wrong. It is stupid, indeed, to defend this indefensible scientific rubbish.

    Egmont Kakarot-Handtke

    #1 Hausman, D. M. (1992). The Inexact and Separate Science of Economics. Cambridge: Cambridge University Press.
    #2 See ‘Failed economics: The losers’ long list of lame excuses’
    #3 See cross-references Employment
    #4 See ‘Economists’ three-layered scientific incompetence’

  7. Really? "We both [economists & doctors] deal with people, and their mind and body can be pretty complicated whether as individuals, or as a society."? Mind and body for medics is not the same even for behavioral economists (including those mostly commited to neuroeconomics). Most things we know about behavior (at least in Scientific Psychology) are just "self-reports and finger movements" (Baumeister, Vohs, & Funder, 2007) made on a computer screen or keyboard or even ratings from questionnaires. One interesting and worth thinking thing when considering disciplinary ouput improvements in this comparison medics/economists is another comparison made between the medicine's error evaluation system and civil aeronautics' error evaluation system (here,)

  8. Speaking as a doctor, and in fact an Addictions psychiatrist, who is a long-standing fan of your blog (but who usually is not in a position to contribute to the discussion), it is not true that 'mainstream' medical practice is never challenged by other policy actors in the UK. In England, opiate treatment policy has changed markedly in the past 9 years because of a change in the way that (pretty robustly) evidence based treatment has been seen by policy makers, the press and patients. If you are interested, here are links to the pubmed abstract of two of the papers that talk about it:
    I hope you find them interesting, if you have time to peruse them.
    Also - thank you for providing clarity, enlightenment and entertainment in these troubling times.

  9. As to "context-dependent" and "historically-rooted", and the difference with medicine consider "laws" like the "law of supply and demand" which is the core of the alleged “science for economists is microeconomic theory”: it does not say by how much supply and demand depend on prices, so it is a scientific law only in a very loose sense.
    In medicine the relationship between quantity A and B is supposed to be “inexact”, that is to have a statistical dispersion, but to be fixed: once it has been estimated it does not change again: the human body chemistry is assumed to be a complex system, and one with a dispersion of details, but is not supposed to change depending on circumstances, e.g. average body temperature is not imagined to be different between today and 100 years ago because human body chemistry had changed.

    This is really very different from studies of the political economy, because essential things do change across time and context: both in the mechanisms that matter and must be included in a model, and in the value of the parameters in the model, even if the mechanisms are the same.
    A model for the political economy of 100 years ago has to be rather different in both shape and value of parameters than that of the current economy.
    In addition there is the "political" aspect, what Soros has called "reflexivity", that is that human choices change the political economy, as it is a social, not a natural construct, and political economy agents use their discretion in teleological way, towards a personal purpose. The human body is a natural, not a social construct. The behaviour of its parts is not discretionary, is not teleological.
    The neoliberal/neocon claim that there is no political economy, no social aspect to its construction, and that therefore Economics is based on immutable statistical laws like medicine, is based on the "because TINA", "because end of history" assumptions and on the fantasy that human agents are not social, and their behaviour can be described once and for all, mechanistically if statistically, within a “science” like “microeconomic theory, now enriched by behavioural economics”, which is implicitly objective being a natural science, like medicine is. It is still the claim of "calculemus", even if statistically rather than exactly.

  10. Put another way I suspect that the argument here is that Economics is a DSGE model: microfounded on the “science” of “microeconomic theory, now enriched by behavioural economics”
    and resulting in immortal “models economists use are derived from this theory” that just need to be calibrated by looking for data, and that calibration is inexact and separate.

    The clever assumption here is in the statement that “doctors have biology as the underlying science behind what they do, they also rely on historical correlations to see if the science is appropriate”, which seems to me a big misdescription: they look at historical data with the assumption that the biology is immutable, both qualitatively and quantitatively, and therefore looking across either time or space with same models with the same parameters gives the same results, only the inputs change. That is if doctors look at the “increase in lung cancer in the middle of the last century” they don't assume that lung biology has changed in the middle of last century, and they need different models for 1980 and 1930, or they need different values of the parameters for the same model; they assume that the models and parameters are exactly the same, and it is the inputs, like the amount of tar inhaled via cigarette smoking, that have changed. The same for lung cancer between France and Denmark in 1980.
    It is I think correct to argue that Economics makes the same assumptions (or at most calibrates the parameters to the context, which is already a big difference with medicine), and in this claims to be a natural science like medicine and that the theory of microeconomics is based on unchanging laws like biology, even if they have to be interpreted statistically.
    But I disagree very much with the implicit claim that actually makes Economics a natural science like medicine and that microeconomics has immutable laws similar to biology and biochemistry, that is that Economics does not need to be concerned with the "political economy", because it is an objective natural science rather than a historical, contextual discipline.


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.