November 2014

Sun Mon Tue Wed Thu Fri Sat
            1
2 3 4 5 6 7 8
9 10 11 12 13 14 15
16 17 18 19 20 21 22
23 24 25 26 27 28 29
30            
Blog powered by Typepad

« One Way of Looking at our Spending Binge | Main | Is the Fat Lady Singing for the US Economy? »

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d83451eb0069e20133ed89dc3d970b

Listed below are links to weblogs that reference Taking the Con Out of Economics: Symposium in JEP:

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

I must say that in my view it is because we aren't "looking out the window" as you have suggested we need to do, before. "Rational choice" theories are not enough - we need to learn from our mistakes -- not through econometrics, but through looking at what we chose to do (policies) and what resulted.

This is why looking at history I think is so important. History can be a period such as 1930-1960 or it can be the policy we put in place 3 years ago - it doesn't matter - but when we look at history, we need to look at what behavioral effects resulted from policy change, what institutional changes we made and what we saw occur as a result. It's common sense, but it works better than aggregating numbers and running regressions on them. Anyway, that is my opinion.

Pete,

The old argument never die. Anyway, IMHO we really need to face up to a basic issue: For the foreseeable future "science" trumps "literature." Analytical narratives are fine don't look as "scientific" as fMRI studies. Now if you think (as I do) that economics is a science that must not eliminate "literary" methods, then you have to ask how to get such unfashionable methods back into the game. It may be hopeless, as David Colander would claim. But if it isn't, then we have to ask how to get there from here. I don't think it makes a lot sense to charge the citadel of science on a literary steed. Isn't it sort of obvious that we have to take the Hayekian path from "science" to "literature"? Start in science, chase down the scientific implications relentlessly, and end up unexpectedly with "verstehende psychology."

If you buy all that, then Austrian economics needs to go all in with complexity theory, but emphasizing problems of computability and data compression. Computable economics gives us a huge opportunity. I really don't understand why we are not grasping it with both hands.

This is no different than what Samuelson and Friedman believed:

" and if theory could produce any result imaginable it would be sophisticated econometrics that would adjudicate between competing hypotheses."

Samuelson later acknowledged that this program was a complete failure.

Roger, this "literature" picture is both false as an account of the causal / explanatory strategy oh Hayekian economic science -- and a dead loser simply from a PR standpoint.

A causal science like economics -- backed by a universal logic -- isn't mere "understanding" or interpretation on analogy with an English Lit professor or any anthropologist. Economic explanation is parallel to Darwinian spontaneous order explanation, not literary interpretation.

This sounds right to me.

"If you buy all that, then Austrian economics needs to go all in with complexity theory, but emphasizing problems of computability and data compression. Computable economics gives us a huge opportunity. I really don't understand why we are not grasping it with both hands."

And understanding the limits of this program will advance understanding and knowledge as well.

I don't much like articles (like the Sims piece in the symposium)that say "economics (or econometrics) is or is not such and so."

It is as if it were written on high (really high) what economists must do. Experiment/not experiment; follow axiomatic formulation, etc.

People have different comparative advantages. Sims, for example, has invested in econometrics. He doesn't want people to start doing narratives.

I wish there were some way to let a thousand flowers bloom -- all respectably.

But our field is dominated by people who elevate what they like to do into a Methodological Principle That All Intelligent Scientists Must Follow. Sorry, I don't buy it.

Explanatory problem / causal explanatory solution sets provide the exemplars from which "methodological" reminders are constructed. I got Larry White to see this pony many years ago on my old Hayek email list, put the lesson never seems to sink in.

It's really the lesson Kuhn taught about science -- and the one both Mises & Hayek grasp when they used Menger' origin of money problem and solution set as an exemplar of "Methodological Individualism". The phenomena -- the satisfying explanation -- comes before the methodological reminder / dictate.

Too much of economics start with the methodological dictate (usually derived from a false image of science combined with a false epistemology taken from the tradition) -- and then tries to cow everyone into bowing before the mighty God of opportunities to crank out endless professional products to advance the career .. All in the name of formally dazzling "science".

A thousand flowers blooming does't necessarily take younto a thousand causally valid explanatory strategies.

Greg's first comment suggests that I may have been cryptic. Sorry about that. In my defense, I was trying to be brief because I've said this sort of thing over and over again on this blog. As I said, it's an old argument.

I was not saying we should give up science and land ourselves in some strictly literary enterprise somehow tied to "understanding." I want to stick with science. But I want to bring "literary" methods, including "understanding," into science. Just as Hume was said to have used reason to whittle away the claims of reason, so too did Hayek use science to whittle away the claims of science (and "science" too BTW). We should stick with science, but the sort of logical limits Hayek explored in The Sensory Order show that we need "verstehende psychology" *in* our science. That's really quite radical and pretty far from what many mainstream economists would recognize as "science."

If Colander is right, then you gotta be mathy to seem scientific. (In so saying, he's making a sociological observation and not a methodological preachment.) So how do you get mathless "literary" methods recognized as "science"? You could attack from the outside and tell mainstream economists how totally wrong they are about science. Good luck with that. Or you could out-math the math-heads. That's my preference. That's the power of computable economics. Your "scientific" method implies that you or the agents in your model can compress data that cannot be compressed, or compute what cannot be computed, or easily solve problems that are "hard" in a technical mathematical sense. That gives you a way of arguing that is parallel to Mises way of arguing about policy. I'll take your method (rather than policy) and show that it does not and cannot produce the results you want it to.

Why aren't we jumping on this stuff?

Because some of us don't know enough maths:)

Part of the problem is the state of philosophy, as desribed by three authors including Barry Smith.
http://www.criticalrationalism.net/2010/04/12/whats-wrong-with-contemporary-philosophy/

Smith is very good on the kind of philosophy that underpins Austrian economics. He calls it "fallible apriorism".

BTW If you are going to go with the philosophy of science, go with the best, that is Popperian critical rationalism, or Popper read as a fallible apriorist, not as a falsificationist. When did Popper ever describe himself as a falsificationist?

But if your method is ok you don't need to bother with the philosphy of science, you just need to get on with economics. Like if you are not sick you don't need to see a doctor.

I read your linked post, Rafe. Is AP really so insular? What about Alvin Goldman and Daniel Dennett? Your post already mentions Searle. I am not a professional philosopher, but I hadn't had the impression of AP as such a desert. Or am I just drinking from the few oases?

Thanks for the reminder, Roger.

My suggestion is that the only hope of getting anyone on board is to use alternative language and recast this using alternative language.

The old language has all sorts of fatal baggage -- there _really was_ a bedrock traditional distinction marked out by the language you are using delineated between "understanding" and "science" science. This distinction was a mistake when expropriated to economics -- and lead to mistakes and confusions. Importing the misleading and fatally compromised language back into the conversation can only turn off ears or provide a barrier to understanding.

Roger asks:

"how do you get mathless "literary" methods recognized as "science"?"

Part of the answer is to get this cast correctly and in a non-contaminated way, dropping the short hand of "literary". We aren't doing rival cultural interpretations of Ulysses.

Roger wrote:

"I was not saying we should give up science and land ourselves in some strictly literary enterprise somehow tied to "understanding." I want to stick with science. But I want to bring "literary" methods, including "understanding," into science. Just as Hume was said to have used reason to whittle away the claims of reason, so too did Hayek use science to whittle away the claims of science (and "science" too BTW). We should stick with science, but the sort of logical limits Hayek explored in The Sensory Order show that we need "verstehende psychology" *in* our science. That's really quite radical and pretty far from what many mainstream economists would recognize as "science."

Greg,

I don't really understand what this "bedrock traditional distinction" is. Is it epistemological or ontological, or what? I think Hayek and Mises both clearly point to Dilthey and Bergson and we in Austrian economics had better understand that fact if we are to know what our own tradition is all about. What's the "mistake" in distinguishing "understanding" from the methods of physics? The behavior of a person has meaning; the behavior of a rock does not. Isn't this point powerfully reinforced by Hayek's 1920 MS, TSO, and is theory of complex phenomena in general?

Awesome. But this is simply part of the case as I've explained it of the explanatory realities of the phenomena and problem at hand that put a laser beam on the difference between good science and bad ...

Roger writes:

"Or you could out-math the math-heads. That's my preference. That's the power of computable economics. Your "scientific" method implies that you or the agents in your model can compress data that cannot be compressed, or compute what cannot be computed, or easily solve problems that are "hard" in a technical mathematical sense."

Roger -- there was a classic distinction between causal explanation (e.g. Mill's model) and "understanding.

Well, Hayek offers an empirical problem and a causal explanatory solution -- entrepreneurial learning in the context of changing local knowledge and relative prices. This ain't "understanding" of the sort found in Windelband or Dilthey.

Roger, in Darwianian biology you have an "as if designed" patterns of order and "as if designed" emergent new communities of order. Darwin offered an open-ended causal mechanism to explain this -- one that could not be limned by the simplest linear relations of physics or chemistry.

Hayek does exactly the same thing, as I pointed out in a paper written for the SAAE many years ago.

Selection across variation in changing environments is an open-ended and never forecastable in advance causal process (as I explain in my "Insuperable Limits to Reduction in Biology" paper.)

In economics the open-ended and never limnable in forecastable "object" physics or chemistry relations is human entrepreneurial learning in a changing environment, aided by the "subjective" logic of marginal valuation.

The explanatory element here is _learning_ and this is a causal component.

The "understanding" / "cause and effect" science distinction of the philosophical / "social science" tradition DOES NOT APPLY HERE. That is one of the core mistakes which has been a barrier to progress.

"Meanings" as cashed out by most modeling programs are fixed Platonic like things. They mislead and confuse things when we attempt to use this construct (as typically modeled) in economics -- we get stuck focusing on explaining "behavior" when what we want to do is explain "as if designed" global economic coordination, not via identifying "meanings" or explaining individual behavior, but through identifying the causal role of entrepreneurial learning in the context of changing conditions (aided by the logic of choice).

On this account, Robbins & Mises and all who came later took a dead end when they pretended that "economics" was the science of economic behavior or "action".

Roger writes:

"What's the "mistake" in distinguishing "understanding" from the methods of physics? The behavior of a person has meaning; the behavior of a rock does not.

Shorter version. People in the market learn and rocks don't. Learning as a causal process can't be limned by simple causal "laws" of the kind we find in simplest parts of physics and chemistry (neither can the causal process of speciation and adaptation in Darwinian biology.)

The mistake is not to distinguish between learning and a false Millian / Popperian / Machian / etc. image of the very simplest aspects of physics and chemistry.

Roger writes:

"What's the "mistake" in distinguishing "understanding" from the methods of physics? The behavior of a person has meaning; the behavior of a rock does not."

Econometricians and macroeconomists DO pretend that the product of the causal process of learning in the market CAN be limned as if they exhibited outcomes of the sort we see in the fixed and determined causal "laws" of the very simplest parts of physics and chemistry.

The "meaning" entity / conception produced more problems than it solves -- Wittgenstein and Hayek (in his most 1940s work) offer an alternative in shared ways of "going on together" or patterns of common behavior (as exhibiting shared understanding and common structures of mind).

That's all you need to "gain access" in a "scientifically legitimate" way to the fact that humans are causal explanatory entities as entrepreneurial learners -- and you avoid all of the pathologies of the philosophy of language and mind that are so well known.

Roger writes:

"The behavior of a person has meaning; the behavior of a rock does not. Isn't this point powerfully reinforced by Hayek's 1920 MS, TSO, and is theory of complex phenomena in general?"

Sorry. Make that "Hayek (in his post 1940s work)"

Gheez, Greg, one comment at a time! Anyway, it is true that some of the old stuff contrasted "understanding" with "cause." I talk about that a bit in a paper in Advances in Austrian Economics from the first Wirth conference. I really don't think that means we should mortify our intellectual roots. Besides, all these sorts of issues arise in the context of papers in methodology and the history of thought, where you precisely must enter into such matters and discuss word histories and so on. If I am doing a paper on error in forensic science, however, I'm not normally going to use words such as "hermeneutics." So I'm aware of the "danger of words" and all, but I'm not sure we get that far with general discussions of what words to use how.

"But I have to admit that this methodological revolution has been slower in coming around than I had hoped. Why do you think that is?"

I'm going to duck the very educational discussion between Prof Koppl and Mr Ransom, which seems to be focused on "how can we get the methodological revolution to go faster?" [Gentlemen: do you think that's a fair characterization? I just wanted to clarify.] I'd like to take a shot at why more people have not come around to analytical narrative.

[Oh yeah--I'm not trying to argue why they *shouldn't* come around to it. Indeed, of the reasons below, I think the most likely explanation is actually a terrible reason not to use a technique. And the list is not meant to be exhaustive--I don't think it's even exhaustive of what I personally think...]

A couple possibilities:

1) Other scholars--those using general, mainstream quantitative methods--have engaged seriously with arguments structured as analytical narratives, and found them wanting.

-This, I seriously doubt. A casual skimming of the latest AER and JPE do not seem to have any such arguments in them, and while they do from time to time, it doesn't seem particularly often. It does not seem to me that people doing the latest and most complex econometrics engage seriously against the argument "why not some descriptive statistics and a well-reasoned qualitative account of the phenomenon?" I think they just accept that they must do the fancy 'metrics without question.

2) These methods do not allow for ready theory falsification.

-I used to think this strongly; now I'm less convinced. Certainly, if your theoretical claim is "the price elasticity of demand for X is less than .5", then a narrative wouldn't be very helpful, but how often is that the Really Important Question? Not very.

3) Analytical narrative relies on case study or example; anyone can pick a case that matches their argument.

-I think that people might believe this--I certainly believe that people *can* do this--but that's no more an argument against narratives than "you can trim your sample/add variables not in your model/weight observations so your regression works". It's an argument against doing a poorly researched narrative, seems to me. Maybe many people hold this opinion, and have not really engaged seriously with "what would a narrative need--in terms of evidence--to convince me?"

4) Narratives do not provide estimates of magnitudes that people want to know.

-I actually think this is it, by and large. Policy debaters in forensics programs are taught that you win arguments by showing that your position saves money & lives and the other one doesn't. In the same way, social scientists are only useful from the perspective of many outsiders (gov't, private industry, think tanks) if we can give them nice, precise estimates of what's going on in the world.

Note: I'm not arguing that this is a good reason to do quantitative work, or even that the estimates are, in fact, nice and precise. But I think it's part of the issue.

5) There are elaborate procedures to judge when a quantitative estimate is "true" or "large"; there is no such procedure for narrative.

-I think this is the other part of it; however positively silly specific procedures to determine whether an estimate is biased or inconsistent, and whether it is statistically significant, and whether it is causal or correlated, such procedures exist. Like the wedding I went to over the weekend, there's a ritual you do for bias, and one for consistency, and so on. It doesn't matter that, in a very real sense, the rules are arbitrary, just that there are rules. Somebody else can examine what you did and see if you fulfilled all the rules.

It seems to me that there are not--cannot be?--such rules for a narrative. And even though we can lay down serious critiques of those rules, their mere existence gives us a process for verification of a result that is more defined than "I find this logical, well-written, and convincing." That doesn't strike me as a particularly defensible reason to eschew narrative, but I think it (and a bit of 4) are why the revolution, as they say, has yet to be televised.

Does anybody else see any good reasons *not* to use narrative? Ones that people who aren't jazzed about narrative might actually defend when pushed? If I step back from my pro-empirical position to ask "why not narrative" (as opposed to "why empirics"), I am hard-pressed for good reasons to avoid narrative. True reasons? Maybe. But good? Not so much.

The focus should be on sound causal explanation / problem - solution exemplars. Focus on "methodology" turns people away from explanatory problem / solution exemplars and towards a pathological philosophical tradition in which most economists have a favored pathology / false view of "science" which they don't want to think about or give up.

We want good science, but "methodology" makes the kiddies eyes glaze over -- and as an easy out they glom on to the most fashionable false view of "science" ready at hand that best promises to produce career enhancing research and publications.

Jared wrotes:

"I'm going to duck the very educational discussion between Prof Koppl and Mr Ransom, which seems to be focused on "how can we get the methodological revolution to go faster?" [Gentlemen: do you think that's a fair characterization? I just wanted to clarify.] "

"But I have to admit that this methodological revolution has been slower in coming around than I had hoped. Why do you think that is?"

Because it doesn't take 10,000 research "scientists" doing endless research to get this right, but it does take extremely hard work and difficult thinking to get the job completed -- and there is essentially no incentive structure (top to bottom) for anyone to do it (Bruce Caldwell and Peter Boettke are working on the problem, and it is not so simple -- although the promised solution as Caldwell and Boetkee characterize it is basic econ explanation.)

Grad students can't risk making their way getting this worked out -- and the academy gets filled with people who don't care and have not competence of background for it.

We find mostly only the oldersters turning to aspects of the topic -- e.g. Douglass North or Vernon Smith or even Paul Samuelson.

What economics needs is its Ernst Mayr.

There is good reason to think that economics has the explanatory / empirical program that is has because of the professional incentives behind it go -- what gives you something easy to teach and grade, what gives you something unrisky and easily tractable to crank out a dissertation on, what gives you something tractable and reliably publishable to do research on, what has a demand within the Federal Reserve and the government bureaucracies, etc.

Incentives matter and there is no incentive for a grad student or top 5 grad program professor to be make himself the Ernst Mayr of economics.

Jared,

I don't know that I want a "revolution." I just think we should self-consciously *add* ("classical," pre-Heidegger) hermeneutics to the scientific toolbox of economists. Plus, it's not so much a question of how hasty the change as how to get it at all. Like it or not, "science" trumps "literature" in fact. So you gotta somehow show why *science* demands the use of methods currently dismissed as "literature" (or whatever). I just don't see how that could realistically happen if we don't place more emphasis on Hayek's diagonal argument at the end of TSO.

Roger, yes I think you may have struck oases in the wilderness. Alvin Goldman may be doing good work outside philosphy where he needs to meet scientific standards but I don't know where he hopes to go with his pre-Popperian epistemology of justified belief. I have encountered people who were excited by Searle and Dennert but it seemed that they were excited by ideas that could be found in principle in Popper and Hayek a long time ago.

Barry Smith went through Oxford and is inside the AP tent so his critique has more cred than my outsiders view.

Something has gone badly wrong with scholarship in philosophy, as demonstrated by Dan Hausman writing on Popper. In one of his papers, possibly in the collection on Popper's contribution to methods in economics, he announced at the start that he was not going to address Popper's ideas as presented by Larry Boland (a trained maths economist who also read Popper) but he would stick to the views that are usually attributed to Popper in the literature. In other words, why bother to grapple with the real thing when you can savage a strawperson. I have discovered in a survey of some dozens of philosophy texts that the "falsificationist" views attributed to Popper cannot be found in the books with Popper's name on them.

In short, Hausman (and dozens of other no doubt highly trained and scholarly philosophers) has written a great deal in criticism of a position that does not exist. A striking example is located in his Introduction to the anthology "The Philosophy of Economics" (1984, 1994, 2008). In each edition he attributed to Popper the view that a theory or hypothesis that has been apparently been refuted should be thrown out forthwith, without making excuses. The citations that purport to support that claim do not do so, and there are clear statements by Popper that (a) no refutation can be decisive and (b) even defective theories can be kept in play in case they can be improved or if no better theory is available (pre Einstein there were known to be problems with Newton's theory but it was still the best around). Decisions on these matters need to be made in context and cannot be decided a priori.

This is a detailed appraisal of some passages of Hausman on Popper.
http://www.criticalrationalism.net/2010/05/02/problems-with-popper-scholarship-dan-hausman/

The reason for my concern with philosphy is that people who want to do economics better may turn to philosphy and the literature on scientific method for help but they will need to be incredibly fortunate in the books they find, or they will waste a lot of time and be no better off.

Rafe -- philosophers are like economists.

They want simple formal constructs that crank out publications & make for easily graded class content.

And they don't want to mess with alternative understandings that require years of soaking to really master.

So they have simple "conversation enders" for all considerations they don't fit their formal crank it out machine. It's not that they don't know the outline or even many details of the alternative, they just don't want to think about it.


This is just a fact of academic life -- people see the consideration of complex alternatives as a waste of their professional time -- time best served producing publications and advancing their career and working on their narrow research program.

The professional institutional system is driving this -- as is self-interest. And I don't see it changing.

On the up side, what if the Austrians make the same progress over the next few decades that they made since 1974? The downside is that the levels of debt and the bipartisan push for big government worldwide may have reached a tipping point where it will be really hard for the entrepreneurs to win the race against the rent seekers and the regulators.

Another positive thought, Austrians from Mises down have wasted a lot of ink arguing that the methods of economics have to be different from the methods of the natural sciences. The argument is based on the false assumption that the methods, the epistemology and the metaphysics of positivism are valid for physics. But that is not the case.

If this is true and all sciences can use the same hypothetico-deductive logic of investigation and testing then Hoppe et al can stop moonlighting as philosophers and stick to their day jobs in economics.

Rafe, Darwinian biology doesn't fit the H-D model, and Popper conceded that Hayek was right that complex sciences like Darwinian biology and economics were in fact science, with an "explanation of the principle" structure not subject to direct falsification.

I like Popper's take down of positivism but his rejection of the Carnap / Mill program was inadequate and incomplete.

Hayek and Kuhn and Hanson arguably get us closer to the truth about the nature of the structure of scientific explanation, than the fatally flawed and inadequate H-D model (even it's advocates can fix it's well known problems).

I'm a big fan of Popper and Bartley on justification, but not so
much on the Carnapian picture of "theory" and meaning at the heart of Popper's H-D model.

Make that:

"even it's advocates can't fix it's well known problems)."

Greg, what do you mean by Popper's H-D model? The hypothetico-deductive model is much older than Popper. It just means deducing consequences from hypotheses for the purpose of testing or for practical and technological applications (which can also be regarded as tests).

It is not the covering law (nomological-deductive or Popper/Hempel) model of explanation which Popper gave away decades ago in favour of explanation by propensities which links his theory of probability with his cosmology - "a world of propensities".

How did Carnap get into this? This is supposed to be a serious discusion:)

Rafe, sorry, I meant only to point to Poppers nomological-deductive model of explanation (and it's background of assumptions).

I never understood (I guess) how the propensities picture does away
with the N-D explanatory machinery in the hypothesis / explanation / falsification process assumed by Popper's demarcation criterion.

And what of Popper's concession to Hayek on Darwin and the sciences of complex phenomena?

How does the propensities picture helpmus with the Darwinian causal explanatin mechanism or the HayekIan causal explanatory mechanism in economics?

It's been some time since I've read Popper's late work on biology.

Pete,

The Boettke Diagram with the formalist/appreciative axis and universal/historical axis, did that ever get published anywhere? I looked on your website but I can't seem to find it. Thanks for any help!


geoffrey

The comments to this entry are closed.