I don’t have time right now to read the essay which is abstracted below. But I’d love to. And I don’t really have time to defend the propositions that I’ll put before you here, nor to get them into a state that I would be confident I wouldn’t have to revise once I’d posted them. But with those caveats, here goes.
Positivism has had a profound influence on the development of what I call the practical social sciences – namely law and economics. In American jurisprudence anti-positivism held sway for most of twentieth century with legal realism being the dominant tradition – something that was an offshoot of American pragmatism – which is the most accessible, useful and democratic of the anti-positivist philosophical schools.
In economics meanwhile, pragmatism was represented by the Institutionalists in economics. Institutionalism was a quite philosophically astute school, but wasn’t much use to anyone because, like a lot of anti-positivist economics today, it spent a lot of its time in ralatively barren spats with positivism rather than getting on with its work. While it won the methodological battles in the sense that it laid bare some of the intellectual sleights of hand on which neoclassical econmics was built it lost the war on account of its own inability to build a useful body of doctrine, practice and technique (someone like Marshall * – or later on Hicks – didn’t really deny they were slights of hand, they just thought – reasonably enough – that the intellectual shortcuts they were taking were worth a try – which they were providing one kept that in mind. Alas our mind only holds so many caveats, and meanwhile there are mountains to climb). Anyway, unlike Marshall and Hicks, institutionalism in economics didn’t get very far in its first incarnation.
Meanwhile positivism kept plugging away reducing economics to the axioms of choice at the margin and gradually throwing away the crucial bridges that were present until around the 1950s between the mathematics and the passages that justified the use of the mathematics and argued that the methodological sacrifices made in making increasingly Herculean abstractions were worth making.
We ended up with a very disturbed discipline in which it was seriously argued that the realism of assumptions could be dispensed with (Friedman argued this largely in defence of the assumption of perfect competition even where it wasn’t quite realistic – a position I’d support – but Friedman ended up unimpressed by more full blooded – and basically nutty – applications of the principle he’d advocated as for instance in the theory of real business cycles (which shows how the Great Depression was really a spontaneous decision by workers to take a holiday). People like Krugman and Delong have been bemoaning the nuttyness and forgetfulness of modern economics of late, though their explanation of the source of the problem is ideology rather than methodology. (I’m not saying it’s all methodology by the way, but I’m dead sure it plays its part).
Anyway, the second wave of institutionalism borrows heavily from neoclassical method and is a kind of positivism, though it is much closer to the ‘soft’ empirically focussed positivism of Hicks and Marshall. Today the strongest cell of anti-positivist thinking is Austrian economics.
Anyway, the article to which I’ve linked above by Becchio Giandomenica (University of Turin) has the following title:
A historical reconstruction of the connections between the Viennese neopositivists and the American pragmatists: economic theory in the project for the International Encyclopaedia of Unified Science.
And here’s the abstract.
Persuaded by the fact that the philosophical debate in Vienna during the 1930s deeply influenced the subsequent American developments of economic theory, my purpose is to reconstruct, from a historical and theoretical point of view, how Austrian interests in economics (which spread in the interwar period) moved to the USA, and what the subsequent developments were. From a historical point of view, I shall focus on the genesis of the International Encyclopaedia of Unified Science promoted by philosophers belonging to the Vienna Circle (Otto Neurath, Rudolph Carnap and Philipp Frank) and American pragmatists (John Dewey and, above all, Charles Morris) between 1938 and 1969. From a theoretical point of view, I shall focus on the development of economic theory within this project, which was regarded as the most important outcome of logical positivism or new empiricism, this being the common philosophical outlook that they shared. The aim of this paper is to show how the original meaning of economics (as formulated by Neurath) within the Vienna Circle was changed in its transition from Vienna to the USA (when the International Encyclopaedia was organized) and was finally radically transformed in the American context during the following decades.
I expect this post will mean very little to lots of people. And the intriguing thing is that sometimes the anti-positivists point predominantly left – as in the case of legal realism and institutionalism – and sometimes right – as in the case of Austrian economics. My interest is mainly epistemological. I think positivism is a kind of epistemological baby talk. But it needn’t matter a great deal. It is simply not the case that you’ve got to get your foundations right to do good economics (Well foundations are a positivist preoccupation, so I guess that’s a victory for anti-positivism right there). But just as great scientists often have a philosophically naive understanding of their own methodology (what matters is how well they do the science, not how well they can theorise their method invariably after the event), Paul Krugman is a damn good economist, even if when he writes about methodology he writes sentences which make me cringe – for instance implying that formalisation is somehow inherently superior to informal (verbal) reasoning. I’d say it always depends, there are costs and benefits of both modes and it depends on their balance in any specific situation. Indeed every useful application of formal reasoning relies to some crucial extent on informal reasoning, both to motivate it and to indicate its relevance to the empirical world.
Anyway, it’s probably taken the time that it would have taken me to read the article to write this post. But now it’s here and a record of these loose thoughts I’ve had for a long time. I’ll leave off here and leave readers to make of this post what they will. Rafe will be interested I suspect, he may even agree with me. This may have some Troppodillians worried. Some will worry about me. Others Rafe.
* Intriguingly, Marshall started out life as a devoted Hegelian – and at least from my limited knowledge, Hegel is the first and archetypal anti-positivist philosopher. Since Hegel there has been positivism and post Hegelian philosophies (though some who have fought the fight against the cruder forms of positivism – and this includes both Hayek and Popper – would describe themselves as post-Kantian. They regarded Hegel as a menance and/or a crank).
“Paul Krugman is a damn good economist, even if when he writes about methodology he writes sentences which make me cringe – for instance implying that formalisation is somehow inherently superior to informal (verbal) reasoning.”
.
Being in an even sloppier area of the social sciences, I’m surprised about this statement. It’s hard for me to think of problems where verbal reasoning is better than formalization when the latter has been done properly (although I can think of lots of problems where formalization is simply too hard to do). All this verbal reasoning has the habit of making me cringe, and is one of the reasons we end up with so many untestable theories. At least in my area, its common, for example, for verbal reasoning debates to go for decades, but then someone will finally bring out a decent model of the underlying processes which then shows a lot of the assumptions that were made via verbal reasoning were simply wrong.
Sorry Conrad,
But your argument is very . . . well verbal. Could you please express it more formally ;)
This issue quite similar to the last essay I wrote in university, through the issue of the spat between New Economic Geography and sections of the geographic discipline. It was frustrating to see a debate which was polarised between anti-fornalist geographers and NGEers (especially Krugman) whom were almost totemistic in their (expressed) attitudes to formalism. The middle ground, which was the most fertile was being emptied by waves of reaction and counter reaction.
I have a feeling that many good economists have been taught in heterodox fashion but falled into neo-classicism through pragmatism. They have been extensively lectured in the limitations and falsehoods of neo-classical toolbox, but when they wanted to do some other research there was nothing else to use. Much of the heterodox world is anti-positivist and disparage the use of formal tools (like Austrianism) and other parts, such as Evolutionary Economics and Post-Keynesianism have an (as yet) incomplete tool box, despite valiant efforts.
So they use neoclassical tools, but understand that they are just tools to make theories to understand reality. They are not the theory themselves.
My own attitude has generally been that formalism is vital as a logic check, and for creating the empirically testable hypothesis which will provide the real kicker. But if an idea doesn’t make sense verbally, it never will, no matter how elegant its maths (I’m looking at you RET!).
Disclaimer: I failed at the mathematics of formal model building.
Yes, I agree Richard, it’s a matter of being pragmatic and making the best use of the tools you can. Not much point in being pro formalism or against it. It’s a case of trying to keep an eye on where it can help more than hinder, and where the converse holds and acting accordingly.
No one reasonable could really object to the point that for practical purposes it is important not to be so stupid as to truly believe ones’ rough abstractions. I am thus in full agreement with the spirit of the above piece. Unfortunately, it is not just a lack of memory capacity that makes us fall into this trap, it is also due to selfish career reasons. If X can find a stick to beat Y out of a market then X wont let that chance go by just because Y happens to be vaguely right. The easiest way to belittle the competition is to point out that they cant do the mathematical models, which, if they truly cant, they will find impossible to deny.
These career motivations are becoming ever more important with more economists and more rewards goin to economists. Hence formalism is safe and sound because the X’s will need it more than ever to beat off the hordes of Y’s. Put bluntly, Nick and Richard, academic economists use formalism to keep the likes of you out of their domains. The problem with being vaguely right is that others who are less erudite as you cannot distinguish you from a charlatan.
There is another reason for formalism. It is that it appeals to the very brightest of the young students. The very brightest invariably do not want to hear you only believe in something for 90%. They want to hear what you believe absolutely and show you how clever they are by following that train of though to an interesting new conclusion via maths. It is a means to distinguish themselves and hence a means for them to beat off the other competition amongst the students.
I resent the implication I am anything but a charlatan!
In terms of using formalism as an exclusionary language, I think there is something there. Just like certain segments of indie music lionise certain bands to distinguish themselves from the teenybopper masses whom like such trivial things as melody and rythm. Just like the comic book fan that insists that a Spider-Man comic need take into account all of his clones and the fact the current Carnage symbiote is from another dimension to distinguish himself from those who never picked up a book until Sam Raimi made a film.
Just like people who read James Joyce and claim to appreciate it!
These elites are exclusionary, but to what end? They wield no power, but stew in resentment because society at large does not recognise their status.
Likewise, formalism totemism doesn’t hold that much sway in policyland. I don’t think politicians appreciate being told “this might not make sense when I explain it, but it has MATHS!”, and they certainly can’t sell that in a soundbite.
I’m sure they’re not having as much effect as Nicholas, or Krugman who for all his own formal totemism has a comparative advantage (exploited) in explaining macroeconomic concepts in purely verbal terms.
This exclusion reduces the supply of good research, but it relegates its adherents to an intellectual ghetto (albeit well paid ones).
Suppose I was trying to teach a macroeconomist how to play snooker. So I start to explain the colours of the balls, and the various positions and angles on the table and how to work the pockets.
The macroeconomist then decides that too many variables just makes it confusing and comes up with an improved analysis using only one variable which is the average position of all the balls on the table. Reducing the entire game down to an average greatly reduces the complexity of the problem and allows the economist to produce charts of how this average moves over time, and correlation of the average position against each player’s score. Unfortunately, the macroeconomist never actually manages to learn to play snooker.
Being a determined fellow, the economist studies a great number of famous historic snooker games and discovers the “Theory of the Great Attractor” which states that the average ball position will generally be attracted towards the middle of the table. Once extensively formalized, this theory is demonstrated to hold for over 90% of all games of snooker!
But actually, it serves no useful purpose in helping you learn to play the game.
“Likewise, formalism totemism doesnt hold that much sway in policyland. I dont think politicians appreciate being told this might not make sense when I explain it, but it has MATHS!, and they certainly cant sell that in a soundbite.”
.
I might point out here that this is half the problem with the global warming debate, but it doesn’t make the nice sets of equations the models are based on bad, and nor does it make the descriptions of how they work better predictors.
John Kay’s current column is on point.
Hi Richard,
yes, exclusionism is an important aspect. It is, unfortunately, almost impossible to measure the ability of a student to ‘muddle through’ successfully. Hence if one would base promotions and exam results of students on the kind of skills Nick and yourself display, one would quickly find ones judgments questioned in court. Like it or loathe it, basic decisions on students have to be based on something that is easily observable and hard to argue with. Memory and mathematical ability are far better at this role than whether someone is broad minded or has a good overall grasp of the relative importance of diverse factors.
Hence the reality on the ground is that formalism is now more important than ever. Students who cant do statistics and maths have very little chance to get a PhD from a good place or a good supervisor. Students who get through the door and fail their technical courses are out. This tendence is clearly spreading through psychology, sociology, management, and all the other social subjects. There is really not much point complaining about this reality because the need to sift through masses of wannabees simply forces our profession in this direction and outweighs the rather vague and not-so-lucrative need to be of overall relevance. Besides, I think Conrad is basically right in his implicit assertion that many of those who cant do formal models usually waffle on at great length without ever discovering how inconsistent and impossible many of their pronouncements are. Once you can do models you start to realise how hard it actually is to think consistently and how easy it is to fall in love with a nice sounding argument.
Whilst I also think Richard and Nick are right in their assertion that formalism does not hold much sway amongst policy makers, I do think they are more influenced by the results of formalism than you might think. The advisers nearly always need technical reports in the background to substantiate their statements, and the constant analysis of data and models is, if you like, the usual training ground for the higher echelons of economic advisors. Most RBA and Treasury people have economic PhDs and quite a few have a mathematical background, if only because of the mental discipline it gave them.
I think the future is not in abandoning formalism. That has no chance. I think the future is more in more explicit policy oriented sub-disciplines within economics, i.e. the science of ‘best guess economics’. The problem will be that none of the other sub-disciplines in economics is going to stop pretending that they already are policy relevant. You’d be an idiot to give up that pretence.
It’s a pity this thread has got caught up in a silly debate, namely formalism versus informal reasoning. That’s a silly debate. The only question of interest is the trade-offs in choosing between them in specific situations.
“The only question of interest is the trade-offs in choosing between them in specific situations.”
.
I’m willing to go out on a limb and say that at least for issues which are not going to be used for policy and the like (i.e., the development of theory versus the practical application), formal models are generally better than informal ones, even if they easily falsifiable (or at least theories where part of the theory is accompanied by a formal model). Here are some reasons why:
1) They are explicit, and therefore the scope of what they can and cannot predict is easier to determine than most verbal theories. We know what is there and what is missing. We also know data they are applicable for and what is out of their scope. (caveat: “hopefully!”).
2) Because they are explicit, one can try and falsify aspects of them more easily.
3) Because they are explicit, it is generally far simpler for people to run and think of experiments to test them than if they are not explicit
4) At least for some models, one can map the entire state space that different parameter values can predict. No-one can do this in their heads in many areas, even with mildly non-linear data (except Stephen Hawking).
5) Falsifying different aspects of models often causes people to try and think of new ways to try and tackle various problems.
6) Knowing what is out of the scope of a model allows people to work on things that are not yet well formalized or even investigated.
7) Knowing what is out of the scope of model is often difficult to determine with verbal theories. At least for me, it’s amazing how many things you realize you have not considered when you actually try and specify a problem down to the last nitty-gritty level.
8) There are some areas where it is simply too hard to predict the results of how multiple levels of representations/feedback etc. happen to work. Alternatively, it might be simple to collect the data to see if what the model predicts happens to be close to the real thing. Using verbal theories may not allow such specific predictions.
9) There is just less arguing with bullshitters (and just the ignorant) who become too attached to their theories and think that no aspect of it could possibly be wrong (these people are common).
So basically, I think that formal models are good for theory development, even if they aren’t very good models — I think they force people to think of better ones, be explicit, and actually try and use reasonable scientific reasoning (such as that found in Conjectures and Refutations) when arguing. Perhaps the latter of these isn’t such a problem in many areas, but it certainly is in the social sciences. I doubt some people even think of the idea of falsification and the like (I know most of our PhD students don’t). Of course, one could argue that at least some of the above should apply to really good verbal theories, but in my experience, forcing people to use formal theories is generally far more efficient, clear, and therefore allows faster scientific progress.
Conrad,
I agree with all your 9 points, particularly point 9. It truly is amazing how quickly a simple model can stop a futile argument that otherwise can rage for ages.
Whilst I think the main debate in science is about formalism versus non-formalism in general (which the non-formalists are resoundly losing all over the place), I do think non-formalism has some specific advantages that give it a place:
– most people think in associations, pictures, and storylines. Very few people think in terms of mathematical abstractions, and you will thus find that communication is more efficient and pursuasive if you use stories (even if they are inconsistent). This goes two ways: a) social scientists themselves usually begin with storylines and pictures and hence their models are only codifications of these. It is then no more than honest to tell the story as well as the model; b) social scientists at the end of the day have to convince others of their theories and hence have to ‘re-translate’ their theories into stories and pictures anyway. I think that is partially why models will never be that popular amongst politicians who are, after all, professional communicators with the rest of the population and thus always will need storylines.
– the curse of dimensionality means the practical distinction between models and ‘waffle stories’ gets blurred. Pretty quickly in formal models, you cant say anything with any certainty as soon as you allow for even a modicum of real-life complexity. ‘Two is stability, three is chaos’ is a telling saying from the world of non-linear feedback loops. Anything becomes possible, depending on parameters you have no hope in hell of ever estimating. What are you to do then? In practise, what I think you then get is that people rely on simpler models for more specific occasions rather than have a formal model for the whole system. And even those simpler models are more to get the right abstractions into focus rather than that they are truly believed. The problem though is that this procedure is, strictly speaking, inconsistent. Conflicting models cant all be right at the same time. Hence, in a sense, models are being used just as if they were unmoddeled piecemeal stories. It is more that they support some stories rather than truly replace them. The main thing they do is to allow the speaker to weed out the concepts he doesnt need for his or her point and they thus shape the language of the stories.
– because most humans dont naturally think in mathematical abstractions, they are more creative in their theorising when they do it in storylines. When using stories, one has the whole of human experience as an analogy, which can be used to find new avenues and new twists. Often creativity just amounts to noticing the similarity between stories and transplanting bits of other stories onto the story being built. It has for instance been argued by Mary Morgan that Fishers’ theory of money derived from noticing that if one visualises the traffick equation as a scale then this scale can be out of balance. The original equation couldnt be out of balance (it was an identity) but the idea of it being out of balance nevertheless lead to new equations. One can also argue that most of market economics relies on the factory analogy and that this image is what we subliminally teach first-years students via the examples we keep using (there is just production with inputs and exchange. No regulator, friction, or service in sight). The main point is that it is probably true that most people are more creative when they use verbal stories than they can be in mathematical models. Hence non-formalism has a role even within pure theory.
Pretty obviously if you’re seeking to pose and test hypotheses you’re model building in some sense and formal methods will be generally superior, though as Paul realises their usefulness is painfully limited by dimensionality. But that’s all us little humangoes have got to go on so we just have to plug away behind the veil of ignorance poking about as we can. (And polish up our appreciation of Socratic wisdom).
Most of the rest of it has to be in words, like the whole of this discussion.
It seems that we’re all in broad agreement here.
The flipside to my comment earlier that “if it doesn’t make sense verbally, it never will” is that “if it doesn’t stay internally consistent formally it never will” and if it’s no falsifiable it never will be.
I always thought it was ironic that once of the legacies of Friedman’s work was the Natural Rate Hypothesis. Whilst it does make sense verbally, which was deemed unimportant, and whilst it is internally consistent, it has no predictive power and is unfalsifiable.
On a side note, I can’t wait til there’s a better theory of preferences, or at least a better toolbox. There were many situations in which I felt guilty using exogenous preferences to explain non-normal goods or many labour supply curves. The explanations make sense verbally, but the appeal to unobserved preferences malleable to the conclusion I already knew was a slight of hand that didn’t explain anything.
This column is also of interest here.
I’m not in broad agreement. I would argue that formalisms are only useful when you have results that genuinely apply in the real world. As in, consistently successful predictions, and useful policy. Economics still has not reached this stage. Economics grapples with fundamental issues such as whether more GDP is a good thing, what type of intervention counts as regulation, and how to put some meaningful upper bound on debt creation.
You can produce wonderful castles of formalism, all based on easily disproven axioms, all useless.
Once you have a theory that works at least more often than it fails, then start down the track of formalism and derivations from that. If you go the other way and start with formal derivations hoping that some useful axioms will pop out of that, let me promise that it won’t happen.
That’s not to say that existing mathematical frameworks (such as chaos theory and system dynamics) aren’t fertile ground for economists to draw from. For that matter, computer models also contain great potential (and that’s the key word “potential”). By definition, a computer model is always a formalism (it’s behaviour is locked to the equations and algorithms in its source code) it can still be completely wrong when compared with any real-world data. These are potential sources of ideas, not answers to the problem.
The other problem is (as Paul points out), it isn’t easy to get real world data, especially if you happen to want details. If a highly trained Engineer is working as a taxi driver, where does he show up in the unemployment statistics?
Conrad’s 9 points presume that Economics is a lot further down the track to being a “real science” than the evidence suggests. The other problem with Conrad’s points is that the building of these formalist castles becomes an art in itself, and people are judged on their building skills instead of whether they build anything useful. Common sense becomes relegated to the dustbin, in favour of ever more elaborate theorising. Chemistry started like this back in the days where people used magic squares and complex rituals to convert lead into gold. No one ever made any gold, but they had such intricate methods of failure!
Tel,
Here’s an alternative scenario for how useful formalism is (not being an economist, it might not be the best example). Let’s say you want to argue for comparative advantage. Actually whether comparative advantage “works” or not is in fact completely non-obvious to many people (just look at all the people that think Australia should manufacture everything). Even better would be an example based on similar ideas, but with a more complex argument, like, if I recall from many Troppo posts ago, someone arguing for an optimal tariff level of a certain percentage on some product (e.g., 8% on product X). Are you more likely to have a convincing argument with or without the ability to show formally, that, at least under restricted conditions, that you end up with the most positive outcome with a level of 8%? It seems to me that of course you have a far better argument with it, otherwise you just end up saying 8% is good for some reason and no-one will take you seriously since you just can’t introspect or verbally reason some number like that (or at least other people can’t). So the model works as evidence.
.
It’s worthwhile noting that the above argument could apply to almost any counter-intuitive predictions. It’s very hard to argue for something that people intuitively don’t believe via verbal arguments. I can’t think of a clean economic argument off the top of my head where intuitive results have been found (perhaps other people can — the Laffer curve comes to mind, but not too many people take it seriously — perhaps other things like the lack-of-effects of the minimum wages do too), but it’s easy for me to think of ones from other areas of science. If for example, you proposed a nice verbal explanation of, say, Quantum physics, people would think you are nut case without the rest. Of course, you can argue that economics isn’t evolved enough for this type of argument to be useful (and that may be right in many areas — again I’m not an economist), but if you want long term predictions where the data is hard to collect, then they need to be based on something, and partial evidence is better than no evidence.
.
The other thing I think you are confounding (I think they are orthogonal issues) is junk science and modeling. Of course there is lots of mathematical junk out there, but that’s true of all areas of science. The same is true of judging people — if people use poor criteria to judges things, that really has nothing to do with formal vs. informal. I have a similar problem in my area — if you do really good work it somehow isn’t as good as really good work with pretty pictures of the brain. So everyone spends oodles of money and time collecting pretty pictures of the brain, which have essentially no use at all.
Conrad,
You’re doing a fair bit of confounding yourself. It seems to me you are taking a series of examples where there is some obvious merit in quantification (though I’ll come back to that because I think even there you’re overstating your case). You’re then taking that as typical of reasoning in the social sciences generally.
Now I may have included a ‘wink’ smiley when I first mentioned it, but I’m serious. This debate is being had in words, not symbols and the reason is that that’s how we do almost all of our reasoning informally – in words – and necessarily so. Formal reasoning can only make any sense when a great deal is already worked out and where there’s some confidence that the formal model has some reasonable resemblance to the world it’s exploring.
Formal reasoning is a bit like a formula one car. We didn’t take one of them to the moon, we took a big boofy four wheeler because we had to navigate tricky terrain and we weren’t quite sure what we’d find. The Formula One car is a like the formal modelling, the words are like the moon buggy.
Of course if one thinks that the circumstances are such that progress will be made with a falsifiable theory, then it makes sense to formalise it in so far as that is possible and the definitions can be agreed upon and so on. (And to the extent that they can’t be, I think they’re probably not falsifiable in any event – so the two things are really cognate.) But take the example of Krugman and strategic trade theory. Until strategic trade theory came along, sensible economists did understand that scale economies were important, but they also knew that to model them one needed to make a whole series of arbitrary assumptions which then meant that when you’d finished you had proved very little – because the ad hoc assumptions made in the modelling made it so brittle to small changes in assumptions that you wondered what you’d achieved.
Now it may have been worthwhile to expend all that effort on strategic trade theory – or some of it. But it’s not immediately obvious. As it was going on various boosters were describing it as a revolution, and some of us knew that it was another assault on problems which had become bogged down and got nowhere much in the 1930s. There were some more techniques around which enabled the new assault but some more seasoned economists were sceptical of what Paul Frijters above has called the problem of dimensionality.
Here’s Ronald Wonnacott an old hand at trade theory – and a sceptic – a decade into Strategic Trade Theory.
So wouldn’t you think it would be sensible for the strategic trade theory literature to contain, fairly centrally, some discussion as to why it was a good place to expend effort, why we had reason to expect better results than we got in the 1930s? Well you won’t find it – or at least I haven’t found it.
A few years later Krugman conceded his disappointment that such “fundamental rethinking of theory can have such modest implications for policy”, but you might wonder how much of the possible disappointment might have been anticipated at the outset, or at least kept under review along the way. If you can find it you’re a better man than me.
This is a small part of the problem of a discipline that is really quite deranged by its unpreparedness to do what it did until the 1950s, which is to painstakingly build bridges between the formal models and the empirical reality they seek to elucidate. So we have whole sub-disciplines not just devoted to what turn out to be relatively harmless wild goose chases like strategic trade theory, but ones like rational expectations, efficient markets and real business cycles that have small kernels of truth in them with whole worlds of fantasy built there-from and policy recommendations wildly asserted from the most brittle of analogies.
We have a discipline which prizes statistical significance (which is technically arbitrary) way above economic significance (which is supposed to be the point of the exercise) – a problem that seems to have grown somewhat worse since Donald/Diedre McCloskey first drew attention to it.
And so it goes on.
Regarding your examples, I certainly don’t want to deprecate what formalism can and should achieve in the areas you’re talking about, but Ricardo expressed comparative advantage in words (IIRC with a small numerical table to illustrate the point). Likewise Quiggin and I put the optimal tariff point verbally – it can be explained equally simply – before I commissioned Peter Dixon to articulate it in a quantitative model.
The fact that people refuse point blank to accept the arguments when expressed in words, doesn’t mean they’ll accept them in mathematics, something demonstrated well enough by my experience in presenting the ‘optimal tariff argument’ in words in these two threads and then backing them up with formal modelling. I didn’t get any retractions from people who had refused to understand the verbal arguments.
Can mathematical constructs be used as haze to fool the gullible? Undoubtedly yes.
But then, Economics exists as both an intellectual discipline, and also as a political discipline. In the later case the conclusion to be reached is well understood (whatever our political masters tell us is true, must be true). The only question is what method do we use to convince the unwashed masses that the economy really is in good hands, no one is cheating the system, and they can go back to working happy in the knowledge that whatever they get is exactly and precisely what they deserve (and look, my mathematical model says so).
OK, there’s also an opposition party, so their pet economists will reliably say that the government screwed up, and we are sitting in the handcart. You know the deal I’m sure. Would you like a chart with that?
Computer graphics get used in the same way, whatever you want people to believe, you just generate some computer graphics showing it happening. Away we go. Most people won’t ask questions because they are too busy oggling at the bright colours, and when one or two people do ask questions you just get some big guys wearing sunglasses to follow them out into the carpark (or however they do this in your particular industry). That’s politics, nothing to do with science or advancing human knowledge.
Don’t get me wrong. Charts, computer graphics, and mathematical constructs all have a useful place too!
By the way, whenever someone does pop up a chart at a meeting or conference, it’s a matter of principle to ensure that at least one person in the audience asks them some technical question about the construction of that chart. Something about the choice of axis, or how the curve fit was done, or the units used… there’s always something. Pretty quickly shows up the guy who pasted in a pretty picture without understanding what it means.
Tel_,
I think you are giving formalism an unnecessary bad rap here. If models are only useful if you can actually trust the predictions they have, then you’d indeed have precious few useful models in economics left. The main use of models is more, in my opinion at least, to get the right concepts in focus when discussing a problem. A model makes you think of the right things. Let me give a stylised example.
A well-known model in labour economics is that of the hold-up problem. This arises when people make investments into the production of a good, whilst they have to agree beforehand what share of the pie they will get. Because they have to share the pie, both will individually underinvest. If there are for instance two people who share the output 50-50, then each individual will invest up to the point where each unit invested begets two additional units (of which they hence individually recoup 1 unit again).
Now, in words the hold-up problem is a pretty streightforward issue to explain, and you’d be forgiven to think the above pretty well sums it up. What are the hidden lessons you learn if you start to model every step explicitly?
1. You need to have a production function. This forces you to think of why each individual cannot produce on his own without this incentive problem. This makes you realise that the hold-up problem needs a notion of complementarities that make the agents more productive if they both contribute. Not just do they have to do complementary activities, but these must furthermore be innate to things you cant take away from them (otherwise an entrepreneur would just buy them). Hence we’re talking about complementarities of innate skills.
2. You have to work out who knows what at each moment in time in a model. This pretty quickly forces you to realise that if the agents could monitor each other’s investment, they could contract on the optimal outcome. Hence you then realise the investments have to be unobservable, or at least not verifiable. This is not a trivial insight, because it also means that if you see a hold-up problem when things are observable, then contracting on the investments could solve it.
3. You have to be precise about timing in a model. If agents would make their investments sequentially, then in principle one of the agents could simply ‘buy’ the intermediate production generated by the other. This would make the investments visible again via the intermediary output, which is clearly not the spirit of the hold-up problem. Hence you are forced to recognise that the investments have to be made simultaneously.
4. You have to be precise about how the output eventually gets distributed, i.e. have some kind of bargaining set-up. The main understanding you get from that is that it makes you realise that the optimal kind of contract is one whereby you trust each other to be reciprocal in your investments, even though this cannot be strictly enforced. Hence it makes you think of social institutions via which the importance of the hold-up model is mitigated. It makes you think of the costs that would be imposed if rewards are entirely driven by observables. It makes you think of the circumstances in which worker co-ownership is useful. Etc.
5. (perhaps most importantly) You get the opportunity to scrap things out of your model if you dont need them. You can thus for instance show by merely formulating a simple hold-up model that you dont need a firm and you dont need property rights or even a market to generate hold-up problems. This makes you realise that even though the model is initially describing two workers, you could re-interpret it to fit other situations, like marriage.
If you are a consistent thinker, you can get at these implictations in words (as I just did), but the discipline enforced by going through the formulation of an actual model ensures that you wont miss a trick and that you thus will have uncovered all the hidden assumptions you need to make for a hold-up problem. Now, you wont find a hold-up model in which people put actual data and via which they make predictions, but it is nevertheless a powerful thing to have built one. The other thing formalism will do is to provide you with a simple mental model via which you will start to see hold-up problems all around you. It is as if you’ve been given the idea of a circle after which you quickly recognise anything looking remotely like a circle.
Nick,
agreed on your story of how our profession is not valuing the right things. I do not think that has much to do with formalism though, but is a simple function of career pressures. For an analogy, just think of theological debates in the Middle Ages. Because there were thousands of philosophers and priests in need of a job, they started arguing about the number of angles on the top of a needle. That debate was fairly useless too, though of course not according to the people debating it.
Paul,
Your comment seems reasonable – though perhaps excessive – that is I expect there are several causes of the phenomenon, but your analogy is (surely) a bit wild. Do you really know that the debate on number of angles on the head of a pin is a product of disciplinary competition?
Paul, fair point, I omitted the idea that models can be useful as toy demonstrations of an abstract principle. There’s a lot to be learned from just starting with a simple idea and tinkering around and looking at what it can do, and provided the limitations and assumptions of a given model are kept clear and honest, I have no problem with it. As you say, a great educational tool, and method for clarifying one’s thoughts.
I guess I’ve just seen too many models turn from a knock-up demo into holy writ without going through much in the way of the difficult stuff in between (experiment, measurement and prediction).
By the way, I’m not so sure about your non-observability criteria in the hold-up problem (I hadn’t heard that particular name before, but I’m familiar with the problem). I would generally say that the problem lies in fair valuation of complementary inputs. Each trade will reliably believe that their particular contribution is highly valuable because their sense of self-worth comes from that. In a marriage for example (ignoring the affairs on the side) both partners have a pretty clear idea what the other one puts in (as a physical observable), but without any available substitute there is no way to attempt a valuation. More than that, I would argue that a fluid marketplace is exactly what provides the consistent valuation over inconsistent inputs, for the hold-up problem to exist there must be some obstruction to fluidity in the marketplace (e.g. in marriage the social convention that frowns upon trade-ins, also makes each party safe from competition, thus shifting their incentives towards shopping or beer as appropriate).
Where there’s genuine non-observability (like using beach sand for concrete instead of river sand, or writing buggy software) you will reliably find someone who tries it on (sometimes a warranty helps that, usually stricter observation is your best friend). I guess that the opportunity of repeat business is an influence, which is where the fluid marketplace comes in… make sure the slackers know they can be replaced (and I’m looking at you Mr Banker).
Hi Nick, Tel_,
of course the hold-up issue arises in many situations and goes by different names. It for instance arises in the electricity sector, between the power generator and the grid and then again with the retail. It also arises with rail, i.e. between the rail grid and the train operators. In each case you will find hard-to-observe investments by the one that affect the output of the other (such as how often the wheels of a train are polished in order to keep them round which apparantly is a crucial issue for how long the tracks on which the trains ride last). The advantage of having it thought trough once in a particular situation, is that you can immediately dismiss it when particular necessary assumptions dont hold. For instance, you shouldnt get a hold up problem between identical entities because they are substitutes, not complements.
As to the issue of angels, hej, why spoil the predictions of a good model with much information :-)
More seriously though, the angel example is not that far-fetched. I obviously cannot prove how it arose to notoriety (for a discussion on that, see here) but it does seem as if discussions like this one basically arose because religious scholars had too much time and too few realword important questions to solve. Like some econ theorists they became pre-occupied with the internal consistency of their own imaginings, looking at questions like whether there are toilets in heaven and if there is a day-and-night cycle in the afterlife.
Paul,
I realise we’re both conducting the conversation with a degree of levity – with the liberal use of smileys to indicate as much. But I think your gung ho use of analogies does indeed point to something important. The usefulness of model building relies just as much on judiciousness in their application as on their quality or the virtuosity of their formal technique. It’s a constant problem in economics that this question of how well the models fit is inadequately dealt with – indeed, it’s more or less dealt with by default. It’s just ignored and regarded as a commonsensical part of the discipline. And analogies are thrown around with reckless abandon. In addition categories which have been hewn in one area – supply and demand in commodity markets – are then thrown around quite recklessly – the supply and demand of political votes, or intellectual ideas etc.
Now there’s nothing wrong with playing with ideas like this. And I guess there are some worthwhile insights in Gary Becker style economic imperialism. But there are a lot of economists running round without much feeling for the fact that such violence to the subject material comes with its own costs and that these costs should be kept in mind as the discussion proceeds.
Your conclusion that the schoolmen had too much time on their hands is essentially tautilogical isn’t it? You’re concluding this from how silly their work seems to us. So that’s not really responding to my challenge. Did Isaac Newton have too much time on his hands – certainly seems to when he got into alchemy and prophecy.
And the schoolmen were not competing with each other in the way modern academics do – or if they were, I’d think we’d need to know plenty more about them than we do (apologies if you really do, which given your breadth of knowledge wouldn’t amaze me too much – it’s certainly beyond my knowledge.)
It seems to me that the schoolmen analogy is probably relevant to this discussion, but beyond taking it as an illustration of the truism that ‘ways of seeing’ are self perpetuating, I’d be loath to take too many lessons from it on the appropriate institutional structure for a discipline without more knowledge. – certainly more knowledge than I have.
Hi Nick,
I would agree that economists are rash in their use of analogies and dont wonder enough about whether they really apply. Its the old story of the hammer and the nails. I would even agree I now and then am guilty of such rashness, but on the other hand it has been one main ways for economists to proceed and to break into new areas. Applying economic thinking to issues within the family, within religions, and within ethnic groups seems to me to have been quite powerful, even though it was bitterly resisted by others who basically said what you said, i.e. ‘dont make such rash analogies’. Hence I suppose you will win some and lose some of you venture ahead with analogies without really knowing all the comparability issues. If there is really no merit in the analogy, then the market for arguments will eventually knock you down.
On the issue of angels and pinheads, it is of course true that I am no expert in medieval scholasticism. Its not something they teach you in econ grad school. Without cheating by googling all the answers, all I can give you is the answers a classical education provided: yes, there was quite a bit of competition amongst Christian scholars. Particularly for the attention of the pope and the cardinals, who elevated some scholastic ideas to dogma (such as the idea of the separation between nobility, clergy and the commoners), but killed off other ideas. Competition between scholastic centres included the competition between the 5 patriarchs (which was won by Rome), the divide between the Catholic and Orthodox church, but also between a whole horde of monasteries. The competition included competition over interpretations of the bible, the divinity of Jesus, the divinity of the church, the divinity of worldly leaders, etc. If these debates became too heated, they were usually decided by papal decree. Within monestaries and centres of learning too, there were more wannabees than going-to-bees. You didnt get to be head inquisitor or bishop without some elbowing. Hence normal human argy-bargy was as much part of the church as it is for modern academia, and scholastic prowess was one dimension of the competition.
Now, one could easily spend a whole PhD thesis on trying to further defend the analogy between the tendency of economic scholars to enter into their own fantasy world, and the world in which medieval Christian scholars worked. I know neither you or I have the time to look that much further. Perhaps someone else already has much more insight on this?
Paul,
I’m certainly not arguing that one shouldn’t make wild associations and see what insights they might provide. That’s quite different from imagining, as many people do, that once you’ve got something down in a model you’ve somehow got closer to the reality. Becker’s analogies are fine, but that’s what they are – analogies, with all the strengths and weaknesses of analogies. Unfortunately we seem to have lost the ability to discuss them as analogies keeping in mind their weakneses.
One of the really big problems is that there is a sense of fatalism about how the discipline uses its resources. This is summed up in your comment “If there is really no merit in the analogy, then the market for arguments will eventually knock you down.” Well what exactly does that mean? It just means that things will turn out the way they turn out. Real Business Cycles and all the other nonsense is all in the mix and will remain so I would imagine, and it will go out of favour at some point. Perhaps the current paroxysm will be enough to chase it away. But it will be replaced by some other idiot savant enthusiasm.
Note that a couple of months ago there was a serious discussion in US policy circles as to whether rising unemployment reflected voluntary withdrawal from work. Right now we have a discipline in which idiotic ideas are lined up with serious ones as if they’re on an equal footing and then told that the market for ideas will sort it out.
In the long run we’re all dead. If you can’t defend yourself against self-evident nonsense it doesn’t seem to me you have much of a market for ideas.
There is a bit of a false dichotomy here. Remember, Mathematics IS a language, itself. If somebody is too challenged to learn mathematics, I would not hold out much hope for their mastery of other reasoning languages, such as English. I don’t have much research in front of me, but a quick back of the envelope reminisce over the scholarship of the formal/mathematically-bereft- particularly those who amplify the “political,” “ideological” utility of formal/mathematically reasoning – is not pretty.
Mind you, the current ruling paradigm in Economics is not long for this world. It has become as monstrous as Ptolemaic astronomy in the late medieval period. But to the formally-deficient, I am afraid the prognosis is not good. Formalism will not go away, it will simply get better. ;)
Hi Nick,
I dont think we disagree on any of this. We both see uses for quick analogies as sources of creative new ideas and would both like to see more and more effort being put into thinking about the validity of the analogy as the arguments get to be made in more serious and more important fora. Whilst truly bad analogies do, I think, get weeded out reasonably fast in the market for ideas, entrenched bad theories do not.
The apparent inability to self-cleanse nonsence once it has grown roots is, IMO mainly due to the fact that they carry a grain of truth in them which is simply overextended but which has famous people’s careers riding on them. You need something pretty big to knock ‘m down. A big recession will weed out a few theories. I think it will take a couple of decades before RBC gets taken seriously again after this recession, though Barro is not letting it go without a fight and, it must be said, many of the textbooks in circulation right at this moment still preach it.
The fact that there is a market for ideas however doesnt mean you simply have to wait for someone else to do the debunking Nick. It is not necessarily fatalistic at all. You and I are part of this market and you seem to be quite happy to knock RBC whenever you can. I am not known to be shy with opinions either. Consider us part of the economic-self-cleaning brigade!
John,
of course mathematics is a language, simply one that is less based on images, storylines and sounds, and thus has different psychological characteristics from English. I dont know about the general relation, but, amongst students, English and math ability often do not go hand in hand.
Paul
Apart from people from non-English backgrounds, I am finding it hard to think of anybody who is fluent in Maths, who is not also very fluent in English.
“I am finding it hard to think of anybody who is fluent in Maths, who is not also very fluent in English.”
Thus spake someone who can’t tell the difference between electrocution and elocution.
The comments function here doesn’t allow for the use of theorem symbols. But if it did, I’d not be alone in running out of such symbols to represent all covalent values of your sheer stupidity, self-obession and spite greenslime.
You could say much the same about computer languages, Java, FORTRAN, etc. They are all mathematical constructs and complete enough that there would be very few mathematical concepts that you could not map readily into a computer language. Some of the efforts at writing theorem-solver programs also produced computer languages that are to all intents and purposes a direct mapping of mathematics (and you can try this at home with Maple-5, Macsyma, Mathematica, etc).
Having said all that, none of these constructs has yet replaced human thinking, other than in highly specialized and ultra-narrow problem domains. That’s because many real-world problems are poorly defined, have no measurable boundry conditions and the biggest part of the problem is getting to an understanding of what is the problem. Even once the problem is defined, the input values are unreliable, and even unreliable to an unknown and inconsistent degree.
And I agree, computers (and other formal constructs, just about all of which depend on computer support as soon as they get even moderately complex) are getting better at solving problems that were once considered only approachable by humans. A good example if massively high dimensional optimisation problems (e.g. upwards of a million unknown variables), the “TimberWolf” algorithm for laying out circuits is probably the best known (and highly successful) example. At the time TimberWolf was published, it was a bit of a shocking left-turn for the Artificial Intelligence community because it does not use any logical inference, nor deductive reasoning to solve the puzzle. It merely throws things down wherever, then shuffles a bit, then reshuffles.
I’ll also point out, that no one has come even close to teaching a computer to speak English.
It’s a pretty common situation in the under-21 computer geek community. I can find some system documentation extracts if you need proof.
Oh you poor thing, the angrier you get the duller you get. Or were you just sober when you posted? Darl, I have noticed that the majority of your tirades across blogs are about posters’ vocab. You are, I suppose, a male version of Miss Hathaway, perhaps even Jethro! But hun, “electrocution lessons” is a globally-renowned funny, ironically usually a weapon of your types. It is a Kath and Kim-esque riff on “elecution lessons”. LOL! Geddit now? Twit.
Tel
Hmmm..you’re probably right. My sample is probably extremely biased. ;)
“My sample is probably extremely biased.”
Lemme guess… glib fuckwits addicted to attention of any kind?
“the angrier you get…”
Nah. Just bitch-slapping a total twat in passing. But boy greenslime you do get excited at the thought you might have aroused in others any emotion other than contempt, don’t you?
“I have noticed that the majority of your tirades across blogs are about posters vocab”
I have noticed you often just make shit up when trying to score a point.
“But hun, electrocution lessons is a globally-renowned funny, ironically usually a weapon of your types. It is a Kath and Kim-esque riff on elecution lessons.”
Shorter greenslime “I meant it on purpose, really I did.”
There you go, your ration of attention for the day. Now you can slither off now like a good little gollum.
To be fair, comment #28 managed to go a whole paragraph without saying “luvvie” and it was a legitimate point on the topic. If the principle of Proportional Response applies to Israel, then is should apply around here to.
Yeah, it was real funny when Kellogg, Brown and Root did it to a dozen their of their own troops (not counting the ones who survived), while loading their pockets with cash handed down by Dick Cheney. I’ll bet they are laughing all the way to the bank. So how bad can the corruption get before the US falls to pieces? And where’s your mathematical model to show none of this is happening?
FDB/Nabakov
Wow, you two really don’t like me, do you? But why? I can’t recall being nasty to either of you, and even if I were, what would it matter as you post anonymously? The stuff you have posted here is not very nice. In fact it’s downright nasty and hateful.
You both should read over the stuff you have posted to me and have a good look inside yourselves.
This all went on while I was out of town. Nicholas, you are lucky that you did not have time to read the whole article, the logical positivism/empiricism movement burned up a lot of high quality brains and many careers without producing much that contributed to science or social policy.
On a historical note, Karl Menger, the son of Carl, was actually conceived with the assistance of a housemaid. Menger senior took the lad in and raised him as his own son. The fact that Menger junior turned into a positivist provides a striking example of unintended consequences.
Carl Menger (founding father of the Austrian school of economics) was a apparently a bit of a lad (a bit like Schumpeter you might say) in his youth he wrote slightly scurrilous novelas and pamphlets under a false name, and he lamented in later life that too much of his youth had been consumed by high living and love affairs.
Well that put’s it about lineball with mainstream economics, except perhaps the economists got better pay.
Rafe,
I’d missed your contribution here. (Or I actually probably didn’t, but have forgotten it.) In any event, reviewing this thread, I’ve (re)discovered it and am glad you ‘shared’ it with us as we say these days.
[…] [fn1]And by “maybe” I mean no, but that’s a debate for another time. […]