Show Me the Empirical Success of the Empirical Method in Economics
In the comments of my latest post, as well as the second front of this war going on here, we’ve been raising the usual issues that come up in this debate on the empirical method in economic science. I have to be brief, but let me summarize my perspective:
(1) Many people simply assume that to be scientific, a discipline must use what we learned as “the scientific method” in fifth grade. Namely, the idea that economists must use their models or theories to generate falsifiable predictions, and then look at the actual observations to see which theories perform the best. We can never know an economic law for sure–just like we can’t “prove” a law of physics–but that’s what it means to “do science.” Well, if that’s your view, then I am saying economics is not a science. However, I would dispute your view; I claim that there can be a science–economics for example–where what we learned as “the scientific method” is not a fruitful approach.
(2) “What’s the alternative?!” you ask in shock. My answer: Everything you get in a good introduction to “thinking like an economist.” Consider the classic essays from the great French liberals. Consider Henry Hazlitt. Consider Steve Landsburg’s Armchair Economist–it’s in the very title, for crying out loud. In fact, go look at the textbook and syllabus that a free-market economist who champions the empirical method uses for his/her first-year students. I will bet you that the vast majority of what they teach involves getting the students to think a new way, rather than saying, “We’ve gone out and tested these propositions numerous ways; trust me, this is how the economy works.” The way you get someone to “believe in” free trade isn’t to show numerous empirical studies. No, you give him something by Bastiat (or Thomas Sowell) and he’ll say, “Whoa, I never thought about it like that before.”
(3) What is the empirical success of the empirical method in economics? Eighty years since central banks caused the last worldwide depression–according to various schools of thought, though the details differ widely–we now have central banks, under the expert leadership of Ivory Tower-trained economists, causing the second worst worldwide economic calamity in modern times, and again according to economists from across the spectrum (with details varying). Is there anything remotely analogous in physics or chemistry?
The supreme claim of the engineers, chemists, and physicists is, “We put a man on the moon.”
In contrast, Ben Bernanke, Larry Summers, and Christina Romer can say with pride, “We put a man in an empty region of space where we erroneously thought the moon would be. Of course that man is dead now, but what do you want from us? This is macroeconomics, not rocket science.”
(4) Don’t misunderstand: I’m not knocking counterfactual analysis. What I’m saying is that if you can’t predict very well, then using the empirical method in constructing your counterfactual is very dubious. It is precisely the logical, deductive approach that makes you very confident in your counterfactuals, in a world of highly unpredictable absolute movements.
Let me give an example: Suppose a guy is on trial for embezzling $1 million from his firm. He doesn’t deny that he took the money, but he claims the company benefited. His evidence is that the shareholder equity and stock price of the firm went up, all during the period of his embezzlement. The prosecutor is at first rattled by this, but then says, “Well of course, the company would be even richer had you not taken that $1 million!” But the defendant says, “How do you know that? Have you conducted empirical investigations of this assertion? I want scientific evidence for your counterfactual claims.”
Now sure maybe the prosecutor could try to come up with something, but can we all agree that the best course of action here–the truly “scientific” thing–would be an appeal to the pure logic of mathematics and accounting, rather than digging up studies of past embezzlement cases, where issues of correlation/causation have been satisfactorily handled?
You can only really argue about the empirical stuff. I think that’s why economists gravitate toward it. The minute you start talking about economic theory, the conversation quickly descends into philosophy. It is no coincidence that Mises wrote so much about epistemology. Philosophical debates often come down to whether or not truth is real, or whether it is knowable even if it is real, or etc. etc.
But you can’t get lost like that in a statistical model because everyone agrees on the rules and it’s easier to just sit around critiquing the integrity of the data or whether a semi-log-linear model would have produced a higher R-squared.
But you can’t get lost like that in a statistical model…
So it’s impossible, say, for Brad DeLong and Paul Krugman to go nuts on Greg Mankiw over the claim that real GDP has a unit root?
No, no, maybe you misunderstand. All I’m saying is that it’s easier to discuss math than it is to discuss philosophy, so a lot of people talk about math to avoid having to talk about philosophy.
Funny. I quite disagree. There’s even a saying, ” math is hard.” MYou seen the average size of classes in pure math versus English, philosophy, sociology or any other talky discipline?
See, in a math class you have to talk about math. In a philosophy class, you don’t really have to do anything at all.
Synthesizing different philosophical ideas, agreeing with one or the other position and arguing why that is true is very different then regurgitating ideas of dead philosophers. Math may be difficult to learn, but philosophy is difficult to DO. Saying “use math and the scientific method” is far easier than debating and justifying epistemology.
Philosophy is, statistically speaking, more difficult for most people than mathematics.
Philosophy is what underpins not only mathematics, but every other field of inquiry. Since philosophy backs all fields of inquiry, philosophy is more complex than any one field of inquiry, including mathematics.
Most can readily agree that 2+2=4.
But how many people agree that praxeology constrains our knowledge of the real world, or that the real world even exists as it is conventionally understood? Philosophy is incredibly complex. It takes 1000 pages of philosophy to show and explicate the theory that humans act. It only takes 300 pages to prove 2+2=4.
That’s my point exactly, MF. Thanks.
“Philosophical debates often come down to whether or not truth is real, or whether it is knowable even if it is real”
Strange, given that such positions are self-refuting. If one purports that “truth is not real”, that is a REAL truth claim. If one purports that “one cannot know truth”… how does one know that?
It is impossible to argue against the Law of Non-Contradiction. To claim it false is to create a paradox of the same kind as “this statement is false”. If things can be BOTH true AND false, then so can the Law of Non-Contradiction… but it cannot be true at all, or it can’t be both.
I agree with your epistimological position Matt, but as you say “the debates still happen”. e.g. I used to go to a philosophy meetup for many years. My main interlocutor would constantly dispute the notion of truth. I would point out (as you did above) the absurdity of it, and his response was to laugh. He laughed. Every time. To say this was frustrating would be an understatement.
Just hit him. When asks what you did that for, you can laugh, then you say, “do what?”
You can at least check any theory for logical consistency, and further check it for circular arguments (which are logically consistent, but useless).
Logical consistency is a necessary but not sufficient condition of a good theory, you can knock out bad ideas this way, and that’s about all.
Gotta run before I fully digest, but here’s a good thing from Grampp (1952) to keep in mind for your #3:
“By reasoning from the actual practices of mercantilist states, economists and historians usually have supposed that the doctrines of the period of mercantilism were a justification of its institutions. It is common in studies of mercantilism for the author to explain, say, the restriction of imports by referring both to the tariff duties of the age and to the concurrent doctrine of a favorable balance of trade, or for him to move freely among expressions of public officials, laws, economic tracts and discourses, and to suppose that because particular controls were exercised, like price fixing, they must have been justified in the economic writing of the time. No one, of course, would write of recent economic policy this way. It would be unthinkable to describe the New Deal by an indiscriminate reference to the works of Keynes and Hansen and to the public papers of Franklin D. Roosevelt and the private memoranda of Harry Hopkins and always to suppose that whatever the state did or wanted to do found its rationalization in economic doctrines. When studies of mercantilism use a method of this kind, they leave an impression with the reader that in many ways is distressingly wrong. He must be led to think that because the mercantilist states did not believe in the market as the mechanism for discharging the economic functions of society, the economists of the age held the same belief and were in favor of the intricate kind of regulation which was practiced.”
Granted, I don’t know of anyone but Time Magazine who says that the Fed even can save the world.
The central bank caused this crisis by failing to follow the empirical method. The evidence of a housing bubble was overwhelming. The consequence of bond agencies giving AAA ratings to junk bonds was overwhelming. The central bank stuck to an ideology in which the market is best left alone to sort itself out.
But Daniel Kuehn and joe, why does this type of thing happen all the time in economics, but not in physics? We have debates over whether to drop bombs on Syria, but there isn’t any dispute among the top mathematicians and engineers about HOW to drop bombs in Syria. Compared to the natural scientists, the economists (following the empirical method) look hilarious.
Because different people value different things Bob, and because the economy (and human society) is a complex system so that even when we are on the same page there may be disagreements on some issues. Is this really the best you’ve got?
DK wrote:
Because different people value different things Bob, and because the economy (and human society) is a complex system so that even when we are on the same page there may be disagreements on some issues. Is this really the best you’ve got?
Oh, so the reason we keep having worldwide depressions is that 30% of the public likes them, and 42% dislike them? In contrast, there was a clear consensus that Americans didn’t want Neil Armstrong to die because his rocket missed the moon?
Yes, “the best I’ve got” is to show you guys the stunning failure of the empirical approach. You don’t win the argument by saying, “Well, the economy is a far more complex system than the material world.” Right, that’s one of *my* arguments.
“Because different people value different things”
Do different people value different things in a constant manner over time, or are different people’s different values not constant over time?
If you are inclined to say the latter, then congrats, you just validated what Murphy is arguing in this blogpost.
Empiricism doesn’t work in economics, because empiricism by its inner structure presupposes a constancy over time. One cannot claim that a theory posed in the past, and tested in the present, is “confirmed” or “falsified” in the future, unless a constancy in relations is assumed.
But as discussed above, since values and hence actions are not constant over time, empiricism cannot do anything other than describe history.
” The evidence of a housing bubble was overwhelming.”
If it was, there would have been no housing bubble. I notice you say this now, comfortably past the bubble.
“The central bank stuck to an ideology in which the market is best left alone to sort itself out.”
If it did that, it wouldn’t exist at all…
Tell that to Ben Bernanke, who repeatedly scoffed at the possibility of a housing bubble before the crash.
The consequence of bond agencies giving AAA ratings to junk bonds was overwhelming.
The rating agencies are a government-created cartel:
Who Captures Whom? The Case of Regulation
[WWW]http://mises.org/daily/2320/
In 1975, the SEC created a designation for credit-raters: Nationally Recognized Statistical Rating Organization (NRSRO.) The SEC rule created a cartel of 3 firms: Standard & Poor’s Credit Market Services, Moody’s Investors Service, and Fitch, Inc. — these being the only firms that qualified for use by broker-dealers for satisfying another of the SEC’s rules on net capital. As North’s model predicts, the credit-raters were only too happy to take the bait and endorse this arrangement.
What’s the empirical way to measure a bubble? What is your reference?
“Namely, the idea that economists must use their models or theories to generate falsifiable predictions, and then look at the actual observations to see which theories perform the best.
The empirical/inductive method is more than that: it involves making sure your axioms, postulates, and assumptions actually do describe the real world and empirically verifying this.
E.g., Mises’s disutility of labor can only be established by empirical evidence
The disutility of labor is not of a categorial and aprioristic character. We can without contradiction think of a world in which labor does not cause uneasiness, and we can depict the state of affairs prevailing in such a world …. Experience teaches that there is disutility of labor. (Mises, L. 1998 [1949]. Human Action: A Treatise on Economics, Ludwig von Mises Institute, Auburn, Ala. p. 65).
And even Mises accepts this basic sine qua non of empiricism:
“Every theorem of praxeology is deduced by logical reasoning from the category of action. It partakes of the apodictic certainty provided by logical reasoning that starts from an a priori category. Into the chain of praxeological reasoning the praxeologist introduces certain assumptions concerning the conditions of the environment in which an action takes place. Then he tries to find out how these special conditions affect the result to which his reasoning must lead. The question whether or not the real conditions of the external world correspond to these assumptions is to be answered by experience. But if the answer is in the affirmative, all the conclusions drawn by logically correct praxeological reasoning strictly describe what is going on in reality” (Mises, L. 1978 [1962]. The Ultimate Foundation of Economic Science: An Essay on Method (2nd edn, Sheed Andrews & McMeel, Kansas City. p. 44).
These points make massive concession to empiricism. It appears you forgot them.
And these distortions are breathtaking:
“Eighty years since central banks caused the last worldwide depression–according to various schools of thought, though the details differ widely–we now have central banks”
*Most schools* say the central banks exacerbated the 1930s worldwide depression by not acting properly to stop financial sector collapse. Even the GMU Austrians like Steve Horwitz concede this.
“we now have central banks, under the expert leadership of Ivory Tower-trained economists, causing the second worst worldwide economic calamity in modern times, and again according to economists from across the spectrum (with details varying)”
Across the spectrum?? You mean *confined virtually to the Austrian school and libertarian fringe*.
Virtually all other schools say central bank intervention averted a worldwide financial collapse and second great depression on the scale of the 1930s.
LK, you’re saying Austrians are the only ones right now saying that central banks are responsible for how awful our economy is? Like, Scott Sumner and Nick Rowe don’t think there’s anything that can be done, and Paul Krugman/Brad DeLong agree that central banks are utterly powerless?
In any event, even Krugman blames the Fed for not stopping the housing bubble. He just thinks it should have acted with regulation, not by raising interest rates.
“LK, you’re saying Austrians are the only ones right now saying that central banks are responsible for how awful our economy is?”
You’re committing a fallacy of equivocation by using different meanings for the “responsible for how awful our economy is” phrase.
And, first, Market monetarists and Keynesians both say the Fed successfully stopped a financial sector collapse through QE. In that respect, its actions were a great success, and *most* economists agree. Libertarian liquidationists who think that not letting the banking system implode was a terrible mistake are a fringe group.
Secondly, the market monetarist sense of “responsible for how awful our economy is” means that they think the Fed should intervene more vigorously.
That sense is the direct opposite of how an Austrian would understand the phrase “responsible for how awful our economy is”.
RIght, MMers and Keynesians are in virtual agreement on the powers and limitations of a central bank during a crisis.
Most Keynesians, when pressed on that point, will revert to the MMT position. I doubt anyone can find a Krugman article from the past 5 years where he explains that there’s only so much and no more that a government can do to “stimulate” an economy.
They might disagree about the mechanical details, how to implement the “stimulus”, etc.
My bad Tel – by MMers I meant Market Monetarists. Though, according to LK perhaps this is a distinction w/o a difference.
Sorry, no my bad. I did think you said MMT.
LK do you really not see the relevance here? Did the Moon Shot miss the moon, with 90% of the scientists saying the thrusters were on too little, and the other 10% saying there were on too long?
Ben Bernanke said, “Thanks to you, we won’t let it happen again.” Oops.
What would the world have to look like if the empirical approach didn’t work in economics?
“Ben Bernanke said, “Thanks to you, we won’t let it happen again.” Oops.”
Are you capable of seeing that this phrase refers to letting the banking system collapse?
Is it your belief that the banking system collapsed in 2008?
Is it your belief that the Fed interventions did not prevent collapse of the banking system?
“with 90% of the scientists saying the thrusters were on too little, and the other 10% saying there were on too long?”
There is a analogy: mainstream scientists versus “creation scientists”. Of course, those of us not blinded by an a priori commitment to the truth of a sacred book can see that the creation science “method” is unacceptable.
Ben Bernanke said, “Thanks to you, we won’t let it happen again.”
Bernanke said that to Milton Friedman in a speech given in his honor – well *before* the ‘subprime’ crisis – NOT during a ‘banking system collapse.
As far as for the rest of what you write – there is a reason for the old joke that if you line up a bunch of economists end to end they will all point in different directions – and it isn’t because of a fringe group of them called the ‘Austrians’ are doing ‘creation science’…
LK wrote:
There is a analogy: mainstream scientists versus “creation scientists”. Of course, those of us not blinded by an a priori commitment to the truth of a sacred book can see that the creation science “method” is unacceptable.
OK LK, let’s take your analogy on its own terms. Why is it that in economics, the analog of the creation scientists are so often in charge of central banks and regulatory agencies?
There is nothing analogous to this in the natural sciences.
There is actually, and that would be environmentalism… in particular the sort of hocus-pocus environmentalist who gets freaked out if anyone attempts to change anything because they (and only they) know the one true purpose of Gaia and somehow it always involves lots of research money and political power going to environmentalists. Somehow it never involves allowing ordinary people to make their own choices about whether they want to cut down a tree, or plant some non-native shrub, or whatever else they decide would improve their personal environment. For some reason the Gaia worshippers always find that normal people can’t be trusted with very important environmental decisions.
The parallels with government economists are quite striking, and the claims of scientific backing are also very similar.
Personally, I don’t regard our present day practice of evolutionary biology, nor environmentalism to be scientific. That’s not to way there is no possibility of useful scientific enquiry down that track, I just happen to think we are doing a bad job of it right now.
“Mises’s disutility of labor can only be established by empirical evidence”
The disutility of labor is a very, very minor point in Austrian theory, with little consequence. Certainly, it’s not really anything more than an assumption that people prefer leisure over effort, which is generally the case. Any conclusion Mises reached via this concept can, in fact, be reached in a similar manner by simply pointing out that people prefer to reach their subjectively valued goals as quickly as possible.
“Any conclusion Mises reached via this concept can, in fact, be reached in a similar manner by simply pointing out that people prefer to reach their subjectively valued goals as quickly as possible.”
So, Mises was kind of an idiot to insist on this principle so strongly?!
Rather strong words, as it isn’t like he was wrong about it – there is, in this universe, a disutility attached to human labor. My point was that you can replace the concept of disutility of labor with simply opportunity cost and time preference without a problem in the conclusions. In a sense, Rothbard did that himself in MES:
“It is possible that included in this “return” of satisfaction yielded by labor may be satisfaction in the labor itself, in the voluntary expenditure of energy on a productive task. When such satisfactions from labor do not exist, then simply the expected value of the product yielded by the effort will be weighed against the disutility involved in giving up leisure — the utility of the leisure forgone. Where labor does provide intrinsic satisfactions, the utility of the product yielded will include the utility provided by the effort itself. As the quantity of effort increases, however, the utility of the satisfactions provided by labor itself declines, and the utility of the successive units of the final product declines as well. Both the marginal utility of the final product and the marginal utility of labor-satisfaction decline with an increase in their quantity, because both goods follow the universal law of marginal utility.”
Were someone to place no value on even the first “unit” of leisure, they would STILL follow the conclusion that “people work only when they value the return of labor higher than the decrease in satisfaction brought about by the curtailment of leisure.” (Mises, Human Action) They just don’t value the leisure. Subjective preference!
Mises:
“The fundamental facts affecting the supply of labor are:
“6. Men prefer the absence of labor, i.e., leisure, to labor, or as the economists put it: they attach disutility to labor.”
So Mises thinks this is a “fundamental fact” affecting the supply of labor. If only he had realized it is a “very, very, minor point,” instead of a “fundamental fact”!
If I reword that to say that men have subjective preferences, and as such will only work as long as that labor helps them achieve the highest valued preferences, the conclusions are the same. In reality, the statements are the same, as there is no one good called “leisure” – leisure is simply the term for “all goals that are not ‘work'”. If I go to the beach, that is leisure. If I read Mises after work, that is leisure. If I engage in discourse on Murphy’s blog, THAT TOO is leisure.
To say “men prefer the absence of labor, i.e., leisure, to labor” is to say nothing more than that subjective preference, including time preference, and opportunity cost exist.
‘To say “men prefer the absence of labor, i.e., leisure, to labor” is to say nothing more than that subjective preference, including time preference, and opportunity cost exist.’
So, it was absolutely incorrect of Mises to insist that the disutility of labor was a fundamental?
Cause you can’t have it both ways: if it is a very minor point that can easily be done away with by just talking about subjective preference, than it was really silly of Mises to call it “fundamental”!
Mises held it to be a “fundamental fact”…affecting the supply of labor.
I do believe that if the context is economics, and/or praxeology, then it is not unreasonable to call it a “minor point.”
Oops, try *research* major: it affects the entire economic world for Mises. See quote below.
It does not affect “the entire economic world” according to Mises, Gene. The whole world for Mises is ACTION, not inaction.
As long as we’re splitting hairs: I believe Matt said “a minor point in Austrian theory”, not perhaps in the supply of labor. Unless the supply of labor is the central point in the entire Austrian theory, as opposed to, for example, the supply of apples.
Also, the number 6 may imply that there are 5 other, perhaps more fundamental facts affecting the supply of labor? Dunno, and too lazy right now to look it up.
So, Peter, it is “splitting hairs” to protest over calling something Mises brought up 60 times in Human Action “a very, very minor point” that Mises easily could have addressed without introducing such a useless concept?
“Dunno, and too lazy right now to look it up.”
And there you have it! Let’s just make up our own, “a priori” view of what Mises ought to have written, cause then we don’t need to bother with difficult books like Human Action.
As near as I can recall, what Mises demonstrates with the disutility of labor is that: “the self-sufficient man who works in economic isolation for the direct satisfaction of his own needs only, stops working at the point at which he begins to value leisure, the absence of labor’s disutility, more highly than the increment in satisfaction expected from working more.” Basically, the whole concept amounts to “men derive some value from ‘doing nothing’ as an end in itself”. In short, “the effort isn’t worth it”.
Item #5 in that list is that “fatigue impairs work quality” (essentially). Is this also a fundamental requirement for deriving the laws of economics? Or rather, does that simply provide an additional constraint on how much labor is available?
For some reason you have taken the “supply of labor” as the whole of economic theory, instead of Mises stating a few facts about things that affect that supply. The number of apple orchards is a “fundamental fact” affecting the supply of apples, too…. does this mean that praxeological conclusions depend on counting apple orchards?
See Matt, as someone trained in *research*, I don’t just guess what someone was saying based on what I think I recall. I look it up. And I find Mises mentions the disutility of labor around 60 times in Human Action. Funny that he would put in a very, very minor point so often, hey?
And he writes:
“In a world in which labor is economized only on account of its be- ing available in a quantity insufficient to attain all ends for which it can be used as a means, the supply of labor available would be equal to the whole quantity of labor which all men together are able to expend. In such a world everybody would be eager to work until he had completely exhausted his momentary capacity to work. The time which is not required for recreation an2 restoration of the capacity to work, used up by previous working, would be entirely devoted to work. Every nonutilization of the full capacity to work would be deemed a loss. Through the performance of more work one would have increased one’s well-being. That a part of the available potential remained unused would be appraised as a forfeiture of well-being not compensated by any corresponding increase in well-being. The very idea of laziness would be unknown. Nobody would think: I could possibly do this or that; but it is not worth while; it does not pay; I prefer my leisure.”
Sounds like a mighty different world, huh? Not a very, very minor point at all.
But what did that Mises fool know?
Again, Matt said “minor, minor point in Austrian theory”. Unless the Austrian theory primarily deals with the supply of labor. Does it? Is Human Action entirely devoted to the supply of labor? Is Human Action the only book ever written on Austrian econ? So my question again is, is the 6th fundamental fact affecting the supply of labor an essential piece of Austrian theory?
All I know is (and please, feel free to ridicule me or put some words in my mouth in your charmingly condescending tone, you seem to be good at that), that the 6th answer on family feud is never a very relevant answer. And in the Olympics, being in 6th place is nice, but not relevant.
That was my only, minor, minor point, wasn’t making up “my own a priori view of what Mises ought to have written”, whatever that means.
This quote does not show what you think it shows.
Yeah, I’ve taken jobs because I wanted to learn about the subject matter and also get paid. Of course I also enjoy sitting on the beach and drinking beer, but I guess there’s a balance between long-term achievement and short term leisure time.
At any rate, money is also typically a requirement to achieve anything in the modern world, so I very much agree that achieving goals is the key point and exactly which goals people tend to set for themselves is a secondary enhancement of the first point.
“These points make massive concession to empiricism.”
If by that, you mean things like when we say “price controls lead to shortages if the legislated price is lower than the market price would be” we can’t know ahead of time whether there will actually be price controls, or the relation of the legislated price as opposed to the market price. Which is not really a “massive concession” at all.
Virtually all other schools say central bank intervention averted a worldwide financial collapse and second great depression on the scale of the 1930s.
Central bank intervention only prevented a “second great depression” if, by that, you mean a depression in artificially stimulated sectors of the economy.
But those sectors are SUPPOSED to crash because they are being propped up through coercion.
In Defense of “Extreme Apriorism”
[WWW]http://mises.org/daily/5195/
Whether we consider the action axiom “a priori” or “empirical” depends on our ultimate philosophical position. Professor Mises, in the neo-Kantian tradition, considers this axiom a law of thought and therefore a categorical truth a priori to all experience. My own epistemological position rests on Aristotle and St. Thomas rather than Kant, and hence I would interpret the proposition differently. I would consider the axiom a law of reality rather than a law of thought, and hence “empirical” rather than “a priori.” But it should be obvious that this type of “empiricism” is so out of step with modern empiricism that I may just as well continue to call it a priori for present purposes. For (1) it is a law of reality that is not conceivably falsifiable, and yet is empirically meaningful and true; (2) it rests on universal inner experience, and not simply on external experience, that is, its evidence is reflective rather than physical;[7] and (3) it is clearly a priori to complex historical events.[8]
Lord Keynes writes: “And even Mises accepts this basic sine qua non of empiricism:
‘Every theorem of praxeology is deduced by logical reasoning from the category of action. It partakes of the apodictic certainty provided by logical reasoning that starts from an a priori category. Into the chain of praxeological reasoning the praxeologist introduces certain assumptions concerning the conditions of the environment in which an action takes place. Then he tries to find out how these special conditions affect the result to which his reasoning must lead. The question whether or not the real conditions of the external world correspond to these assumptions is to be answered by experience. But if the answer is in the affirmative, all the conclusions drawn by logically correct praxeological reasoning strictly describe what is going on in reality’ (Mises, L. 1978 [1962]. The Ultimate Foundation of Economic Science: An Essay on Method (2nd edn, Sheed Andrews & McMeel, Kansas City. p. 44).
These points make massive concession to empiricism. It appears you forgot them.”
These points do not make a massive concession to empiricism, at least the empiricism you have in mind. In the case of leisure, the empirical verification of the postulate “Humans prefer the absence of effort, or humans prefer leisure to labor” is performed before the application of that postulate to the theory. However, you seem to take Mises to mean that empiricism is then appropriate to be used on the macro level of the theory, rather than the more sound approach of restricting empiricism to those tasks it is suited to handle in the social sciences, such as the verification of certain postulates.
A good example of the failure of ’empirical’ economics is the Phillips curve. It’s ‘discovery’ was atheoretical, and was also short-lived. There are hoards of examples of this sort of thing in finance, going all the way back to the sunspot theory (18th century) that suggested when there are more sunspots, stock prices will increase, and when there is a comparative dearth of sunspots stock prices will decrease.
Even some of the leading asset pricing models are empirically founded, which leads them to fit the data in some periods, and not in others. For example, the size effect comes & goes, depending on the time frame chosen.
“A good example of the failure of ‘empirical’ economics is the Phillips curve. It’s ‘discovery’ was atheoretical, and was also short-lived.”
Hmm. Do we really want to say that every case in which a scientific theory fails is a failure of the science itself? Did “empirical astronomy” fail when Kepler’s theory of planetary solids turned out to be nonsense?
I wasn’t talking about every science, just economics & finance.
I think empirical evidence can much more easily become theory in the natural sciences because those sciences deal with fixed phenomena – or at least much more fixed than anything dealing with human action. Empirical studies work very well for developing a theory when we can expect the same thing to happen over & over – like orbit patterns.
“I wasn’t talking about every science, just economics & finance. ”
That is completely irrelevant to the point I was making. You proposed an argument of the form:
*****
One theory in (possible) empirical science X has failed.
Therefore, empirical science X has failed.
*****
It doesn’t matter how you limit it, that argument is unsound.
(And my point here is not about whether economics can or should be an empirical science: I am only saying your argument here doesn’t work.)
If my phrasing led you to think I was making such a broad statement, then I apologize for being misleading.
My argument is: several theories (I named only one from economics, and pointed out that finance is rife with them) that were derived from empirical observation have failed. Therefore, empirical observation is likely not the way to derive theory in this particular field of inquiry.
No, Jeff. I don’t think you need to apologise. Clearly, Gene has misread what you have written. You implied no theory of the kind he has explained. At least, your wording does not imply any such underlying theory. “One example of…..” is just a sentence that highlights an instance, as an illustration, of the general phenomenon identified later in the sentence.
OK, Bala, then a good example of the failure of empirical astronomy is the failure of Kepler’s theory of the planetary solids?
Note well, Jeff did NOT say, “One example of a failed empirical economic theory is…”
Gene,
Please do explain how “One example…” is so very different from “A good example…” in other than the introduction of the postive adjective.
“Therefore, empirical observation is likely not the way to derive theory in this particular field of inquiry.”
Still suspicious, though. Look: every single physical theory before the very latest one “failed” in one way or another. So is empirical testing also bad in physics?
Well, as I suggested, natural sciences are different from social sciences in the sense that phenomena are repeated in the natural sciences. That means strict observation will eventually lead to theories. That doesn’t mean the current theories are correct, true. It just means they fit the data best. I would think social sciences have an advantage here sense some laws can be derived through logic (whether verbal or symbolic) that are true always and everywhere.
LK wrote (in commenting on an insight by Mises):
“These points make massive concession to empiricism”.
I understand economics to be a theoretical explanation of the real world. There can be no theoretical explanation without the ‘data’ of the real world. We could say 1+1 = 2, but without the real world experience what would be the use in knowing 1+1 = 2. Recognising the necessity of ‘reality’ I feel is not a concession to empiricism, it is simply the apriori methodological approach reflecting on the soundness of its theories when comparing them with the ‘data’ (ie. reality).
Flipping the coin, no historical period under investigation is ever identical to another. Therefore drawing empirical evidence to determine an economic law is impossible because it cannot be universally valid through time and place as any law attempts to be.
Have I understood the questions that Professor Murphy and his critics are raising?
If empirical studies are performed, then heavy caveats must be listed.
Most empirical research “disproving” Austrian economics comes from the fact that ending the profligacy of gov’ts and central banks will allow a depression to ensue. They ignore the cause of the business cycle and ignore the fact that the recession is the return to stable footing. Austrian economics is the hardest to prove, when looking at simple correlations and ignoring causation.
Austrian economists are the only economists that look at observations from every possible point of view. ex: “Contrasting Views of the Great Depression | Robert P. Murphy”
The spendthrift schools (MMT, Keynesian et al.) see the world in a very narrow, immature way. This is the only way their schools can survive. Using logic would explode their prescriptions.
Mark Thornton talks about this very topic in great length on this Podcast
http://academy.mises.org/courses/bbb/
This is off topic, but I remember listening to a lecture once in which it was explained that Hayek wrote something about the possible benefits of empiricism regarding some topic, and he was expecting Mises to be pretty pissed about it. But as it turned out, Mises was fine with the statement and agreed.
I’ve been trying to recall what this was about for some time now with no luck, does anyone know what I’m referring to? Thanks.
The whole “Economics and Knowledge” essay.
Great, thank you!
There are a few things to bear in mind here.
First off, the comparison of economic empirical methods to chemistry and physics is pretty lackluster. Those are (largely) deterministic sciences where there are clear agents at work. However, even within that there are limits to what can be done. But even then, simple deterministic physical structures can become far too chaotic for forecasting given a long enough time horizon. If I asked any Austrian economist whether or not the economy was a deterministic system I hope they would laugh heartily before answering. Thus it’s unfair to compare *predictive* results of physics to predictive results in economics. The two aren’t the same, and how we consider the value of empirical methods to economics needs to be weighed against this.
Secondly, the use of empirical methods in economics is largely descriptive. Most articles I’ve read so far have a lot of theory, but as ultimately consumer utility preferences cannot be met with existing statistical methods- though I’m sure Hal Varian is slowly debunking that with big data work using the weak axiom of revealed preference- end up using aggregate data as a proxy for what they’re missing. This tends to both weaken results and generate excess null or factually wrong results in my experience- especially in tax policy where I held a lot of early interest. I doubt few PhD level statisticians have even figured out the best way to approximate various ‘actual’ coefficients of non-deterministic systems consistently, and PhD economists in my experience tend to be trained even worse. Using your benchmark based around preditive analysis undermines the value of empirical methods to the field.
Thirdly, as a positive response to your inquiry: Questions such as what has driven real wage growth among certain sectors of the economy are routinely asked within labor economics, and fairly robust empirical methods give reasonable and consistent answers. The same thing has occurred with human capital attainment. Things like the “mincer equation,” have been incredibly robust benchmarks for individual wages across various countries and time periods, almost stupidly so in explaining about the same variance, and having routinely approximated coefficient estimates. Yes, I’m aware of the Austrian critiques against such extrapolation, but used descriptively they do offer us a better insight into how things like education attainment compared to raw skill growth tends to affect life time earnings of workers, and other extrapolated knowledge.
A lot of tax policy work has slowly come to a larger consensus as empirical methods have improved in how to approach discontinuities.. Same thing goes among estimating how various health care policies can impact inter-hospital level of care. The main factor here is that these studies focus on smaller areas than are typically mentioned in the WSJ or reported by groups such as Macroeconomic Advisors. As we extrapolate into larger systems empirically more non-deterministic elements get thrown into the mix which require researchers to make more unreasonable assumptions about the industry structure and what the distribution of possible effects are.
Within the two fields I’ve had some graduate level exposure to, tax policy and labor economics, empirical methods have provided a better measure of counterfactual analysis estimating current elasticities using improving statistical and computational methods. It has provided descriptive tools to better explore theory with that might allow us more reasonable estimates of various partial equilibrium effects for future predictive analysis. Is it there yet? I don’t think so, but that doesn’t mean we can’t discount empirical methods existing value to the field.
Finally, It is easy to find things like macroeconomics and point out all the woes with trying to extrapolate ‘truth’ out of it. But throwing out the empirical method entirely in that critique is like nuking a small village because it has the bubonic plague. Further I also feel that non-statistical methods such as agent based modeling offer a new modeling technique that through moment matching may prove to be empirically validated. There are other ways of being empirically validated.
To respond to the author: how do you feel about articles such as this, http://mises.org/journals/qjae/pdf/qjae16_1_4.pdf that appeared in the latest Quarterly Journal of Austrian Economics? I’m personally of the opinion that this provides a lot more robustsness in arguing for the ABCT then the current verbal tradition present by people trying to spread Austrian economics.
“I will bet you that the vast majority of what they teach involves getting the students to think a new way”
Yes, I’ve noted several times that Mises is not nearly as far from the mainstream as both the mainstream and some of the Misesians think: there is a lot of “praxeology” in, say, Landsburg.
The fault is on both sides: Mises made some marketing errors in overstating his case and choosing a weird word for it, and the mainstream really wanted to believe their empirical rhetoric.
Steve Landsburg does use the assumption that people are rational in his writings, but he takes great pains to point out that “all models are false”, and that the assumption of rationality is a simplifying tool, akin to how physicist assume no friction when calculating the trajectory of a baseball.
I always thought it was Hiroshima… but yeah if your null hypothesis is that a person got to the moon by random chance or that a city demolished itself purely by accident, then I think it’s fairly safe to demolish this null hypothesis and believe that physics, chemistry and engineering do have a tangible effect.
If your null hypothesis for a modern economy is that it muddles along either with or without the economists in much the same way, then it’s a much more difficult one to knock over.
If you want to go one step further and try to claim that central government economic management has overall led to greater growth than otherwise… well that’s a very shaky claim indeed.
I like how Bala and Daniel Kuehn are probably surprised, but with two very different takes on it.
“What?? Murphy’s talking about methodology?
+1
How do I process the irony of Bob asking for empirical proof of successful empiricism?:) His point (4) on deductive logical approach is spot on. It seems like in most of the empirical work we do, we end up needing to apply a priori deductive reasoning to make sense of it, especially with incomplete data and measurement errors etc.
gofx wrote:
How do I process the irony of Bob asking for empirical proof of successful empiricism?:)
How is that ironic? I’m saying the people claiming you need to use the empirical method as a criterion of success, can apply that very criterion to their own rule.
I mean, it’s ironic for them, is that what you’re saying? Or are you saying, “Bob doesn’t realize he typed the same phrase twice in his blog post title”?
I am an Austrian in Economics and a Quant (who said an Austrian is unable to do math himself?). I completely disagree (for once) with most Austrian bloggers about the utility of mathematics and quantitative techniques in economics 1/ Better one eye than quite blind! Quantitative techniques help us to make better predictions in many (not all) cases but of course cannot be fully relevant and accurate (not even in natural sciences). 2/ Do not oppose mathematics and economics. Math is simply a language. Is a Japanese less relevant in economics because of his particular language? All economics can be expressed in math language as well as in any other language. The problem is not the math but the thinking underlying it. Do not blame the math instead of the bad economists.
I totally agree with you.
“Math is simply a language.”
This is not actually true. Math is a subset of logical analysis relating specifically to quantitative questions. If, as the Austrians hold, economics has no sensible quantitative analysis, then math is not useful. It would be like trying to hammer a nail in with a screwdriver.
It’s better to say that math *can* be a language. I think Matt is right that math is more, or at least other than, a language. But you can use it as a more precise language and rely on it’s theorems to prove your case. So if you can represent something by a smooth function you can express your results in calculus and rely on the correctness of the reasoning. But of course that does not reilieve you of the need to verify and defend your argument by other means; nothing in the world of econ is exactly described by a smooth function to stick with that example.
It’s in the assumptions. Math is correct, but once one starts assuming certain criteria to particular functions, that’s often where they fall into trouble.
Is the scientific method used to choose these criteria, and do they even make sense in that realm of knowledge (math) as compared to human actions?
Yes. Using math is a good idea to make your arguments precise and often quantifiable. But just phrasing an argument in math is not in itself enough to prove the argumet is sound.
Logic is also math, by the way.
Just applying boolean algebra instead of other types of algebra. Any argument expressed in logic is still mathematical, and it can still have wrong assumptions just the same.
There are plenty of topics in mathematics that have literally nothing to do with quantity. Numbers are not the foundation of modern math, but merely one of its subjects of study. One might even get away with defining mathematics as “formal a priori reasoning.” Go back to Euler, etc. and you can see examples of math without symbols (although at that point I think it was strictly quantitative).
It seems like the more broad-minded Austrians realize this, and focus their criticisms on empiricism and positivist epistemology rather than math.
“There are plenty of topics in mathematics that have literally nothing to do with quantity.”
Such as? Keep in mind that mathematics is DEFINED as “the abstract science of number, quantity, and space”.
Set theory, mathematical proofs, model theory, etc. – these are all quantitative in nature. You might have a case with something like Boolean logic, but that is still technically operations upon quantity. X AND Y is equivalent to X * Y. NOT X is equivalent to 1 – X. X OR Y is equivalent to X + Y – X * Y. In reality, all boolean algebra consists of is a set of useful operations to represent “true” and “false” via a number line constrained to the integers 0 and 1.
I’m not gonna wade into this except to note you are quite wrong about boolean algebra. Boolean algebra applies to much broader structures thatn {1,0}
I would say group theory, but you would not be convinced if you believe that set theory is “quantitative in nature.” Why do you say that?
“You might have a case with something like Boolean logic, but that is still technically operations upon quantity.”
That’s a valid isomorphism, but it’s a stretch to claim that it makes Boolean logic quantitative in nature. Is economic theory quantitative in nature because it deals with prices?
“Why do you say that?”
Set theory involves quantitative collections. Each basic function of set theory can be rewritten as algebra performed on numerical quantities. In reality, set theory is simply a combination of standard algebra performed on the characteristics of the set and Boolean algebra on the characteristics of the elements.
“That’s a valid isomorphism, but it’s a stretch to claim that it makes Boolean logic quantitative in nature.”
All operations are algebraic: sums, multiples, etc.. By definition, these are performed on quantitative data.
“Is economic theory quantitative in nature because it deals with prices?”
Ignoring for a moment that this is not all economic theory deals with, we still find that price analysis is not really quantitative in nature. The price is a quantity, but the laws relating to it (all of them requiring “ceteris paribus” caveats) are not.
Let B be the set of all human behavior, and A be the set of human action. A is a subset of B. Arguing about meta-math in blog comments is presumably an element of A, so I will stop now.
“Let B be the set of all human behavior, and A be the set of human action. A is a subset of B.”
Certainly. But there is a cardinality of the sets, the cardinality of A is less than or equal to B, and so forth. Even the demonstration that A is a subset of B requires first comparison, and then mathematical operations of an algebraic nature.
It is fundamentally different, in as much as a price is very finely divisible, but a logic statement is not.
Which one better represents the world depends a bit on application, but I doubt you are going to be able to model planetary orbits, or motor vehicle dynamics with boolean algebra. Sometimes you just have to sit and carefully select the right tool for the job at hand.
Lio wrote:
Quantitative techniques help us to make better predictions in many (not all) cases but of course cannot be fully relevant and accurate (not even in natural sciences).
In this blog post, I am saying that “make better predictions” is barking up the wrong tree when it comes to figuring out the content of economic laws or principles. To say that quantitative techniques are essential for making better predictions has no bearing on this blog post.
Sigh. These squabbles about methodology are always so vague. Can you phrase your argument as an equation please?
In discussions of the scientific method, you rarely see someone come right out and honestly say “The scientific method relies on empirical observation, which is, in turn, a position of faith in the relative reliability of sensory data.” I suspect that, if empirics-only economists would stare at that simple truth long enough, they would be more open to the notion that there are Other things they know but either can’t prove, or don’t need to.
IOW, Descarte was wrong. Doubting your senses is silly and gets you nowhere. Likewise, doubting the relevance of incentives, even as a part of a counter factual explanation, where incentive effects cannot be statistically separated from an event, is silly.
Reliance on empirical observation also requires that one accept truths that do not – causality, constancy, etc.
There is another possibility which is to say that the Austrian approach is actually scientific in the same sense as the approach of the “natural” sciences while the mainstream approach is not. This is possible to demonstrate if we abandon the positivist version of science (i.e. inquiry based on controlled experiments and their analogues) and the constructivist understanding of mathematics as something that is just in the mind.
In this realist sense any science is based first and foremost on certain fundamental observations, or qualitative inductions of the world without which no experiments (or their surrogates) are possible. E.g. in physics you have to have an intuitive grasp of what space, time, force, energy, work, light, movement (in general and at the level of ordinary reality) are to even start a fruitful inquiry. It’s the same in economics. Unless we have an intuitive grasp of basic phenomena like choice, value, goods, money, etc. you can’t make sense of economic history.
This dose of realism allows to realize that while the physical phenomena are in a certain sense mathematical (e.g. material objects have dimensions) and deterministic, the economic ones aren’t. So attempting to apply mathematical concepts to the study of economic phenomena is an antithesis of science because it contradicts even the most fundamental bits of experience. The realist approach also allows to realize that economic phenomena happen inside human minds, not outside, so you can’t test economic hypotheses by purely behavioral observations (of which statistical data are part).
So to wrap up. the Austrian approach is pretty scientific. It’s based on fundamental observations and it remains faithful in the analysis to the features of the economic world that those observations point to.
Yes, Bob. Ironic for them. I assumed you had a wry smile when you wrote the headline.
gofx OK. Sorry for doubting you but…look at some of these comments.
How else would we know whether our methodological criticisms applied to the real world?
How would we define “empirical method”? For I am pretty sure that most, if not all, economists do not practice what I would understand as such. Take for example Feynman’s lectures on “The Character of Physical Law” (here: http://www.physicsteachers.com/pdf/The_Character_of_Physical_Law.pdf, go read it, at least the first lecture). He describes in the first lecture how the law of gravity was discovered as the result of measuring something visible and tying various phenomena together through prediction. For this purpose the measurements were summarized in math. This to me is the “empirical method”.
I am pretty sure that no economist does such a thing. For us purpose is invisible, go ahead try and measure the ends that cause people to act. The only empirical generalizations I know of are so-called “stylized facts” such as “Okun’s Law”. The fundamental difference between Okun’s law and gravity should be obvious to any economist trained today.
Bob, I have a question about praxeology, that I asked in a previous comment thread as well: has there been any attempt to systematize praxeological reasoning and approach the rigor of Euclid? By that I mean having numbered axioms, theorems, proofs, etc, laying out step by step derivations of various statements from Mises’ action axiom (and what whatever supplementary axioms Mises and Rothbard use). And have there been attempts to express praxeology in symbolic form, similar to how people nowadays use algebraic language to do geometry, rather than expressing geometry in verbal form as Euclid did?
has there been any attempt to systematize praxeological reasoning and approach the rigor of Euclid? By that I mean having numbered axioms, theorems, proofs, etc, laying out step by step derivations of various statements from Mises’ action axiom (and what whatever supplementary axioms Mises and Rothbard use).
I know some people have talked about doing this; I don’t know if anybody has.
Michael Oliva Cordoba has started to –
http://mises.org/media/7903/Getting-Mises-Right-On-the-Philosophy-and-Logic-of-Praxeology
He told me he started this after a colleague told him about what Mises had tried to do. Cordoba had never heard of Mises before and after reading some of his work was skeptical about one of Mises’s foundational arguments. But, he put the argument in symbolic form and became convinced Mises was right.
What was funny was he told me he presented his findings to his peers and they rejected it; but without showing him where he had gone wrong or giving any specific criticism – they just said it couldn’t be right.
Thanks, but I’ve already seen his paper here:
http://www.philosophie.uni-hamburg.de/Team/Cordoba/Materials/oc_Foundations-PraxEcon-05-final_2013-05-28.pdf
He proves two rather trivial propositions that I think most people would readily admit logically follow from Mises’ definition of action. Something like this would only be convincing if it proved some substantive propositions, like the law of demand, and especially the controversial macroeconomic claims of Austrians, like the nature of recessions and the effects of fiscal and monetary policy. I’m less convinced that such claims can be proven just by assuming that human action is purposeful.