July 2014

Sun Mon Tue Wed Thu Fri Sat
    1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30 31    
Blog powered by Typepad

« Pete Boettke on Austrian Economics | Main | The Troubling Transition to Democracy In Russia Continues »

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d83451eb0069e2016760ae2a94970b

Listed below are links to weblogs that reference Mises and the Importance of Context:

Comments

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

Right. So lets STOP calling it "information".

It's not stuff sitting out there as a univocal "given" in a box with a "given" search price tag.

When our understanding advances our conceptual scheme changes -- we are NOT simply adding to our collection of "information" boxes, with "given" search price tags attached.

Peter writes,

"And the epistemic point is not an information theoretic argument. It is not that the information is too costly to gather, it is that the knowledge required to make the rational calculations does not exist outside the context of the private property market economy."

Would you agree that central to Mises' scheme is the idea of purposeful behavior? If so, it also follows, does it not, that purposeful behavior is absent in conditions of equilibrium? (In Mises' words, the individual is now in a vegetative state.)

That, I think, is all well-settled and uncontroversial. But what of the idea of equilibration? Does it follow that the further the market "equilibrates" (i.e. moves in the direction of equilibrium), the less purposeful human action becomes? If that is true, then why do we champion equilibrating market properties?

In reading Mises, I see his argument as NOT being so much concerned with the *efficiency* of market exchange AS IT IS being concerned with retaining the purposefulness in human action (which is necessarily separate from effiency arguments).

One can in fact also make an information theoretic argument, too. Socialism requires perfect information. However, according to information theory, perfect information is the same thing as no information. Therefore, socialism has no information with which to make decisions. So even if one could input all the information, socialism is impossible.

This is one of many points I make on my own blog posting on the colloquium.

What I quote below from Peter never been well clarified, i.e. the various dimensions for exactly why this "knowledge" is not univocal across space & time and across individuals has not been clearly isolated and explicated, and their relations articulated and clarified. All sorts of specific dimensions of this get muddled and squished together in the multi-used word "subjective", which has all sorts of alternative jobs and varying significance in the "Austrian" literature.

Hayek made a start in his _The Counter-Revolution of Science_. And others have talked abou the role of alternative individual judgments, and the role of the relational logic of valuation, and the role of changing historical price relations, in the market process. But no one has isolated each one of these (or others) and shown how they are related to univocal natural kind or standard economic good categories in the world, or how they relate to each other.

It just hasn't been done -- and may believe that the job the word "subjective" does is just one job, when in fact, people are using it to do several jobs in several alternative domains, e.g. the real world of entrepreneurial evaluation vs. the pure logic of marginal valuation, etc.

Until someone makes is clear in rigorous and precise detail how and why knowledge does not equal information in economics, no one in the top Grad departments is going to "get it".


Peter writes,

"And the epistemic point is not an information theoretic argument. It is not that the information is too costly to gather, it is that the knowledge required to make the rational calculations does not exist outside the context of the private property market economy."

Another minor wrinkle in this discussion is the odd case of "hostile nationalizations," which rarely occur. The most prominent examples were those of certain French private firms after WW II whose owners were accused of collaborating with the Nazis. These owners were "punished" by having their firms taken away from them without compensation.

The upshot of this is that these firms were also viewed in a hostile manner by the rest of the government for many decades, with no subsidies or other soft budget constraint sorts of support that most firms in market socialist contexts get. This means that they must compete in open markets in a profit-maximizing way, more or less, in order just to survive, and in the case of these French firms, they did so for many decades, although eventually this animus left over from the war faded, and they began to exhibit the classic syndromes.

The most important example of this was Renault, which was nationalized, in contrast with Peugeot, whose owners had apparently avoided collaborating and remained private. The two performed approximately equally in terms of profitability, innovativeness, and so on, for quite a few decades.

@Rosser:
Could you elaborate on your point. I find it interesting, but I don't see how a state company, subject to the whims of state bureaucrats, could perform well.

We had some experience here with state owned car manufacturers who, although they tried, couldn't compete even with second hand cars. The only reason they didn't disappear was that they were owned by the state.

As a matter of fact Trabant could be considered a huge success since it managed to survive for decades and almost always had a huge backlog.

If you could provide more data about Renault, much appreciated. Thank you.

Troy,

You said, "However, according to information theory, perfect information is the same thing as no information."

I definitely don't understand that one. Can you elaborate, please? (Disclosure: I pretty much assume that I'm going to disagree with you.)

XYZ: Your question about equilibrium highlights for me what I dislike about the term. What in economics is generally referred to as "equilibrium" is actually a steady state, rather like Mises' evenly rotating economy. Purposeful action continues, but it follows an unchanging routine. People still want to eat, they want to replace worn-out shoes, etc, so they get off their butts and go to work, but the routine does not change. "Equilibration" makes sense -- and serves a good purpose -- as incremental refinements of the routine that make it more efficient. But real progress stems from disequilibrating Schumpeterian entrepreneurship.

Niko,

Trabant operated within the command socialist economy framework of the CMEA and the former DDR. Renault operated in the market capitalist framework of postwar Western Europe. It is still around, as is Peugeot, and Trabant went out of business almost immediately after the Berlin Wall fell and the former DDR was opened to western capitalism. The state controlled prices and outputs and wages and so on at Trabant. None of that was controlled by the state for Renault, and for decades it received zero subsidies or support.

BTW, I think that it has been at least partially privatized in more recent years, so this is really a story about the pre-1990 and more definitely the pre-1980 era following WW II. Prior to Mitterand in the early 80s, it definitely received zero support, but did as well as its fully private counterpart, Peugeot.

Roger,

My claim is straight out of information theory. The equation for information is the same as the one for entropy.

S = k log W, where S = entropy, k = Boltzmann's constant, and W = the number of ways the system can be arranged. Low W = low S; as W approaches infinity, so does S. With zero noise (S = 0), W = 0, meaning the system can be arranged 0 ways.

Thus, perfect information (S = 0) means the system cannot be arranged in any way at all. It is formally equivalent to there being no information available at all. There has to be noise to get complexity, to get any information communicated at all.

What would zero noise look like? According to the math, it looks like nothing.

This is, at least, how I understand everything I've read on information theory, anyway.

Troy, you seem to be saying that when Shannon's H is zero, information is "perfect." But that just means the system can take on only one state. The "information" socialist planners would require concerns alternative hypotheticals. To say that the planners would need "perfect information" -- well that's not even true, but let that go -- just means that they would have to know all those hypotheticals. Well, that's nothing to do with Shannon's H being zero. I think you may be guilty of equivocation on the phrase "perfect information."

@XYZ

> Does it follow that the further the market "equilibrates" (i.e. moves in the direction of equilibrium), the less purposeful human action becomes? If that is true, then why do we champion equilibrating market properties?

It is not the case that, as the market equilibrates, the remaining actions become "less purposeful." I don't think Mises's praxeology has any concept of "purposeness spectrum/degrees."

Instead, think of each voluntary trade as an attempt to put the world into a "more valued" state. Mises (nor any other Austrian) does not favor equilibration for the sake of reaching a vegegative state. He favors (equilibrating) trades because they are attempts to reach these better states.

@Troy

When S=0, don't you get W=1? [Not W=0]. Exactly one configuration is possible.

This doesn't change the overall argument. Information can only be encoded when there is a space of possible states [as a subset of this space?]

I think this is actually one of the fuzziest (most subtle?) points in all of Austrian economics. Can't we perform hypothetical, but logically consistent, thought experiments where a good configuration (maybe even the configuration a market *would* pick) is computed?

Unless we can prove intractability, can't we only say that Mises offers *intuitions* of why this is a hard problem?

Actually, the planners would need perfect information, because even the slightest difference in input (from error or rounding) results in diverging outputs (as per chaos theory) in a complex system. Further, isn't Hayek's critique of socialism precisely that they can never have the information needed, that socialists need complete information to do what they want to do, and that information is inherently unavailable? Yes, he puts it as "knowledge," but I think it can be expanded to information in general. My point is that there is not only no way we can know what is necessary, but we can neither have the information necessary -- and if we did have all the information, information theory says you use it to do anything.

marris,

Yes, I made an error. If W = 0, then S = -infinity.
If S = 0, then W = 1. Which still boils down to no information.

"information theory says you [can't] use it to do anything"

I don't think that makes any sense. I'm reminded of Euler's "donc, Dieu existe," except that Euler (assuming the story is even true) knew perfectly well that his "proof" was pure silliness.

There is a serious point to be made here, quite beyond negating Troy's seeming errors on information theory. There are no shortcuts. I do believe that Mises, Hayek, and others have shown that socialism cannot work the way its advocates seem to imagine. Yep. But the arguments getting you there are not trivial, quick, or easy. The socialists erred, but they were not stupid. On the contrary, many were among the best minds of their times. It simply will not do to pick up some half-understood equation from here or there and assert that this equation somehow disproves a large body of very serious thought produced by fine, trained minds. There are no magical wormholes in science. You have to do the long, hard slog and there's no getting around it. Period.

Chaotic dynamics are locally unstable in the short term, but bounded in the longer term, indeed eventually returning arbitrarily close to their original starting point. So, it is not the case that a small error in a plan means unbounded divergence. The plan will indeed get off, but how far off it gets will tend to be bounded somewhat.

In the actually existing command planned socialist economies such as the former Soviet Union, in fact they showed substantial stability over long periods of time. They were even efficient in a simplistic production sense of producing outputs well for the inputs provided, given the existing technology. The bigger problem was that what was being produced, particularly for final goods, was not always what was wanted, and more importantly, there was no incentive technological improvement, with the Brezhnevian stagnation relative to the US economy being the major manifestation of this.

OTOH, there was arguably a resiliency-stability tradeoff. While market capitalist economies exhibit all sorts of fluctuation, including possibly chaotic ones, command planned socialist ones exhibit all kinds of short-term stability. However, they are not resilient, as seen in their propensity to just suddenly fall apart and collapse under sufficient stress, in contrast to the market capitalist ones. In this regard, chaos may be a good thing. It is catastrophe that is the bigger problem.

Folks,

It is NOT a complexity point, and it is not a computability point --- that was Pareto's contribution! --- it is an epistemic point about contextual nature of knowledge in economic processes. Order is defined, as Buchanan has pointed out, in the process of its emergence. Outside of that process, and the knowledge itself DOES NOT exist. This is also Polanyi's point about moves in chess being always within a context and not abstract.

This is not fuzzy, but it is subtle. And too often missed. But this is why institutional analysis is so critical to doing good economics -- both the incentives economic actors face and the knowledge utilized by them in the market are context dependent and not abstract. Absent the context, and those incentives and that knowledge is non-existent --- something else is in existence and operating.

> Actually, the planners would need perfect information, because even the slightest difference in input (from error or rounding) results in diverging outputs (as per chaos theory) in a complex system.

[This point is in addition too, and not clearly in conflict to what Rosser wrote above].

Chaotic properties may be present in an economy, but I don't think this is sufficient to counter the socialists. The socialists don't need to pick THE SAME path that the market finds in the chaotic space. They just need to find SOME path which is not too horrible.

Maybe the space is structured so that as soon as you lose market prices, your search diverges into la-la-land. But why should this be? What is the argument?

I think this is analogous to claims that we can't build AIs because a brain is a chaotic system. Penrose argues that this is irrelevant because we're not trying to build a PARTICULAR mind. We're trying to build A mind that does reasonably well.

I know this way of framing the problem does not align with Buchanan and Polanyi's points, but why MUST we align with them? The socialists may argue that at the end of the day, we're talking about physical things being pushed around [ores being assembled into iPods, etc]. Why is this not expressible in computation or complexity terms? Or is Mises saying that the socialist economy COULD produce iPods, but they won't because there's no computational way to decide whether iPods will be valued consumer goods without testing them on the market?

What Pete said.

Everyone should read the Buchanan paper Pete references. Perhaps he can provide a link.

Taylor and Lerner thought they'd found "worm holes", didn't they ...

That's the whole point behind Hayek's ridiculing of "given data" .. the math misleads people into thinking the "givens" they have access to as stipulated assumptions are parts of the world that can be accessed by people in the world.

What's the title of the paper professor Boettke references?

The Buchanan "paper" is actually a letter to the editor in response to another paper:

http://oll.libertyfund.org/?option=com_content&task=view&id=163&Itemid=282#xx01

Greg, I might be completely wrong, but this is how I think about the "knowledge" vs "information" distinction. Information is about the transmission of knowledge, whereas knowledge is a more epistemic issue about whether a thing is even available to be accessed by anyone's mind.

For example, I don't personally know how much my car weighs, but I could find out. Now I guess we could argue about whether or not the knowledge existed before I weighed my car or not, and that's a legitimate epistemic point to make, but in some sense, it is knowable because not only do the tools of measuring weight exist, but the concept of "weight" itself exists.

However, without market prices, a calculable economic value of a good simply is not knowable. This is because "economic value" only exists as a market phenomenon. If you don't have markets, you don't have "value." There is literally no way to determine how many pounds of coal one barrel of oil is worth without a market in order to calculate how much of one you should use in place of the other, because "worth" emerges from markets. You can no more figure out that value than you can figure out what would have happened in the 20th century if the Germans had won WWI.

Pete: Yes. Mises said that preferences do not exist apart from action, which implies Buchanan's statement.

My point is that it is a complexity point, a computability point, and an epistemological point. Socialism fails on all fronts. I have no problem with pointing out every possible way socialism fails to be possible. The more, the merrier, I say.

Telling a man who knows his shirt is blue that his shirt is blue is providing him with no information. That's the bottom line. There has to be some element of ambiguity -- and redundancy -- for information to be communicated. Socialism is the economics of telling a man who knows his shirt is blue that his shirt is blue. No action is created; no action is required. Nothing happens.

marris,

What if the economy is by necessity a complex system? And what if one cannot have AI without self-organized complexity? In both cases, then, topology matters.

I would like to tie my original remark about the odd Renault example of a hostile nationalization back to Pete's orignal point about context. Yes, it matters. So, because of context, what was nominally a socialized company behaved in a thoroughly market capitalist way, whereas there are many companies that are privately owned but get subsidies and direction from governments that are much more "socialist" in effect than Renault.

Keep in mind the diferences between "market socialism" (former Yugoslavia) "social market economy" (Germany) and "socialist market economy" (China).

Josh, this is not how I look at it -- this way of looking at it really assumes the "'given' information in box" view.

The same fallacies underly the debate between Kuhn / Popper /Hansen vs Carnap / Nagel / Hempel, etc.

There is no univocal set of "given information" which stands behind everything that we can have an unmediated access to. And meanings are not passed simply like a hat in a box.

Read Wittgenstein to get some sense of how the "beetles in the box" picture is false, and how a shared social background make the communication of "information" possible.

Things get more complicated, not less, when we add the context of changing relative prices and alternative judgments of economics goods and entrepreneurial production, profit, and consumption opportunities thru time.

Josh writes,

"Greg, I might be completely wrong, but this is how I think about the "knowledge" vs "information" distinction. Information is about the transmission of knowledge, whereas knowledge is a more epistemic issue about whether a thing is even available to be accessed by anyone's mind."

Josh, I roughly agree with the rest of what you say.

No, it not dependent on the "beetles in a box" view of information. In fact, the whole point of the subjective value revolution is that "value" is only something that exists in the context of market-based social organization. It's not a thing residing within an object, but something that only exists in the communication of one human being with another as they make an exchange.

If anything, the socialists take the "beetle in a box" view of value. They thought it was an intrinsic property inhering in an object, like its mass (yes, I know the meaning embedded in the word "mass" is also a social construct, but if there are space aliens who are equally capable of putting satellites in orbit around their planet, they're going to need an equivalent concept). Boehm-Bawerk's whole critique of Marxism is showing that this is false--if you don't have the market society, you don't have "value," period.

Mises goes one step farther. He shows that the whole concept of "efficiency" arises from market society as well, that it's only an intelligible, meaningful idea in the social context of a market economy because you have the ideas of "profit" and "loss." Not only is the labor theory of value false, but there is no way to say if you are wasting resources or not, because the phrase, "A is more wasteful than B" only has meaning in the capitalist social context. So Mises issued the Marxists a challenge--construct a meaningful concept of economic efficiency from within socialism that allows you to compare nonhomogeneous things.

They never did.

Yola. Is Sautet still at Mercatus or has he moved on?

Josh, I get this. We're talking about stuff that is hard to articulate simply and quickly, and unfortunately I don't have the time or space here and now to do this topic justice.

Josh writes,

"In fact, the whole point of the subjective value revolution is that "value" is only something that exists in the context of market-based social organization. It's not a thing residing within an object, but something that only exists in the communication of one human being with another as they make an exchange."

Josh, I get and agree with what you say in your last comment.

Perhaps we are simply talking past one another.

The comments to this entry are closed.