Friday, June 22, 2012

Some economists need to accept the fact that we're scientists

Social scientists can have really weird hang-ups about this. I don't know why. Science is a method for understanding the world around you, it's not some secret club that you're not allowed into because you have some inferiority complex around physicists. Get over it. We're just highly evolved apes. Do you think other people who study other apes with the scientific method are scientists? OK - then you're a scientist.

Yes, you do science that has immediate implications for human welfare and lots of normative questions lurking around the corner. So do lots of other scientists.

Anyway - a very strange hang-up.

In discussing the Manzi quote yesterday about whether the number of stars in the galaxy was odd or even, I was reminded of this news piece from 2010.

A Yale astronmer believes that the traditional method for counting (counting, nothing fancier than that - just counting) stars is providing biased estimates (umm... sound familiar to you econometricians?), and he wants to triple the number of stars out there. The idea, while not entirely accepted, is being taken very seriously. The guy is making a good point apparently - not a crackpot or anything like that. Whether it's a correct point needs to be sorted out, but it's apparently a good point.

Other astronomers are a little worked up over it, and the guy making the claim even admits "it's a big pain". Why? Because they have to rethink a few things and recheck things, etc.

Notice what you don't see them saying. You don't see them saying "astronomers have a huge range for the potential number of stars out there - this is clearly not science" or "this guy is a hack he's clearly not a scientist".

The fiscal multiplier is a very tricky thing to pin down, both because we don't observe a lot of data and the data series that we do observe are simultaneously determined. Tricky stuff. But multiplier estimates still seem to fall in a fairly narrow range. Of the estimates people generally trust, the largest estimate is no more than triple the smallest estimate.

And that's pretty good, particularly since multipliers might be expected to vary with institutional settings anyway.

Astronomers don't go into soul searching over this sort of thing - why do economists? Why do we have people out there making such a big deal over uncertainty and error?

If you can't deal with uncertainty and error, you shouldn't be doing science.

I think there are two major reasons social scientists do this - one of which I've stated before:

First, people get weird because they think humans are special. Humans are special, of course - there's a lot that's unique about us. But we're not special in a way that makes us unamenable to scientific study.

Second, I think a lot of social scientists got interested in social science because they were first interested in political or social philosophy, and social science seemed more practical than philosophy. But they've still got a philosopher's mindset. I think you see this hang-up about science less among psychologists precisely because fewer of them come at the field from an interest in more philosophical issues.

Philosophy's fine. That was the entry point for natural scientists too at one time, and that's how social science started. But let's not let it get in the way of doing good science.


  1. I don't dispute that economics is a science.

    Why is astrology unscientific? One might say it's because the predictions of astrology--besides those shared with astronomy and psychology--are, entirely or nearly, unfalsifiable. That is, its predictions are sufficiently vague that they're compatible with almost any set of observable facts. The explanatory framework of astrology is capable of "explaining" pretty much anything at all and, therefore, explains nothing in particular.

    This is all quite true, in my view. The ease with which predictions can be "confirmed" and the arbitrariness of explanations are hallmarks of pseudoscience like astrology. However, astrology was not always so vague and wishy-washy. Once-upon-a-time, its predictions were more precise and its explanations less arbitrary.

    Science is not just a word to describe the logical structure of hypotheses but also a social enterprise. What made astrology unscientific was also the reaction of the community to recalcitrant evidence--astrology became a degenerative research program. The reaction was to indefinitely evade refutation through ad hoc amendments to theories; the empirical content of astrology was slowly strained out by successive generations defensive moves.

    The explanatory framework of astrology aided its fall. One could easily vary the particulars of its explanations without compromising coherence--something not true of good scientific theories. It had the seeds of a degenerative research program from the beginning. Add to that a irrational community, less concerned with with good explanation, truth, and precision than with defending the orthodoxy come what may, and astrology ultimately suffered a worse fate than being falsified--it became pseudoscience.

    The problem with astrology is not just the logical structure of its explanations and predictions, but rather the absence of real scientific institutions and traditions of rational criticism. Indeed, the former was in large part a consequence of the latter.

    While it's clear to me that economics is not astrology, it also seems fair to say that it does not exemplify the scientific ideal. Even basic theories in economics are often prohibitively difficult to test empirically, and many of its explanatory theories are, perhaps, a little too good at explain too much. Moreover, the corrupting influence of politics tears at the seams of rational traditions and scientific institutions. There are elements within economics that resemble degenerative research programs (modern Austrian economics?)

    Perhaps some economists need to accept the fact that economics is a science, but they'd also be right to suggest that economics should not rest on its laurels. The critical institutions of a science require continual maintenance if it is to avoid becoming another astrology; people don't take to such institutions and traditions naturally; economics, in my view, too often teeters close to the brink.

  2. 1. Here is Richard Feynman explaining why astrology is not science.

    2. Daniel, your "economics" is not science as explained by Feynman for you let people like Hayek (the original sinner when it comes to vagueness) in under the tent, when their vagueness is deliberate.

    3. Given your fascination with libertarians, etc., how do you have room to write, "Philosophy's fine. That was the entry point for natural scientists too at one time, and that's how social science started. But let's not let it get in the way of doing good science?"

    4. Feynman also talks about when scientists can use estimates.

    5. You write, "Astronomers don't go into soul searching over this sort of thing - why do economists? Why do we have people out there making such a big deal over uncertainty and error?"

    The reason is that Astronomers are in agreement there are stars.

    By contrast, when you have economists assume that markets are rational or that there is an efficient market, they are making stuff up which doesn't exist.

    Even when experiments and experience show them to be wrong, you, for example, continue to speak of such people as economists.

    Similarly, you have economists like Barro who make up Ricardian Equivalence, but you continue to recognize him as an economist.

    How would Astronomers react if someone, today, said, "There are no stars?" That is where Barro and Cochrane, to take two, are today.

    Similarly, as per Feynman, Astromers today do not truck with deliberate vagueness. See about 5:45.

    Today, however, the greatest skill of an economist is deliberate vagueness, whether it be Bernanke, Greenspan, or Taylor and Cochrane on the campaign trail talking about too much regulation but never, ever saying which parts of the Code of Federal Regulations is most in need of repeal or replacement.

    I will prove that you have such a disease.

    List the five laws that Congress ought to pass into law by July 1, 2012 to start creating jobs, good high paying jobs with full benefits?

    I can list five but I doubt you will list one.

  3. @Daniel:

    I don't think the question of whether economics is or isn't a science is interesting at all. As you said, science is a process. Some economics is unquestionably scientific and some isn't. I'm mostly a micro guy and I definitely have seen some theories which appeared entirely unfalsifiable to me. (Or some work that involved nothing more than playing around with math or computers without providing explanatory power) Those are cases where the work being done isn't scientific. But I've also seen lots of stuff which is good science.

    In general, I think economic researchers are doing science while economic policy makers are not. That's for a simple reason: economic researchers want the truth while economic policy makers want to make policy. Should we do a stimulus tomorrow? Only one of these two groups can respond: give me 5 years and I'll answer your question rigorously. The other has to make a decision right now whether they have sound science to back their decision or not.

    1. I definitely agree with your second paragraph. Policy makers are social engineers, and aren't scientists in the way that other engineers aren't scientists (although they ought to have good science to work from).

      On the first paragraph - I just find it interesting that so many people don't think it's a science or that so many people draw this sharp line between "hard" and "soft" science.

    2. I think the problem might partially be terminology. Economists are always presented as economists. So when an economist goes on TV to talk about policy options without hard science behind their proposals, that's what the public sees. And when their policy proposal (or prediction) turns out to be wrong, it gets blamed on economics not being a science rather than being blamed on policy makers/analysts making mistakes while operating under uncertainty.

  4. "But let's not let [philosophy] get in the way of doing good science."


  5. We're just highly evolved apes.

    If I were a scientist, I would figure out a way to get you to stop using this phrase in order to make a point that utterly depends on you using it the way you deny to me and Gene that you are using it. (Sorry for the long italicization.)

    1. The reason you are not a scientist is that you choose to deal in generalities that are deliberately stated so that they cannot be tested. Feynman's last lecture at Cornell, thanks to Microsoft for putting the lectures on its website as Project Tuva, makes the case plain that people ought to disregard what you say and write, for that reason.

      In about 40 words, Feynman lays open the entire racket, with his discussion of A being either loved too much or not enough by his mother.

      Until economics finds a way to exclude such opportunism, it should be reclassified as advanced fiction writing.

      A prime error, for example, is to pay attention to people like Hayek and Friedman, people who were skilled at saying nothing, while sounding like they are saying a lot. Modern adherents are Taylor and Cochrane, who rant against "regulation," but cannot find a section in the Code of Federal Regulations in actual need of repeal or replacement.


All anonymous comments will be deleted. Consistent pseudonyms are fine.