If you are ever puzzled by the phrases “a priori” or “a priorism”, in connection in particular with the writings of Ludwig von Mises, and would appreciate becoming less puzzled, then let me now recommend to you a recent essay by Detlev Schlichter entitled The a priori method in economics – In defence of Ludwig von Mises.
Perry Metzger’s recent demolition job, here at Samizdata, of Thomas Piketty, made me think of Schlichter’s essay. What Perry Metzger was giving us was surely a perfect example of the a priori way of thinking about economic events and economic evidence.
Economic events, even if alluded to with pertinent statistics, can only be said to make sense once you have made sense of the human judgements that gave rise to these events. Future events cannot be predicted merely by looking at numbers and at graphs and then guessing at where these numbers seem to be heading. The economic future is made of human judgements, and to get a handle on what that future will most probably be like, you have to interpret that future in terms of rational human judgements and reactions.
Piketty points towards what Metzger names and shames as an “Investment Event Horizon”, a world in which investment is done far too much, with all other economic activities being curtailed in the service of this one obsession. Piketty says there must be political action to impose rationality on such otherwise irrational events. Metzger says that it is Piketty who is being irrational. Piketty’s interpretation of his supposedly supportive statistics is a perfect example of the kind of thing that Detlev Schlichter is criticising.
In general, says Schichter:
… We can neither verify nor falsify the a priori laws of economics … Even more important (and potentially disappointing to those who derive their expectations as to what science is all about from the natural sciences) we cannot derive the laws of economics from mere observation, and that includes even the most extensive collection of data and the most elaborate and sophisticated analysis of it. The economists who claim to do this are either confused or simply play to the gallery (Piketty?) and are frequently not really proper economists, although some of them may even win Nobel Prizes. This may sound harsh but I believe it is true. The reasons for why we must fail to achieve these two things (test/verify/falsify economic laws and discover economic laws through statistics) are fundamental and I will give them below. Of course, if economics were a natural science, if it were an empirical science, these two things would not only be possible, they would be essential to its modus operandi as a science. Crucially, economics is not an empirical science in the sense that the natural sciences are.
This is in fact the reason why no amount of data mining and statistical analysis will ever settle disputes in the field of economics. Keynesian economists will forever quote historical data from around the Great Depression as evidence of their crisis theories and policy recommendations, just as those who subscribe to monetary explanations of the business cycle (as we “Austrians” do) will forever cite the same or similar data in support of their theories. It is a common complaint that anything can be proven with statistics, and in the field of economic debate this seems to be true to a large degree. (I subscribe to the “Austrian” explanation of economic crises not because it fits the data better but because it fits the principles of economics, the laws of economics that allow us to analyse the cycle in the first place. A detailed analysis of Keynesian theories leads to conflicts and mismatches with some key economic principles. This makes this theory much less convincing.)
I consider that last bit there in particular, the final three sentences in the brackets, to be SQotD-worthy.
In the first of the two paragraphs quoted above, Schlichter actually mentions Piketty, albeit with his name between brackets and with a tentative question mark attached, merely as a possible example of the kind of thinker he is criticising. But if Metzger is right about Piketty (and I would be amazed if anyone is able persuasively to show him not to be) then that question mark can be dispensed with.
The intellectually corrupt Karl Marx, having promised his publisher a proof of impending proletarian revolution caused in the meantime by proletarian immiseration, then felt obliged to fiddle his numbers to prove that things were going from bad to worse for the nineteenth century proletariat rather than from bad to better. (Metzger might describe Marx’s yearned-for revolution as something like a “Discontent Event Horizon”.) The Piketty story has the look to me of a farcical rerun of that intellectually sordid Marx episode. (On the subject of the bad faith of Karl Marx, I recommend the late and much missed Antony Flew’s 1991 Free Inquiry/Libertarian Alliance article entitled (by the LA) KARL MARX WAS NOT A SOCIAL SCIENTIST (the LA likewise subtracted a question mark). Scroll down to where it says “FALSEHOODS OF IMMISERATION”, on page 5.)
But it isn’t just that Piketty’s numbers are wrong, although it would seem certain that they are. It is, as Metzger points out, that even if Piketty’s numbers were correct, Piketty’s understanding of what he believes they must be pointing to is defective.
Which just goes to show that if you come to a big pile of economic statistics with some a priori principles that are wrong, because they don’t make sense, then you cannot hope to make any sense of the statistics.
Piketty would presumably say that he is not doing this, merely letting the statistics speak for themselves. But neither can you make any sense of a pile of economic statistics if you expect those statistics to tell you everything you want to learn.
I met Detlev Schlichter earlier this week, and he told me that he had not then read Perry Metzger’s posting. I recommended to him that he might particularly enjoy it.
As my tutor once said to me (in frustration, I admit): ‘Look, facts aren’t just things lying around on the floor’.
And similarly, ‘the economic data’ isn’t just lying around on the internet.
Chris Giles, in a detailed analysis at the Financial Times about Piketty’s use of data, has this devastatingly mild paragraph: (Item behind the FT’s paywall).
My point is that if someone is claiming to have found a fundamental contradiction of capitalism, predicts the result is a rising share of wealth inequality, and uses apparent recent rises in recent wealth inequality as evidence that his theory is correct, those data lie at the core of the book’s argument. Without the rising wealth inequality data across all advanced economies, I would submit that Prof Piketty has a theory without all the necessary evidence.
There is no way to say this more diplomatically than Chris Christie would, so i won’t try.
Schlichter displays a disturbing amount of mental confusion about induction (sensu Aristoteles/Francis Bacon/JS Mill) and Popperian epistemology.
So much so, that his post cannot be read intelligently without being aware of the incompatibility between the 2 positions.
Let me point out just one short sentence from Schlichter’s post:
Indeed it is, but it might have escaped Schlichter’s notice that it is just as impossible in physics as it is in economics.
(Actually William Whewell is even better than Popper about this, see especially his dispute with JS Mill.)
So when Schlichter bizarrely claims that
he is showing that his view of natural science is little more than a Baconian mythology. Starting from this mythology, he is in a poor position to distinguish between the methods of the natural sciences and of economics.
That is not to say that i view economics and the social sciences as “scientific” in the same way as the natural sciences: it’s just that Schlichter makes a very bad argument against this view.
(Possibly even worse than Farage’s argument against immigration: i likewise am wary about mass immigration, but it seems that Farage makes arguments especially designed to make me repudiate them.)
Maybe i’ll articulate my dissent more fully tomorrow.
Snorri:
You pre-empted me, but you probably said it better than I could have. In fairness to Schlichter, though, his belief about the role of inference in science is extremely common today, and is probably subscribed to by the majority of practitioners in the physical sciences, too.
I strongly suspect that this obsession with empiricism and inference is largely a reaction to the cultural war raging between science and religion. Scientists see empiricism as their trump suit, but it is really the every-increasing explanatory power of theory that gives science the edge. Empirical verification is just a tool that assists that process.
Tedd: thanks for the support. My experience is different from yours, though. Before switching from neuroscience to AI (and therefore away from empirical science) i found that my anti-induction remarks usually fell on deaf ears … because nobody believed in induction anyway!
OTOH it is true that, when talking to laymen, scientists often fall back on dogmatism, eg Dawkins claiming that “evolution is a fact” … which it is, except that all facts (including the fact that Dawkins claims that evolution is a fact) are theory-laden.
I could go on about climatology but i won’t.
Thanks Snorri and Tedd – something did sound fishy, but I didn’t stop to think about it until you mentioned it. But as Tedd mentioned in Schlichter’s defense, this understanding of “science” has become so widespread even among some of the scientists themselves, as to take non-scientists such as Schlichter off the hook, for the most part. (Cue in the Climate Debate…)
I don’t dispute these criticisms. I linked to Schlichter because he does seem to understand a priorism, not because he understands regular science, which I am guessing he has not thought about as much.
On the other hand, Tedd’s point about how inference dominates the “philosophy” of the average scientist (quotes because it isn’t usually thought through very thoroughly – I have been told that scientists who think seriously about the philosophy of science are widely dismissed by other scientists as failed scientists) applies also, I should surmise, and in particular, to economists. They believe that in practicing inference,or trying to, they are doing proper science. But, I agree with SG’s complaints about that aspect of what Schlichter says.
The other odd thing about Schlichter’s piece is his assumption that animals don’t ever make choices. Only humans. Vegetables, yes, that I would readily concede. But anyone who has ever watched a cat or an urban fox going about its business would surely at least have severe doubts about such a claim. And a refutation of the idea that cats and urban foxes make choices might end up applying also to humans!
Again, this criticism is not central to what Schlichter says about a priorism. The conclusion might instead be that economics could also have things to say about the behaviour of cats and urban foxes. And I bet it does.
Maybe when asserting that only humans make choices, Schlichter is merely channeling von Mises. In which I think von Mises is wrong about that.
I still think Schtichter is right that economics and regular science are entirely distinct forms of knowledge. Maybe my brief and sketchy descriptions of this difference are actually better than his!
Indeed!
I disagree. My understanding of Popper’s thinking is that it is based on the notion of fundamental agnosticism, and that qualitatively applies to any sort of knowledge. I think that the difference between “scientific” and “non-scientific” knowledge is quantitative – i.e., the only real difference is in the degree of uncertainty (which is never a perfect zero).
The problem with economics as science is that economic experiments cannot be conducted in anything even approaching the precision controlled conditions required for the replication of similar previous experiments in order to attempt to refute a theory. The same of course goes for all other Social “Sciences”.
…as well as for Climatology. I’ll shut up now…
A good and important post.
It can be summed up as follows………
If something does not make sense (if it violates the laws of reason) it is FALSE.
Contrary to Milton Friedman’s early 1950s essays on “Positive” economics – it DOES matter that one’s theories make logical sense.
“Empirical data” and the “power to predict” are not enough – not in economics they are not. Especially (given the incredible complexity of the human world) “empirical data” can be piled up SEEMINGLY in support of any theory – and of the OPPOSITE of that theory.
So one is forced back to the laws of reasoning.
Does what is being suggested MAKE SENSE?
If it does not make sense (if it violates the laws of logical reasoning) it is FALSE.
Snorri:
It may be that my perspective has been warped by too much exposure to scientism, and to a few high-profile scientists (such as Dawkins) who make a bigger deal out of empiricism than I’m comfortable with.
It strikes me that economics has some parallels to neuroscience and AI: tons of data that isn’t necessarily helping all that much.
“It strikes me that economics has some parallels to neuroscience and AI: tons of data that isn’t necessarily helping all that much.”
Indeed, it has long been my contention that economics belongs among the “social sciences”; it is merely a branch of human psychology. Unfortunately, for the last century or so economists, envious of both the intellectual rigor of the physical sciences and the academic standing of its practitioners, have been indulging in the conceit that economics is itself one of the physical sciences, subject to the same methodologies. Clearly it is not, but the fantasy persists, and manifests itself in the elaborate mathematical models which are now de rigueur in economic treatises. (By comparison, examine a nineteenth century article on economics; rarely will you see any mathematical formulae whatsoever.)
And I think this is Schlichter’s central point. He may not be as fluent as Snorri and Tedd in the minutiae of the philosophy of science (he is, after all, a bond trader and practical economist, not a scientist), but the point he is making is sound. Complaints about his “mental confusion about induction” are irrelevant distractions.
‘Neuroscience’ is not a science, it is not even stamp collecting. I pray in aid Lord Rutherford.
http://www.quotationspage.com/quote/26219.html
It is bullshit with electrodes.
Re. “Karl Marx was not a social scientist”
He probably THOUGHT he was.
Generations of his murderous minions have declared him to be.
A long time ago I brought a veritable Niagara of opprobrium on myself by declaring to a group of lecturers and fellow students, that, “”social science” was a contradiction in terms”.
Instant outcast!!
Note that the full spectrum of bolshevist, totalitarian sociopaths cleaves to the “principle” of “Scientific Socialism”.
Given their obsession with semiotics and other quaint linguistic “sleight of mouth”, it is just a “jump to the left” to arrive at such a state of mind.
It would be interesting to see what “culture” does to this “science”; do their “immutable laws” work in Cambodia the same way they work in the “zoo” of Berkley, or, say Argentina or Sudan?
Economics is simple: Demand and supply
Keynesian Economics is not valid as proven in 1938.
In America there was a recession after WW1 but the economy recovered quickly and the Great Depression was the result of overproduction and excessive investment in the Stock Market i.e. investing in inflation, what can go wrong? Fucking lots!
It may interest you proud Englanders to know England started building concrete highways long before the Seppos and Germany but long distant lorries were only introduced during the sixties by British Leyland with power steering, synchromesh gear boxes and power steering. The irony about British Leyland is that the trucks were reliable but the cars crap and Margaret Thatcher who won four elections was blamed for the decline of manufacturing in the Midlands when in fact it was the mindset of Trade Unions and using crap parts i.e. Lucas the Princes of Darkness.
That should have included heaters and demisters, senility strikes again.
Mr. Ed could not be more right on this matter and that quote is fantastic. Amen.
Actually, yesterday i was already irritated for personal reasons, before reading Schlichter’s post. Apologies to Schlichter in case he reads this. I stand by the substance of what i wrote, though: DS is correct that economics does not, and cannot, use induction, and he is correct that economics is different from physics; but these are 2 entirely separate issues.
Just last week, i was reading something by Popper that is worth quoting here:
Source: Objective Knowledge, chapter 1.
(Italics in the original replaced by boldface here.)
The first part is pretty definitive: all our knowledge is psychologically a priori; no learning “directly” from data, not in physics, not in economics, not even in everyday life.
The second part raises the issue: is economics a priori valid?
Looking at Schlichter’s essay, i find something that is valid by definition, hence a priori; eg the law of diminishing returns, as DS says:
But i also find a couple of predictions by any other name:
The reason these are not strictly falsifiable is the word “lastingly”: if a minimum wage improves the conditions of the poor, it can always be said that it won’t last forever. (I think so too, of course.)
Perhaps the best counterexample to Schlichter’s argument, actually, is the title of his own book: how long is he willing to wait for paper money collapse before giving up his prediction that it will happen?
PS: looking at the comments, i see that Alisa already made a similar point to mine, about the line between physics and economics being blurred.
I also agree with Brian that animals do make choices (though with a much shorter time perspective); we might also agree that humans often do not make choices, but act out of habit.
Laird:
The problem with Schlichter’s argument is not his ignorance of some minutiae, but his basic misunderstanding of the epistemological principle involved. This a real problem, because when one makes a bad argument in support of a good cause, one is likely to damage the cause itself – the cause in this case being promoting the notion of fundamental agnosticism, of the understanding that everything and anything at all we perceive should be subject to a certain degree of uncertainty and taken with a grain (or a shovel, as the case may be) of salt. This goes way beyond both physical sciences and economics, it is about constantly questioning absolutely everything. Our current sorry state of human affairs, in any field, really – be it economics, science, politics, religion, you name it – is nothing more than a symptom of the (non-coincidental) failure by the Education System to instill this simple principle in the minds of the population.
I think we are making a mistake that because mathematical/statistical/empirical methods are hard to apply to ecomnomics therefore a priori is the only way we can make progress. I would contend a priori reasoning is also hard for economics. A priori reasoning certainly exists in Physics. Einstein’s derivation of Special Relativity is a brilliant example, proved to be correct empirically many years later.
Physics is “easy” because it concerns a relatively simple set of objects that appear to follow matehmatical laws and form simple enough systems to be very tractable to mathematical/empirical falsification. Chemistry (how lots of atoms interact to form chemicals) is a bit harder because there are more moving parts, biology (how lots of chemicals interact to form organisms) much harder, ecomonics (how lots of organisms interact to form societies) harder again. The level of mathematical complexity in these higher order systems is so overwhelming at the economic level that you can’t produce solvable equations without massively simplifying and your simplifications may be material. The systems are almost certainly chaotic which means you can mathemtaically prove that you can’t makre predictions of anything other than broad patterns. However a priori reasoning also becomes more and more difficult due to the range and complexity of different interacting elements in the system. Hence lots of economists create theories based on a priori arguments that disagree with other economists’ a priori arguments and because of the mathematical problems above it’s darned difficult if not impossible to make mathematical predictions based on the theories or to conduct experiments which would conclusively falsify. So bottom line is there is a lot of uncertainty as to what the truth is. I think we are best to use a mixture of a priori reasoning and empirical data because if your theory clearly falls foul of either reason or the facts then its wrong. But being in a situation where it is indeed “clear” that your economic theory is wrong rare (at least since people stopped trying to go for total central planning). Having said all that I think it is certain that any type of economic reasoning is uncertain and therefore I also think there are very good (a priori) reasons not to attempt central planning. However whether or not there are economic (as opposed to moral) reasons not to attempt moderate regulation and “nudging” I think is not comething we can conclude from these epistemological difficulties.
Animals and choices. The demonstration is fairly simple. Take one cat, and place it in one bedroom with not too much furniture. Wait until cat is in corner of room (intersection of wall A and wall B). Say “trip to the vet”. Walk purposefully towards cat, following a line at a 45 degree angle from either wall. What happens next is interesting. As you approach, cat makes a feint to run to safety along wall A, but stops when it sees you moving towards wall A in a line of interception. It will then make a feint to escape along wall B, but again stops on seeing you about to intercept it. As you move closer, cat hesitates, clearly realizing that neither the wall A nor the wall B escape route offers any advantage over the other. At this point, a simple algorithmic choice mechanism would get stuck in a loop of indecision. The cat, however, will invariably end up making a dash for it along one or other escape route, chosen, as far as we can tell, at random, in a way eerily reminiscent of discredited UK political party leaders.
Endivio: that might be true of your cat, but not of Schrödinger’s cat. Schrödinger had a cat in a superposition of 2 states: run-left and run-right. When Schrödinger approached the cat, the wavefunction collapsed into 1 of the 2 states.
Quote of the year.
Collapsing a wavefunction? Shucks, that’s nothing. Even my tortoise can do that. You tell Schrödinger from me, my cat can conduct the New York Philharmonic doing Night on a Bare Mountain with its tail. Beat that.
Inspired by this, I went looking for Mises’ “Human Action”. (I presume from the title that he, too, unfairly excludes feline and canine, etc., action.) Within three minutes I had the scholarly edition for Kindle – for the astonishing price of £2.92.
I’d willingly have paid more if it had a supplement dealing with the economics of lolcat behaviour.
Believe it or not, i worried over the weekend, on and off, how i’d answer the question that i put to Detlev Schlichter at the end of my comment @8:06.
Now let me put it off my chest: Schlichter should never declare that there is no more danger of paper money collapse; but if he does, that’s the time to buy gold if you are a contrarian.
That’s yet another difference between economics and the hard (or: slightly less soft) sciences: nobody in their right mind would start disbelieving in evolution just because there are no more creationists.