The Pseudoscience of Paul Samuelson


In 1993 Reason asked a bunch of libertarians to nominate "a book published in the last 50 years that is significant because it has helped promote wrongheaded ideas with serious consequences." One participant, the economist and historian Robert Higgs, wrote about Foundations of Economic Analysis, the influential textbook by Paul Samuelson. Since Samuelson just died, it's worth looking back at Higgs' comments from 16 years ago:

Samuelson fashioned his models, which set the standard, after 19th century physics. Functions were assumed to be smooth and continuous. Economics was reduced to various types of the same calculus problem: finding a constrained extremum. The economist's job was to state the objective function and the constraints, then grind out the solutions. This required considerable mathematical ability and stomach for tedium but little imagination and no familiarity with economic reality.

By the 1960s, if not earlier, academic economists who quarreled with this way of doing the job were, as Roy Weintraub put it, "regarded by mainstream neoclassical economists as defenders of lost causes or as kooks, misguided critics, and anti-scientific oddballs." By aping 19th-century physicists, neoclassical economists convinced themselves and others that they were doing science, but the effort was basically misguided, not so much scientific as, in F.A. Hayek's term, "scientistic." Human beings, purposeful and creative, are not like atoms; nor is a market analogous to a physical or chemical system. In the view of Hayek and his teacher Ludwig von Mises, neoclassical economics is, in critical respects, pseudo-science.

The whole thing is here. For more from Higgs on Samuelson, go here and here. For more from Reason on Samuelson, go here.


NEXT: Praying Lieberman Does Not Alter This Deal Any Further

Editor's Note: We invite comments and request that they be civil and on-topic. We do not moderate or assume any responsibility for comments, which are owned by the readers who post them. Comments do not represent the views of or Reason Foundation. We reserve the right to delete any comment for any reason at any time. Report abuses.

  1. I am too just a mindless atom awaiting the proper stimulus from A Scientist to achieve my pre-ordained outcome.

  2. Functions were assumed to be smooth and continuous.

    Eat it, bitches!

    1. Q: Did you hear about Cauchy’s dog?

      A: It left a residue at every pole.

  3. I like kicking the shit out of corpses as much as anyone else, but this many posts on the same dead statist is a bit much.

    Unless Samuleson’s death and the cowardly disappearance of Lonewacko are somehow linked.

    1. cowardly disappearance of Lonewacko

      Odd, too, that shortly before he disappeared he started leaving posts with actual content and no links to his site.

    2. LoneBoner didn’t disappear; Jabba had him frozen in carbonite. For being an annoying asshole. You can free him if YOU’RE HOLDING A THERMAL DETONATOR

      1. This poster is my kind of scum

  4. Of course, human beings may not be atoms, but we’re made up of them… one day all those variables will be able to be accounted for! Until then, RIP Sam.

  5. Another of Samuelson’s great achievements was his “proof” that stock prices are normally distributed. Mandelbrot totally eviscerated this theory. If you want to read the details, I recommend his excellent book “The Misbehavior of Markets”. I’m not really sure what Samuelson has actually contributed to economics. He has certainly changed the way economics is done, but has he really added anything to our understanding of the economy?

  6. “Human beings, purposeful and creative, are not like atoms; nor is a market analogous to a physical or chemical system.”

    Let’s ignore the begging and assume that human beings are purposeful and creative. Still, there are these regular patterns we find among them in the aggregate. For example we can find consistent patterns in things like suicide or arrest rates (where every year place x or group y will have higher rates than place a or group b). This is what the social sciences focus on. Even more there are certain predictable ways that individuals will think and behave, this is the foundation of psychology. So while it is true that human beings are not like atoms I’m not sure why that means we can’t study human beings using the scientific method more profitably than we could using some other method. From what little I know about it sub-atomic particles are’nt like atoms either but we can develop a science about them.

    1. “ignore the question begging” that is

    2. I suppose in theory you could. The problem is that people are not lab rats. So you can’t conduct controlled experiments on them. Therefore it is very difficult to make predictions and study results that are not free of other variables. Social science is often reduced to just crude correlation. We know that criminals tend to be poor people. But the world is much to complex to determine which way the causality goes. And we can’t ethically capture a bunch of poor people and do controled experiments on them.

      We have been doing this kind of crap for over a 100 years now and haven’t achieved one positive result that I can see. It is just a cargo cult.

      1. This just means social science is hard, not impossible. And that it should be careful and humble in its conclusions. That’s sensible. But that its all useless pseudo-science is nonsense imo.

        1. It is not pseudo science like say astrology since it is possible that it could someday be usful. But it is psuedoscience in its current form since it has yet to produce anything but bunk results. I can’t think of a single area of life where the social sciences have produced knowledge that has made it better.

          1. I’m not sure what you mean by “bunk results.” Do you mean false or useless? For example take the well documented “aging out” phenomena in criminal offending. We can’t predict that kid x or adult y will commit a crime this year, but we can be certain that the age distribution of whoever does commit crime will be skewed towards the young end. A huge chunk will be committed by 15-29 year olds and a much smaller chunk for older groups. That’s certainly true, its been replicated from culture to culture and timre period to time period. And its kind of useful (we know which groups have the highest “risk” of committing crimes).

            1. All that is crude correlation. Anyone can read the newspaper and figure that out. It says nothing about why young people commit crimes or what to do to prevent them. Indeed, the moment social scientists step beyond crude statistical analysis and try to provide something usful, they fall on their faces.

              1. I’m not sure why the fact that it is “crude correlation” makes it uninteresting or useless. And I’m not sure it doesn’t tell us something about why people commit crimes (it fits nicely with the idea that impulsivity is a large cause of criminal behavior). It’s something true about the world. Part of what science is involves making statements about what is the case and what is not.

                1. It does tell us something. But what it tells us is very fleating. I am not saying that we shouldn’t think about these issues. I am just saying that when we do, we are doing philosophy not science.

                  1. I actually kind of agree with MNG here.
                    There’s plenty of useful things you can study and say about society.

                    Where Samuelson goes wrong is in attempting to write precise mathematical formulas using only a few variables, and then base policy on that. It’s a conceit that the economy is reducible to a few simple interactions in the aggregate. That doesn’t mean you can’t use math to say some theoretical things about how economies function, for instance. You just don’t want to make specific predictions about the future demand for blue jeans, for instance.

              2. Why does gravity work?

                1. To support freeloadin’ SOBs!

    3. The problem is that there are simply too many variables, and economists like Samuelson pretended that there were only a few, and that a nice smooth continuous function could be derived from them.

      Please go read Hayek’s “The Fatal Conceit”.

      1. Agreed. It’s valid to say that a market seeks a constrained optima. It’s a conceit to think that a human being can accurately calculate the position of that optimum in a real economy. Complex systems and all that.

        That’s precisely the point market economists make – that the distributed computing power of a billion human beings making locally optimal choices for themselves acorrding to billion of local variables gets you closer to the global optima than one guy attempting to compute the optima from an office in DC using an equation with only 4-5 variables.

        1. Your thinking disturbs me.

  7. The rise of the pseudo science known as “social science” in the 20th century will be remembered as one of the century’s biggest and most damaging blunders. Economics, political science, sociology, and the rest have foisted any number of horrible ideas on the world backed up by “expert opinion”.

    1. The Science Is Settled!!

    2. Postmdernish on the social sciences. Gotcha.

      1. No not at all. Just know a cargo cult when I see one.

        1. If it’s really “no not all” then you should watch such broad generalizations as “The rise of the pseudo science known as “social science” in the 20th century”.

          1. Go read your ass some Allen Bloom.

  8. People that say social science is impossible then usually within a minute will say something like “minimum wage laws reduce employment”, in other words social science…

    1. Anyone who says minimum wage laws always reduce employment is overstating the case. The can reduce employment and usually do depending on the circumstances and the nature of the laws.

      Regardless, how is knowing that raising the price of labor will usually reduce the demand for it really science? It is just common sense but it is not science. You could imagine cases where that wasn’t true. At best it is a case where you say “usually it works this way”. That may be usful but it doesn’t strike me as a natural law.

      1. Well, no, actually you can collect statistics on minimum wages and employment, and draw some pretty straightforward correlations between minimum wages and employment levels.

  9. The very fact that you could imagine cases where that wasn’t true makes it amenable to science, it’s falsifiable.

    1. Okay, even if supply and demand are in some sense scientific, so what? It seems like there are a pretty limited range of things in social science that are as rock solid as supply and demand.

      1. Well, at least you now agree that there are some areas of social life amenable to scientific examination. Progress?

        1. Not really. Everythign that is true is not scientific. Somethings are self evidently true without being within the realm of science. Supply and demand is true. But I am not convinced it is scientific. It might be. You can do controled experiments on it and do some pretty neat mathematical analysis of it. But even if it is, it is the only thing in social science that is.

  10. On the topic of lieberman analogies……..lieberman/

  11. I am surprised that Samuelson has a following considering the amount of work that has been done on recursive systems.

    The problem is not merely that there are a lot of variables, it is that some variables are mutually dependent. As one changes, the other changes and that cause the first one to change.

    Economies are a perfect example of Chaos Theory in action.

  12. I’m just going to suggest that people who have problems with Samuelsonian mathematical economics take a look at an essay by a man with unimpeachable libertarian credentials, Friedman’s “The Methodology of Positive Economics”, as excellent a defense of modern economics as has ever been written.

  13. John,

    I think you are off-base with your assessment of entire fields of scientific inquiry.

    That said, economics is the softest of the soft sciences, despite all the fancy math.

    But the reality of science is that any question worth knowing the answer to is probably going to involve complex systems that are not amendable to naive reductionism. That doesn’t mean that observational (as opposed to experimental) science can’t move us forward significantly in our understanding.

    1. I dunno, economics has more in common with the biological sciences – at least ecology and psychology, than it does with (say) culturual anthropology. Ecology and (obviously) climate science are both on the same shakey mathematical grounds. All trying to reduce hugely complex systems to a few variables.

      At least microeconomics has something going with game theory and psychology. It’s macroeconomics that is conceited.

      1. Hazel,

        To me it seems that the way to judge the softness of these endeavors is to build a chain from the physical sciences to the science at hand. To get to economics, you have to move through the data from psychology, which relies on neurobiology, which relies on chemistry and physics, etc…

        It seems to me that climate science is on par with neurology more than economics as the variables it is working with are all physical measurements.

        1. The important point being that climate science can use our understanding of chemistry and physics to validate their proxies. Economics has to rely on psychology to validate its proxies.

          1. But the climate isn’t solely dependent on chemical and psysical inputs. It’s also dependent on ecological processes – algae in the oceans, plant growth, animal life.

            I agree with your general hierarchy of dependencies between the different sciences. But I don’t think you can necessarily make direct comparisons between degrees of complexity based on the place in the hierarchy. The sciences aren’t just a big pyramid, they are also a branching tree. Ecology is dependent on a lot of complex interacting variables, just as neurobiology is. But some variables are more nonlinear than others, and some things have more different kinds of elements interacting(as opposed to millions of similar elements). I don’t think you can just say “economics is based on human behavior, which is based on psychology, which is based on neurobiology, therefore economics must be more complex than neurobiology”. usually when you move from one “level” to another, you’re abstracting the low-level stuff away and dealing with emergent elements that might be very simple compared with the underlying substrate they are based on.

            1. Hazel,

              I have no quibbles with any of these comments. I was talking broad strokes. Of course when we look in more detail it is a more complex issue.

        2. Softness isn’t well defined, so we can each define it as we like, but what’s useful to get at here is which fields quantitative approaches are or can be most fruitful in. I don’t think that measure of proximity has very good correlation with quantifiability. Population genetics requires taking assortive mating into account, which requires that it goes through behavioral sciences, yet population genetics takes quantitative methods very well since assortive mating can be treated fairly well with a few empirical constants.

          With regard to economics specifically, most of the psychology is smooshed into preference functions, which are treated as essentially arbitrary and not directly measurable, so for the most part it doesn’t matter what the psychology says. You don’t need a detailed psychological model of preferences if all your model needs to know about them is that they’re monotonic.

          1. Again, I have no quibbles with anything you are saying here.

            I think economics faces a lot of the same challenges as psychology, however, in defining what their proxy measures are actually measurements of. So, for an example from psychology, a score on an attention test is used as a proxy of attention, but there is no way to verify that the performance is a result of attention rather than say, motivation. In the same sense, measures in economics may have a nice name like “preference” but when you look at the details of the measure, there is no way to validate that preference is what is being measured.

            Climate sciences, on the other hand can draw a path through physics and chemistry to detail the relationship between their proxy and the thing they claim it measures.

            1. I’ll agree that economics is much more complex than climate science. But that doesn’t mean that the climate is simple enough to be reduced to a few variables, or be accurately predicted.

              We pretty much know that the climate as it currently exists is almost entirely a product of life on the planet. So, it’s safe to say that biological variables play a huge role in it’s maintainance.

  14. “That said, economics is the softest of the soft sciences, despite all the fancy math.”

    [citation needed]

    1. No need for a citation when someone is clearly stating their own opinion. What’s your beef?

  15. Neu,

    This struck me as exceedingly odd:

    Still, with so few people to govern, New Mexican authorities can really go to town on the minutia of regulatory laws. Years ago a friend’s mother hit an elk on her way home from work. When she called in to report the incident, an official from the Department of Game and Fish informed her that the elk she hit was technically the property of someone else on the waiting list for claimants of roadkill; she, by virtue of hitting her own large animal, would now be put on the queue as well. The origin of the waiting list is unknown, though we suspect that whoever first hit an elk in New Mexico failed to reap what he sowed. It’s been years since that particular elk met its end, but my friend and his family has yet to experience the joys of elk jerky and elk stew.

    Ever heard of the roadkill registry?

    1. I have not. But it sounds very much like a NM thing to do. NM’s legal tradition is a bit different than most states in that it has influence from the Spanish system. That system took a much more colletivist approach than English common law. Communities had rights rather than individuals (to over simplify in the extreme).

      I also wonder if the source of this may have something to do with tribal treaties.

      1. Tribal treaties? I wonder if it’s about where you run over an elk…

        This can only be solved by you going out and running over a series of elks in various locales and seeing which ones you get to keep.

      2. Communities had rights rather than individuals (to over simplify in the extreme).

        Which is sort like saying that the “big man” in town got to decide what to do with windfalls, isn’t it?
        The Spanish system also seems especially prone to takeovers macho men.

  16. I think you people misunderstand what I am saying. I am not saying that we cannot study or think about society. Nor am I saying that we cannot draw usful conclusions about it.

    What I am saying is that those conclusions can never rise to the certainty of scientific inquiry. Yes, you can draw lots of interesting correlations about the actions of people. You can even predict those actions occasionally in a very macro sense. But what you can never do, or least no one has yet been able to do, is take the study of people and society and reduce it to mathematical or even probable certaintude. It is not the art of the social sciences I object to. It is the conceit that they are able to produce results and conclusions with scientific certainty.

    And that is even the best of social science, of which economics is probably the best. At their worst, social sciences have produced any number of horrible and useless ideas and practices that have done more harm than good. For this I think psychology and the entire pschoanalysis and treatment racket is the best example.

    1. But “scientific precision” is not an absolute. It is relative to the question at hand. I think many social science questions can be answered with the tools available.

      If you are concerned with people who over-interpret their findings or apply methods incorrectly to a specific problem, then I guess your complaint resonates, but you original post was casting aspersions on the whole enterprise.

      Psychology as science needs to be kept separate from clinical interventions in this kind of discussion, it seems to me. Not that science can’t be used to demonstrate that a particular treatment works…it is just that is a different category of science than psychology aimed at understanding how the mind works.

  17. The problem with economics at the peak of it’s physics envy wasn’t the endeavor to create mathematical models, but a failure to recognize the limitations of those models. Often only boundary cases (equilibrium, prefectly stable prices, etc) are resolvable to usable models but none of the boundary conditions are reasonable assumptions for the economy at a given time.

  18. nor is a market analogous to a physical or chemical system

    Indeed. It is much more analogous to a biological system.

Please to post comments

Comments are closed.